{"id": 0, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_2/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_9/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[784, 312], [688, 344], [575, 352], [496, 320]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the grapes."}, {"from": "gpt", "value": "<point>[[900, 353], [771, 344], [632, 278], [500, 195]]</point>"}]}
{"id": 2, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_17/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_17/frame_15.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[824, 261], [765, 315], [723, 391], [697, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the croissant on the plate."}, {"from": "gpt", "value": "<point>[[655, 357], [513, 332], [412, 332], [319, 357]]</point>"}]}
{"id": 4, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_20/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_29/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[789, 312], [737, 352], [643, 374], [532, 365]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the grapes."}, {"from": "gpt", "value": "<point>[[797, 327], [717, 313], [650, 333], [602, 354]]</point>"}]}
{"id": 6, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_29/frame_15.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_34/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the grapes towards the plate."}, {"from": "gpt", "value": "<point>[[437, 276], [352, 313], [297, 378], [250, 472]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the edge of the cloth."}, {"from": "gpt", "value": "<point>[[769, 288], [669, 353], [579, 487], [523, 665]]</point>"}]}
{"id": 8, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_36/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_37/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[774, 308], [712, 353], [622, 389], [498, 400]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[776, 305], [718, 367], [647, 404], [523, 436]]</point>"}]}
{"id": 10, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_56/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_56/frame_15.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[748, 321], [695, 331], [656, 352], [618, 404]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the croissant on the plate."}, {"from": "gpt", "value": "<point>[[581, 346], [500, 323], [423, 346], [394, 372]]</point>"}]}
{"id": 12, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_70/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_71/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the grapes."}, {"from": "gpt", "value": "<point>[[917, 353], [793, 372], [669, 421], [553, 515]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the grapes."}, {"from": "gpt", "value": "<point>[[828, 319], [739, 408], [672, 475], [588, 559]]</point>"}]}
{"id": 14, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_71/frame_15.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_75/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the grapes towards the plate."}, {"from": "gpt", "value": "<point>[[429, 265], [370, 357], [319, 399], [290, 458]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corner of the cloth."}, {"from": "gpt", "value": "<point>[[789, 308], [761, 350], [707, 383], [639, 391]]</point>"}]}
{"id": 16, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_76/frame_15.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_85/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the croissant on the plate."}, {"from": "gpt", "value": "<point>[[709, 216], [633, 274], [530, 308], [402, 329]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corner of the cloth."}, {"from": "gpt", "value": "<point>[[793, 316], [692, 291], [613, 244], [524, 162]]</point>"}]}
{"id": 18, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_93/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_93/frame_15.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[835, 344], [797, 365], [759, 365], [714, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the croissant on the plate."}, {"from": "gpt", "value": "<point>[[650, 195], [573, 212], [509, 258], [434, 357]]</point>"}]}
{"id": 20, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_106/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_106/frame_15.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the grapes."}, {"from": "gpt", "value": "<point>[[807, 332], [739, 357], [693, 399], [664, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the grapes on the plate."}, {"from": "gpt", "value": "<point>[[445, 492], [353, 487], [286, 504], [269, 550]]</point>"}]}
{"id": 22, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_107/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_107/frame_15.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the grapes."}, {"from": "gpt", "value": "<point>[[798, 307], [693, 218], [580, 193], [462, 210]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the grapes towards the plate."}, {"from": "gpt", "value": "<point>[[454, 139], [382, 223], [319, 328], [303, 479]]</point>"}]}
{"id": 24, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_110/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_114/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corner of the cloth."}, {"from": "gpt", "value": "<point>[[705, 275], [659, 303], [627, 350], [604, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[798, 248], [714, 307], [664, 391], [618, 500]]</point>"}]}
{"id": 26, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_114/frame_15.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_141/frame_15.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the croissant on the plate."}, {"from": "gpt", "value": "<point>[[588, 349], [521, 332], [454, 332], [395, 361]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the grapes on the plate."}, {"from": "gpt", "value": "<point>[[811, 286], [640, 304], [487, 367], [340, 484]]</point>"}]}
{"id": 28, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_145/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_145/frame_15.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[798, 277], [748, 319], [697, 366], [605, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the croissant on the plate."}, {"from": "gpt", "value": "<point>[[555, 303], [445, 340], [361, 374], [298, 433]]</point>"}]}
{"id": 30, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_158/frame_0.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_174/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[765, 239], [622, 336], [538, 399], [445, 483]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the grapes."}, {"from": "gpt", "value": "<point>[[808, 345], [773, 359], [740, 408], [718, 510]]</point>"}]}
{"id": 32, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_174/frame_15.png", "rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_180/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the grapes towards the plate."}, {"from": "gpt", "value": "<point>[[504, 433], [408, 419], [337, 452], [318, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[769, 239], [630, 324], [542, 382], [496, 450]]</point>"}]}
{"id": 34, "image": ["rtx_frames_success_0/10_utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds#episode_189/frame_0.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_0/frame_0.png"], "height": [128, 224], "width": [128, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the grapes."}, {"from": "gpt", "value": "<point>[[805, 323], [748, 327], [671, 356], [634, 380]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[598, 362], [608, 411], [608, 472], [593, 522], [577, 569], [555, 608], [504, 667]]</point>"}]}
{"id": 36, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_0/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_7/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[341, 602], [264, 679], [163, 738], [79, 764]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[483, 370], [560, 487], [594, 624], [596, 784]]</point>"}]}
{"id": 38, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_7/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_10/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[630, 650], [538, 707], [408, 748], [269, 759]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[392, 352], [439, 441], [478, 535], [510, 624], [514, 691], [516, 752], [508, 827]]</point>"}]}
{"id": 40, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_10/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_11/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[522, 852], [472, 858], [415, 862], [358, 856], [264, 829], [219, 817], [179, 785]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[432, 342], [541, 421], [624, 523], [703, 688]]</point>"}]}
{"id": 42, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_11/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_12/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[553, 579], [449, 647], [316, 688], [182, 695]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[459, 391], [551, 502], [622, 624], [669, 786]]</point>"}]}
{"id": 44, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_19/frame_0.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_25/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[424, 366], [424, 466], [429, 584], [387, 676]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white bowl."}, {"from": "gpt", "value": "<point>[[388, 398], [445, 472], [492, 565], [522, 624], [555, 679], [571, 740], [563, 815], [557, 854]]</point>"}]}
{"id": 46, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_30/frame_0.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_30/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[622, 366], [632, 431], [632, 486], [624, 549], [608, 585], [591, 628], [563, 657], [528, 681]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[490, 524], [449, 573], [421, 602], [394, 640], [364, 663], [317, 695], [256, 738]]</point>"}]}
{"id": 48, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_31/frame_0.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_34/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[312, 460], [428, 514], [526, 590], [587, 737]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[391, 340], [445, 462], [483, 609], [508, 786]]</point>"}]}
{"id": 50, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_34/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_36/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[294, 592], [206, 664], [118, 735], [42, 807]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[420, 357], [424, 479], [445, 576], [454, 702]]</point>"}]}
{"id": 52, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_36/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_38/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[471, 521], [328, 559], [210, 643], [105, 739]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[433, 353], [437, 445], [471, 559], [513, 702]]</point>"}]}
{"id": 54, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_38/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_39/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[408, 508], [277, 567], [202, 626], [113, 786]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[374, 340], [391, 462], [416, 592], [437, 752]]</point>"}]}
{"id": 56, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_39/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_42/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[555, 718], [433, 752], [298, 782], [101, 811]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the white plate."}, {"from": "gpt", "value": "<point>[[466, 294], [437, 466], [424, 643], [412, 794]]</point>"}]}
{"id": 58, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_42/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_61/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[235, 508], [118, 605], [67, 651], [34, 752]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[357, 286], [424, 412], [487, 538], [521, 668]]</point>"}]}
{"id": 60, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_61/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_62/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[584, 601], [429, 660], [277, 718], [109, 777]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[346, 329], [434, 427], [487, 549], [513, 722]]</point>"}]}
{"id": 62, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_62/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_66/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[402, 673], [331, 714], [216, 756], [100, 776]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[378, 340], [466, 487], [559, 622], [630, 769]]</point>"}]}
{"id": 64, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_66/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_70/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[546, 727], [429, 769], [303, 819], [155, 853]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[380, 417], [421, 498], [444, 598], [455, 737]]</point>"}]}
{"id": 66, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_70/frame_15.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_72/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the white plate on the red plate."}, {"from": "gpt", "value": "<point>[[368, 588], [333, 662], [246, 729], [130, 782]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[412, 332], [445, 450], [454, 534], [462, 639]]</point>"}]}
{"id": 68, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_73/frame_0.png", "rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_78/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[353, 408], [366, 584], [370, 723], [374, 836]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the white plate."}, {"from": "gpt", "value": "<point>[[487, 336], [450, 559], [399, 685], [336, 853]]</point>"}]}
{"id": 70, "image": ["rtx_frames_success_0/11_utokyo_xarm_pick_and_place_converted_externally_to_rlds#episode_80/frame_0.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_1/frame_0.png"], "height": [224, 480], "width": [224, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the white plate."}, {"from": "gpt", "value": "<point>[[521, 445], [479, 559], [437, 630], [387, 718]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[342, 697], [447, 642], [581, 612], [699, 627]]</point>"}]}
{"id": 72, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_15/frame_0.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_15/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[336, 697], [464, 599], [596, 454], [695, 273]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to open the cabinet door."}, {"from": "gpt", "value": "<point>[[703, 276], [673, 303], [630, 313], [596, 296]]</point>"}]}
{"id": 74, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_16/frame_0.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_24/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[350, 699], [436, 762], [581, 822], [739, 852]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[338, 652], [370, 679], [425, 694], [461, 682]]</point>"}]}
{"id": 76, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_24/frame_15.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_33/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the bowl in the below cabinet."}, {"from": "gpt", "value": "<point>[[461, 684], [506, 767], [570, 842], [656, 912]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[346, 714], [468, 617], [590, 446], [673, 278]]</point>"}]}
{"id": 78, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_39/frame_0.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_49/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[336, 712], [427, 774], [538, 817], [686, 855]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the teapot."}, {"from": "gpt", "value": "<point>[[333, 612], [481, 566], [628, 549], [789, 551]]</point>"}]}
{"id": 80, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_54/frame_0.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_54/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[325, 697], [380, 689], [423, 652], [462, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl in the below cabinet."}, {"from": "gpt", "value": "<point>[[517, 622], [564, 682], [600, 810], [598, 937]]</point>"}]}
{"id": 82, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_61/frame_0.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_61/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[346, 684], [451, 622], [603, 596], [786, 619]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl in the below cabinet."}, {"from": "gpt", "value": "<point>[[788, 617], [716, 672], [664, 779], [639, 915]]</point>"}]}
{"id": 84, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_79/frame_0.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_80/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[342, 697], [491, 554], [566, 456], [673, 278]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[336, 697], [402, 689], [477, 654], [541, 609]]</point>"}]}
{"id": 86, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_86/frame_0.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_114/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[327, 682], [477, 589], [613, 426], [694, 281]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[338, 702], [387, 649], [462, 617], [562, 612]]</point>"}]}
{"id": 88, "image": ["rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_114/frame_15.png", "rtx_frames_success_0/12_ucsd_kitchen_dataset_converted_externally_to_rlds#episode_132/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the bowl in the below cabinet."}, {"from": "gpt", "value": "<point>[[549, 551], [543, 662], [575, 817], [620, 930]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[346, 694], [417, 744], [555, 815], [690, 865]]</point>"}]}
{"id": 90, "image": ["rtx_frames_success_0/15_dlr_sara_grid_clamp_converted_externally_to_rlds#episode_64/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_15/frame_0.png"], "height": [480, 224], "width": [640, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the grid clamp."}, {"from": "gpt", "value": "<point>[[277, 746], [271, 757], [270, 774], [275, 786]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[193, 399], [231, 433], [261, 466], [269, 546]]</point>"}]}
{"id": 92, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_18/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_19/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[203, 437], [327, 459], [443, 474], [531, 502], [563, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[219, 435], [278, 465], [344, 492], [402, 553]]</point>"}]}
{"id": 94, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_41/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_49/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[214, 378], [315, 483], [395, 571], [454, 702]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the red object."}, {"from": "gpt", "value": "<point>[[222, 421], [254, 463], [272, 522], [276, 575]]</point>"}]}
{"id": 96, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_56/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_75/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[210, 382], [282, 458], [353, 538], [454, 706]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[210, 374], [235, 466], [252, 542], [273, 660]]</point>"}]}
{"id": 98, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_76/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_135/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[197, 370], [202, 458], [210, 567], [210, 685]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[197, 429], [246, 455], [313, 488], [415, 531], [476, 565], [522, 620], [537, 663]]</point>"}]}
{"id": 100, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_145/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_151/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[219, 425], [325, 463], [455, 547], [514, 620]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[256, 395], [340, 441], [416, 500], [504, 609]]</point>"}]}
{"id": 102, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_169/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_180/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red hexagonal prism."}, {"from": "gpt", "value": "<point>[[218, 403], [324, 487], [433, 576], [538, 693]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[215, 437], [232, 500], [252, 583], [246, 636]]</point>"}]}
{"id": 104, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_186/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_187/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[211, 437], [278, 455], [348, 488], [411, 522], [478, 559], [522, 602], [565, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[217, 429], [317, 486], [388, 531], [425, 583], [449, 634]]</point>"}]}
{"id": 106, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_195/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_251/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[209, 447], [244, 494], [278, 567], [283, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[211, 408], [368, 447], [492, 509], [624, 605]]</point>"}]}
{"id": 108, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_257/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_264/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[195, 414], [338, 470], [464, 541], [575, 658]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red block."}, {"from": "gpt", "value": "<point>[[214, 414], [274, 453], [297, 511], [297, 590]]</point>"}]}
{"id": 110, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_268/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_272/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[207, 414], [244, 459], [259, 508], [254, 583]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[201, 423], [195, 487], [211, 545], [244, 586]]</point>"}]}
{"id": 112, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_278/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_286/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[197, 409], [272, 470], [317, 531], [348, 583], [364, 648]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[207, 443], [287, 490], [339, 541], [372, 581], [392, 634], [396, 691]]</point>"}]}
{"id": 114, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_296/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_301/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[191, 421], [228, 478], [246, 528], [268, 589], [270, 650]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[217, 427], [327, 437], [445, 478], [516, 541], [569, 610]]</point>"}]}
{"id": 116, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_304/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_316/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[201, 435], [380, 445], [490, 480], [579, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[197, 427], [335, 453], [459, 478], [579, 553]]</point>"}]}
{"id": 118, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_325/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_337/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[199, 415], [236, 484], [238, 571], [220, 634]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[211, 415], [238, 506], [264, 587], [281, 634]]</point>"}]}
{"id": 120, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_340/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_349/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[215, 433], [281, 461], [348, 516], [382, 565], [390, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[164, 399], [336, 454], [458, 534], [588, 693]]</point>"}]}
{"id": 122, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_371/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_394/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[197, 374], [328, 408], [420, 441], [529, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red block."}, {"from": "gpt", "value": "<point>[[202, 387], [315, 458], [395, 550], [399, 685]]</point>"}]}
{"id": 124, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_424/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_442/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[222, 425], [344, 470], [463, 530], [557, 591], [612, 691]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[207, 419], [215, 457], [220, 514], [207, 585]]</point>"}]}
{"id": 126, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_470/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_471/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[207, 447], [358, 498], [474, 553], [518, 628]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[215, 439], [246, 492], [274, 547], [281, 602]]</point>"}]}
{"id": 128, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_490/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_509/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[205, 439], [281, 472], [396, 533], [431, 596]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[202, 366], [227, 445], [235, 559], [261, 655]]</point>"}]}
{"id": 130, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_511/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_516/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the red block."}, {"from": "gpt", "value": "<point>[[231, 349], [252, 408], [277, 441], [277, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[217, 421], [281, 453], [321, 478], [364, 518]]</point>"}]}
{"id": 132, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_526/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_537/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[227, 374], [298, 408], [370, 450], [479, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[222, 441], [333, 453], [421, 478], [516, 514], [539, 567]]</point>"}]}
{"id": 134, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_540/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_549/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[217, 421], [278, 447], [325, 476], [382, 512], [435, 553], [457, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[205, 404], [408, 461], [564, 534], [726, 658]]</point>"}]}
{"id": 136, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_551/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_553/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[199, 417], [274, 464], [335, 541], [370, 643]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[227, 382], [353, 420], [492, 466], [622, 559]]</point>"}]}
{"id": 138, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_555/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_573/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[162, 164], [199, 227], [235, 267], [301, 305]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the red object."}, {"from": "gpt", "value": "<point>[[206, 378], [303, 433], [403, 492], [525, 613]]</point>"}]}
{"id": 140, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_582/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_602/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[218, 400], [188, 459], [205, 508], [244, 553]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[212, 419], [256, 442], [282, 474], [282, 545]]</point>"}]}
{"id": 142, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_611/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_614/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[209, 427], [311, 467], [374, 537], [374, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[201, 437], [305, 467], [372, 498], [433, 531], [449, 594]]</point>"}]}
{"id": 144, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_615/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_619/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[213, 431], [299, 482], [384, 533], [482, 587], [541, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[207, 415], [301, 437], [376, 461], [449, 508]]</point>"}]}
{"id": 146, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_634/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_643/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[203, 439], [281, 476], [339, 528], [398, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[207, 421], [331, 486], [380, 569], [407, 659]]</point>"}]}
{"id": 148, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_674/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_674/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot on the stove."}, {"from": "gpt", "value": "<point>[[126, 177], [209, 211], [265, 263], [318, 344]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot on the stove."}, {"from": "gpt", "value": "<point>[[348, 333], [427, 314], [526, 318], [618, 342]]</point>"}]}
{"id": 150, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_678/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_689/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[223, 395], [319, 424], [361, 458], [399, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[206, 391], [328, 450], [429, 517], [546, 609]]</point>"}]}
{"id": 152, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_726/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_766/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[210, 370], [277, 424], [307, 496], [315, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the red object."}, {"from": "gpt", "value": "<point>[[211, 414], [376, 466], [528, 551], [694, 724]]</point>"}]}
{"id": 154, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_775/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_780/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red object."}, {"from": "gpt", "value": "<point>[[192, 414], [218, 519], [297, 613], [432, 688]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the red object."}, {"from": "gpt", "value": "<point>[[218, 432], [299, 427], [378, 449], [453, 496]]</point>"}]}
{"id": 156, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_780/frame_15.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_785/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the red object."}, {"from": "gpt", "value": "<point>[[472, 528], [549, 547], [600, 585], [635, 648]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[220, 410], [382, 432], [523, 491], [707, 615]]</point>"}]}
{"id": 158, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_788/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_800/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[218, 425], [323, 432], [449, 468], [560, 524]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[234, 448], [379, 477], [506, 523], [645, 639]]</point>"}]}
{"id": 160, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_814/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_833/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[202, 387], [244, 466], [277, 550], [319, 676]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[201, 437], [262, 494], [299, 553], [325, 587]]</point>"}]}
{"id": 162, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_835/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_861/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[209, 427], [313, 514], [364, 557], [427, 610], [480, 679]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[215, 429], [232, 482], [256, 581], [246, 654]]</point>"}]}
{"id": 164, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_881/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_883/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[218, 425], [301, 468], [359, 551], [376, 675]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[199, 414], [323, 481], [415, 556], [504, 667]]</point>"}]}
{"id": 166, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_919/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_922/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[195, 419], [289, 444], [352, 483], [414, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[207, 425], [305, 459], [376, 521], [444, 609]]</point>"}]}
{"id": 168, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_926/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_942/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot on the stove."}, {"from": "gpt", "value": "<point>[[139, 139], [202, 189], [244, 235], [298, 298]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[193, 423], [262, 476], [305, 530], [343, 600], [368, 683]]</point>"}]}
{"id": 170, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_946/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_950/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[201, 429], [325, 453], [437, 510], [520, 581], [585, 665]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[195, 427], [323, 459], [407, 492], [474, 563]]</point>"}]}
{"id": 172, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_962/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_972/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[197, 439], [319, 445], [409, 472], [455, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[197, 439], [323, 467], [461, 516], [600, 565], [693, 622]]</point>"}]}
{"id": 174, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_974/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_983/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[222, 433], [376, 474], [455, 502], [526, 531], [581, 565], [616, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[220, 421], [281, 486], [309, 573], [333, 632]]</point>"}]}
{"id": 176, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1003/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1008/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[189, 415], [246, 465], [303, 508], [333, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[205, 417], [335, 492], [421, 559], [470, 687]]</point>"}]}
{"id": 178, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1044/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1056/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[185, 378], [248, 458], [282, 517], [298, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[193, 366], [315, 424], [403, 483], [496, 601]]</point>"}]}
{"id": 180, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1071/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1103/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[202, 374], [340, 458], [471, 542], [580, 702]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[213, 435], [321, 465], [394, 514], [457, 596]]</point>"}]}
{"id": 182, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1117/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1122/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[193, 366], [223, 462], [256, 525], [294, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot on the stove."}, {"from": "gpt", "value": "<point>[[134, 147], [189, 206], [210, 235], [244, 303]]</point>"}]}
{"id": 184, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1126/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1134/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[209, 410], [289, 451], [359, 521], [398, 618]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red block."}, {"from": "gpt", "value": "<point>[[219, 437], [319, 437], [411, 467], [486, 514], [577, 593]]</point>"}]}
{"id": 186, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1140/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1152/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[172, 408], [303, 483], [437, 559], [643, 676]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[214, 374], [214, 437], [214, 496], [231, 550]]</point>"}]}
{"id": 188, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1168/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1168/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot on the stove."}, {"from": "gpt", "value": "<point>[[124, 175], [199, 230], [246, 276], [260, 305]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pot in the sink."}, {"from": "gpt", "value": "<point>[[524, 289], [594, 311], [642, 352], [693, 415]]</point>"}]}
{"id": 190, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1173/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1196/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[213, 427], [289, 449], [360, 504], [388, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[199, 425], [307, 472], [406, 539], [476, 598], [516, 661]]</point>"}]}
{"id": 192, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1202/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1209/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[203, 423], [352, 467], [449, 528], [502, 628]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[193, 419], [268, 463], [319, 531], [352, 632]]</point>"}]}
{"id": 194, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1220/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1254/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red block."}, {"from": "gpt", "value": "<point>[[206, 387], [244, 450], [265, 500], [265, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[214, 387], [311, 458], [370, 550], [403, 714]]</point>"}]}
{"id": 196, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1263/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1284/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[235, 382], [315, 420], [378, 500], [382, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[197, 361], [332, 416], [437, 483], [534, 592]]</point>"}]}
{"id": 198, "image": ["rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1344/frame_0.png", "rtx_frames_success_0/19_ucsd_pick_and_place_dataset_converted_externally_to_rlds#episode_1348/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[220, 429], [201, 459], [197, 510], [199, 608]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[193, 433], [285, 484], [388, 547], [439, 612], [453, 679]]</point>"}]}
{"id": 200, "image": ["rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_1/frame_0.png", "rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_13/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[427, 982], [412, 772], [447, 388], [500, 173]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[427, 982], [357, 797], [316, 584], [327, 421]]</point>"}]}
{"id": 202, "image": ["rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_103/frame_0.png", "rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_107/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[232, 702], [211, 629], [223, 579], [261, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the trash can door."}, {"from": "gpt", "value": "<point>[[261, 657], [354, 500], [455, 421], [539, 382]]</point>"}]}
{"id": 204, "image": ["rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_115/frame_0.png", "rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_116/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[408, 992], [389, 878], [379, 732], [387, 628], [418, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[436, 992], [430, 927], [462, 837], [506, 782]]</point>"}]}
{"id": 206, "image": ["rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_127/frame_0.png", "rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_128/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[442, 992], [414, 912], [414, 812], [436, 717]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[429, 990], [453, 887], [536, 744], [620, 639]]</point>"}]}
{"id": 208, "image": ["rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_136/frame_0.png", "rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_144/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[427, 987], [461, 850], [532, 732], [626, 659]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[388, 955], [388, 899], [421, 820], [430, 747]]</point>"}]}
{"id": 210, "image": ["rtx_frames_success_0/25_nyu_door_opening_surprising_effectiveness#episode_148/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_25/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[434, 949], [383, 758], [333, 567], [299, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handles."}, {"from": "gpt", "value": "<point>[[472, 780], [463, 730], [443, 653], [423, 598]]</point>"}]}
{"id": 212, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_28/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_34/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handle."}, {"from": "gpt", "value": "<point>[[449, 826], [492, 680], [573, 541], [695, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[460, 780], [449, 649], [420, 525]]</point>"}]}
{"id": 214, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_42/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_44/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[452, 749], [408, 602], [347, 490], [301, 398], [266, 197]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handles."}, {"from": "gpt", "value": "<point>[[397, 969], [371, 876], [356, 776], [374, 680], [414, 510]]</point>"}]}
{"id": 216, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_49/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_60/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[463, 749], [437, 676], [405, 595], [365, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[446, 784], [420, 718], [388, 656], [347, 602]]</point>"}]}
{"id": 218, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_61/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_64/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[403, 977], [376, 865], [379, 737], [397, 660], [434, 564], [475, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[472, 741], [446, 595], [403, 471], [397, 317]]</point>"}]}
{"id": 220, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_67/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_70/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handle."}, {"from": "gpt", "value": "<point>[[469, 680], [463, 579], [463, 486], [414, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[365, 981], [342, 819], [350, 726], [417, 579]]</point>"}]}
{"id": 222, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_72/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_79/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[423, 819], [423, 683], [423, 571], [414, 467]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[449, 799], [504, 753], [565, 714], [614, 664]]</point>"}]}
{"id": 224, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_156/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_164/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[434, 896], [452, 822], [446, 737], [411, 699]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[484, 710], [458, 602], [426, 479], [350, 255]]</point>"}]}
{"id": 226, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_165/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_166/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[443, 846], [492, 726], [553, 687]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[397, 988], [403, 772], [414, 633], [501, 440], [611, 305]]</point>"}]}
{"id": 228, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_167/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_169/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[434, 784], [458, 668], [417, 606]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[437, 718], [437, 606], [455, 448], [489, 305], [515, 216]]</point>"}]}
{"id": 230, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_172/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_174/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[452, 826], [394, 722], [417, 622], [507, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[478, 757], [515, 660], [553, 575], [602, 521]]</point>"}]}
{"id": 232, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_175/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_192/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[411, 977], [362, 811], [336, 637], [333, 483], [397, 170]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[403, 799], [408, 645], [382, 517], [290, 340]]</point>"}]}
{"id": 234, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_196/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_197/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[414, 985], [426, 927], [458, 880], [495, 834], [536, 799], [597, 749]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handle."}, {"from": "gpt", "value": "<point>[[408, 861], [388, 753], [434, 645], [510, 602], [585, 645]]</point>"}]}
{"id": 236, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_201/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_202/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[460, 761], [475, 618], [507, 463], [542, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[463, 734], [492, 591], [524, 479], [573, 359]]</point>"}]}
{"id": 238, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_213/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_223/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[408, 977], [388, 903], [374, 865], [362, 741]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handle."}, {"from": "gpt", "value": "<point>[[446, 853], [458, 753], [434, 683]]</point>"}]}
{"id": 240, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_227/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_229/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[431, 803], [452, 718], [486, 664], [507, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handle."}, {"from": "gpt", "value": "<point>[[411, 965], [376, 819], [379, 653], [403, 602]]</point>"}]}
{"id": 242, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_230/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_234/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handle."}, {"from": "gpt", "value": "<point>[[449, 861], [489, 772], [547, 691], [588, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[420, 822], [388, 761], [388, 699], [437, 614]]</point>"}]}
{"id": 244, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_236/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_253/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[397, 915], [434, 826], [507, 757], [582, 753]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[388, 965], [359, 865], [345, 726], [353, 583], [466, 278]]</point>"}]}
{"id": 246, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_257/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_260/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[437, 911], [460, 857], [492, 784], [486, 691]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[423, 822], [397, 726], [397, 614]]</point>"}]}
{"id": 248, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_263/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_264/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[446, 811], [431, 633], [397, 548], [356, 467]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[449, 826], [481, 741], [455, 683]]</point>"}]}
{"id": 250, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_278/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_280/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[434, 807], [394, 718], [362, 672], [319, 637]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[458, 861], [420, 776], [359, 664], [321, 618]]</point>"}]}
{"id": 252, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_282/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_283/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[405, 973], [400, 826], [420, 672], [498, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[434, 822], [400, 753], [397, 633]]</point>"}]}
{"id": 254, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_284/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_292/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[458, 819], [426, 687], [400, 595], [368, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[460, 884], [440, 776], [469, 660], [550, 629]]</point>"}]}
{"id": 256, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_301/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_323/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[437, 815], [405, 730], [379, 618], [374, 533], [397, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[455, 753], [492, 699], [565, 664], [617, 637]]</point>"}]}
{"id": 258, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_328/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_333/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the cabinet door."}, {"from": "gpt", "value": "<point>[[460, 741], [452, 649], [446, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handles."}, {"from": "gpt", "value": "<point>[[434, 811], [365, 703], [391, 610]]</point>"}]}
{"id": 260, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_338/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_340/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[440, 834], [426, 780], [443, 695], [568, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[403, 977], [371, 896], [368, 807], [391, 753], [440, 710]]</point>"}]}
{"id": 262, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_367/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_368/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[397, 996], [382, 927], [342, 757], [327, 622], [347, 386]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[379, 973], [429, 876], [472, 815], [588, 757]]</point>"}]}
{"id": 264, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_369/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_392/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[408, 981], [376, 876], [374, 757], [388, 602], [458, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handles."}, {"from": "gpt", "value": "<point>[[426, 977], [382, 834], [388, 710], [417, 564], [507, 367]]</point>"}]}
{"id": 266, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_395/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_400/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handles."}, {"from": "gpt", "value": "<point>[[429, 996], [408, 903], [437, 718], [481, 614], [553, 533], [657, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[417, 958], [385, 865], [371, 768], [368, 645], [385, 521], [423, 417], [478, 332], [513, 309]]</point>"}]}
{"id": 268, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_406/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_409/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[458, 822], [443, 656], [423, 463], [394, 232]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[443, 830], [481, 761], [565, 680], [620, 622]]</point>"}]}
{"id": 270, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_411/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_415/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[414, 977], [391, 803], [382, 703], [382, 544], [400, 398], [437, 286], [504, 185]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[443, 830], [489, 749], [536, 695], [565, 614]]</point>"}]}
{"id": 272, "image": ["rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_423/frame_0.png", "rtx_frames_success_1/25_nyu_door_opening_surprising_effectiveness#episode_425/frame_0.png"], "height": [720, 720], "width": [960, 960], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the trash cabinet door."}, {"from": "gpt", "value": "<point>[[446, 776], [469, 680], [513, 571], [536, 490]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[417, 857], [376, 699], [394, 537], [463, 375]]</point>"}]}
{"id": 274, "image": ["rtx_frames_success_1/27_jaco_play#episode_12/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_14/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the sponge."}, {"from": "gpt", "value": "<point>[[439, 254], [500, 246], [572, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the apple."}, {"from": "gpt", "value": "<point>[[520, 133], [428, 197], [367, 318], [370, 506]]</point>"}]}
{"id": 276, "image": ["rtx_frames_success_1/27_jaco_play#episode_27/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_31/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow cup."}, {"from": "gpt", "value": "<point>[[910, 382], [812, 341], [702, 338], [616, 387]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the black bowl."}, {"from": "gpt", "value": "<point>[[231, 110], [390, 104], [552, 133], [717, 188], [879, 298]]</point>"}]}
{"id": 278, "image": ["rtx_frames_success_1/27_jaco_play#episode_35/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_39/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the apple."}, {"from": "gpt", "value": "<point>[[460, 434], [569, 376], [697, 376], [887, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the steak."}, {"from": "gpt", "value": "<point>[[121, 182], [309, 176], [419, 260], [543, 379], [656, 613]]</point>"}]}
{"id": 280, "image": ["rtx_frames_success_1/27_jaco_play#episode_57/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_64/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the soap in the blue container."}, {"from": "gpt", "value": "<point>[[497, 468], [671, 431], [789, 431], [910, 457]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the black bowl."}, {"from": "gpt", "value": "<point>[[962, 393], [858, 324], [757, 301], [630, 295], [552, 358]]</point>"}]}
{"id": 282, "image": ["rtx_frames_success_1/27_jaco_play#episode_104/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_104/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the apple."}, {"from": "gpt", "value": "<point>[[971, 332], [887, 240], [763, 179], [598, 116], [483, 90], [298, 124], [194, 246]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the apple towards the black bowl."}, {"from": "gpt", "value": "<point>[[483, 104], [355, 110], [182, 246]]</point>"}]}
{"id": 284, "image": ["rtx_frames_success_1/27_jaco_play#episode_109/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_118/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the butter stick."}, {"from": "gpt", "value": "<point>[[676, 471], [751, 434], [835, 408], [948, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the steak."}, {"from": "gpt", "value": "<point>[[578, 107], [592, 205], [604, 298], [610, 497]]</point>"}]}
{"id": 286, "image": ["rtx_frames_success_1/27_jaco_play#episode_139/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_151/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the butter stick."}, {"from": "gpt", "value": "<point>[[462, 396], [564, 382], [682, 393], [763, 436], [867, 503]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bowl."}, {"from": "gpt", "value": "<point>[[283, 107], [379, 223], [442, 370], [448, 474]]</point>"}]}
{"id": 288, "image": ["rtx_frames_success_1/27_jaco_play#episode_170/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_181/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green cup."}, {"from": "gpt", "value": "<point>[[572, 113], [561, 205], [532, 309], [491, 396], [410, 486]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the black bowl."}, {"from": "gpt", "value": "<point>[[876, 295], [905, 367], [905, 413]]</point>"}]}
{"id": 290, "image": ["rtx_frames_success_1/27_jaco_play#episode_194/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_203/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the tomato."}, {"from": "gpt", "value": "<point>[[399, 436], [405, 347], [445, 277], [491, 249]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the apple."}, {"from": "gpt", "value": "<point>[[919, 260], [757, 243], [587, 301], [477, 390], [460, 575]]</point>"}]}
{"id": 292, "image": ["rtx_frames_success_1/27_jaco_play#episode_218/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_227/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the steak."}, {"from": "gpt", "value": "<point>[[679, 384], [760, 399], [829, 439], [867, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[179, 246], [286, 269], [329, 358], [361, 488]]</point>"}]}
{"id": 294, "image": ["rtx_frames_success_1/27_jaco_play#episode_236/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_240/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the black bowl."}, {"from": "gpt", "value": "<point>[[587, 191], [691, 185], [812, 280], [908, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[168, 205], [257, 234], [321, 315], [379, 396], [416, 454], [451, 526], [471, 575]]</point>"}]}
{"id": 296, "image": ["rtx_frames_success_1/27_jaco_play#episode_251/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_259/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the black bowl."}, {"from": "gpt", "value": "<point>[[442, 231], [526, 263], [575, 280], [592, 370]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the green cup."}, {"from": "gpt", "value": "<point>[[578, 90], [558, 168], [523, 269], [506, 303]]</point>"}]}
{"id": 298, "image": ["rtx_frames_success_1/27_jaco_play#episode_263/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_265/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the orange."}, {"from": "gpt", "value": "<point>[[214, 171], [272, 237], [335, 335], [387, 451], [419, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the steak."}, {"from": "gpt", "value": "<point>[[616, 465], [763, 422], [928, 474]]</point>"}]}
{"id": 300, "image": ["rtx_frames_success_1/27_jaco_play#episode_271/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_271/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[566, 92], [564, 194], [543, 338], [517, 465], [451, 566]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[486, 306], [512, 393], [503, 480], [445, 569]]</point>"}]}
{"id": 302, "image": ["rtx_frames_success_1/27_jaco_play#episode_275/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_277/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the black bowl."}, {"from": "gpt", "value": "<point>[[483, 405], [581, 384], [685, 390], [844, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the orange."}, {"from": "gpt", "value": "<point>[[367, 408], [399, 275], [477, 237], [587, 243]]</point>"}]}
{"id": 304, "image": ["rtx_frames_success_1/27_jaco_play#episode_279/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_321/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the gray bowl in the oven."}, {"from": "gpt", "value": "<point>[[832, 237], [725, 188], [587, 139], [425, 116], [272, 127], [179, 136]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the soap."}, {"from": "gpt", "value": "<point>[[486, 92], [503, 162], [514, 223], [514, 298], [514, 384], [509, 477]]</point>"}]}
{"id": 306, "image": ["rtx_frames_success_1/27_jaco_play#episode_323/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_334/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the yellow cup."}, {"from": "gpt", "value": "<point>[[575, 98], [526, 194], [483, 283], [480, 379]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the steak."}, {"from": "gpt", "value": "<point>[[471, 358], [566, 408], [607, 451], [627, 512]]</point>"}]}
{"id": 308, "image": ["rtx_frames_success_1/27_jaco_play#episode_338/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_351/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bread."}, {"from": "gpt", "value": "<point>[[974, 361], [867, 370], [754, 416], [697, 468], [656, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bowl."}, {"from": "gpt", "value": "<point>[[879, 306], [749, 324], [627, 405], [561, 645]]</point>"}]}
{"id": 310, "image": ["rtx_frames_success_1/27_jaco_play#episode_357/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_369/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cooked meat."}, {"from": "gpt", "value": "<point>[[159, 228], [306, 223], [439, 277], [535, 353], [647, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the milk carton."}, {"from": "gpt", "value": "<point>[[980, 237], [861, 223], [720, 289], [645, 514]]</point>"}]}
{"id": 312, "image": ["rtx_frames_success_1/27_jaco_play#episode_373/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_381/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the butter stick."}, {"from": "gpt", "value": "<point>[[659, 185], [575, 277], [540, 373], [540, 512]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the black bowl into the blue sink."}, {"from": "gpt", "value": "<point>[[251, 136], [355, 87], [512, 101], [647, 136], [720, 197], [786, 251], [882, 338], [945, 451]]</point>"}]}
{"id": 314, "image": ["rtx_frames_success_1/27_jaco_play#episode_385/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_405/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the bowl."}, {"from": "gpt", "value": "<point>[[497, 283], [630, 179], [746, 133], [841, 142], [908, 168], [951, 188]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the gray bowl."}, {"from": "gpt", "value": "<point>[[257, 87], [465, 29], [607, 40], [702, 52], [818, 92], [939, 205]]</point>"}]}
{"id": 316, "image": ["rtx_frames_success_1/27_jaco_play#episode_407/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_407/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the bowl containing an orange into the blue container."}, {"from": "gpt", "value": "<point>[[355, 234], [517, 199], [743, 220], [867, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl into the sink."}, {"from": "gpt", "value": "<point>[[662, 121], [780, 211], [879, 425]]</point>"}]}
{"id": 318, "image": ["rtx_frames_success_1/27_jaco_play#episode_410/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_420/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sponge."}, {"from": "gpt", "value": "<point>[[867, 341], [725, 355], [659, 474], [618, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[572, 272], [503, 329], [451, 477], [434, 642]]</point>"}]}
{"id": 320, "image": ["rtx_frames_success_1/27_jaco_play#episode_421/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_425/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the butter stick."}, {"from": "gpt", "value": "<point>[[543, 321], [679, 335], [798, 376], [908, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the apple."}, {"from": "gpt", "value": "<point>[[607, 358], [760, 382], [910, 457]]</point>"}]}
{"id": 322, "image": ["rtx_frames_success_1/27_jaco_play#episode_437/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_439/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the sponge."}, {"from": "gpt", "value": "<point>[[538, 436], [656, 399], [751, 402], [887, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the sponge from the blue container."}, {"from": "gpt", "value": "<point>[[942, 344], [873, 275], [760, 257], [601, 309]]</point>"}]}
{"id": 324, "image": ["rtx_frames_success_1/27_jaco_play#episode_442/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_445/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the plate."}, {"from": "gpt", "value": "<point>[[558, 98], [561, 147], [569, 205], [575, 283], [590, 387], [595, 523]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the steak."}, {"from": "gpt", "value": "<point>[[621, 457], [720, 419], [815, 425], [936, 486]]</point>"}]}
{"id": 326, "image": ["rtx_frames_success_1/27_jaco_play#episode_447/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_448/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the yellow cup."}, {"from": "gpt", "value": "<point>[[884, 251], [818, 234], [691, 237], [621, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the steak to the left of the blue container."}, {"from": "gpt", "value": "<point>[[936, 347], [841, 425], [772, 488], [671, 520]]</point>"}]}
{"id": 328, "image": ["rtx_frames_success_1/27_jaco_play#episode_459/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_461/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the yellow cup."}, {"from": "gpt", "value": "<point>[[370, 263], [436, 246], [506, 251], [555, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the gray bowl."}, {"from": "gpt", "value": "<point>[[850, 98], [749, 136], [645, 199], [526, 301]]</point>"}]}
{"id": 330, "image": ["rtx_frames_success_1/27_jaco_play#episode_479/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_479/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the black bowl."}, {"from": "gpt", "value": "<point>[[263, 156], [390, 145], [509, 237], [613, 370]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the black bowl."}, {"from": "gpt", "value": "<point>[[540, 228], [578, 263], [610, 329], [616, 379]]</point>"}]}
{"id": 332, "image": ["rtx_frames_success_1/27_jaco_play#episode_485/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_488/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the bread."}, {"from": "gpt", "value": "<point>[[673, 500], [647, 390], [558, 335], [399, 301], [199, 292]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the green cup."}, {"from": "gpt", "value": "<point>[[410, 497], [503, 465], [546, 413], [564, 321]]</point>"}]}
{"id": 334, "image": ["rtx_frames_success_1/27_jaco_play#episode_492/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_493/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the apple."}, {"from": "gpt", "value": "<point>[[517, 249], [434, 338], [416, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the black bowl."}, {"from": "gpt", "value": "<point>[[197, 162], [277, 275], [344, 399], [370, 512]]</point>"}]}
{"id": 336, "image": ["rtx_frames_success_1/27_jaco_play#episode_494/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_499/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the butter."}, {"from": "gpt", "value": "<point>[[575, 260], [642, 327], [650, 431], [642, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the apple."}, {"from": "gpt", "value": "<point>[[584, 303], [517, 338], [480, 442], [488, 592]]</point>"}]}
{"id": 338, "image": ["rtx_frames_success_1/27_jaco_play#episode_500/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_511/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[621, 133], [737, 142], [853, 199], [936, 298]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[925, 347], [812, 306], [671, 289], [494, 315], [416, 408]]</point>"}]}
{"id": 340, "image": ["rtx_frames_success_1/27_jaco_play#episode_523/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_523/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the orange in a bowl."}, {"from": "gpt", "value": "<point>[[916, 355], [809, 315], [650, 329], [491, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the orange in a gray bowl."}, {"from": "gpt", "value": "<point>[[699, 220], [642, 228], [549, 289], [500, 390]]</point>"}]}
{"id": 342, "image": ["rtx_frames_success_1/27_jaco_play#episode_527/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_528/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the burger patty."}, {"from": "gpt", "value": "<point>[[627, 298], [402, 194], [194, 231]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the green cup."}, {"from": "gpt", "value": "<point>[[899, 361], [772, 338], [601, 338], [416, 402]]</point>"}]}
{"id": 344, "image": ["rtx_frames_success_1/27_jaco_play#episode_556/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_558/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bread."}, {"from": "gpt", "value": "<point>[[110, 104], [194, 113], [367, 182], [575, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the dish rack while holding the green cup."}, {"from": "gpt", "value": "<point>[[509, 176], [572, 142], [671, 139], [789, 159], [931, 223]]</point>"}]}
{"id": 346, "image": ["rtx_frames_success_1/27_jaco_play#episode_563/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_567/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green cup."}, {"from": "gpt", "value": "<point>[[827, 81], [746, 87], [653, 133], [590, 182], [532, 266], [503, 329], [497, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the apple."}, {"from": "gpt", "value": "<point>[[584, 462], [480, 457], [376, 592]]</point>"}]}
{"id": 348, "image": ["rtx_frames_success_1/27_jaco_play#episode_580/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_586/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the plate while holding the apple."}, {"from": "gpt", "value": "<point>[[434, 350], [503, 338], [566, 327], [590, 335]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the gray bowl."}, {"from": "gpt", "value": "<point>[[569, 104], [555, 171], [543, 246], [526, 329]]</point>"}]}
{"id": 350, "image": ["rtx_frames_success_1/27_jaco_play#episode_594/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_598/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green cup."}, {"from": "gpt", "value": "<point>[[532, 332], [584, 390], [604, 445], [616, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the dairy product."}, {"from": "gpt", "value": "<point>[[422, 251], [514, 272], [587, 338], [633, 503]]</point>"}]}
{"id": 352, "image": ["rtx_frames_success_1/27_jaco_play#episode_608/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_621/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green cup."}, {"from": "gpt", "value": "<point>[[240, 422], [318, 451], [387, 506], [436, 575], [454, 613]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green cup."}, {"from": "gpt", "value": "<point>[[977, 399], [887, 312], [789, 292], [636, 280], [494, 321]]</point>"}]}
{"id": 354, "image": ["rtx_frames_success_1/27_jaco_play#episode_624/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_629/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the oven."}, {"from": "gpt", "value": "<point>[[512, 168], [422, 162], [327, 182], [260, 217]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the steak."}, {"from": "gpt", "value": "<point>[[205, 133], [344, 162], [460, 237], [558, 396]]</point>"}]}
{"id": 356, "image": ["rtx_frames_success_1/27_jaco_play#episode_630/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_632/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[581, 92], [590, 194], [598, 283], [587, 393]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the breadstick."}, {"from": "gpt", "value": "<point>[[656, 549], [595, 451], [497, 367], [402, 324], [301, 280], [202, 269]]</point>"}]}
{"id": 358, "image": ["rtx_frames_success_1/27_jaco_play#episode_632/frame_15.png", "rtx_frames_success_1/27_jaco_play#episode_633/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the breadstick towards the toaster oven."}, {"from": "gpt", "value": "<point>[[350, 301], [272, 263], [197, 251], [147, 251]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the soap."}, {"from": "gpt", "value": "<point>[[332, 344], [465, 445], [543, 532], [621, 621]]</point>"}]}
{"id": 360, "image": ["rtx_frames_success_1/27_jaco_play#episode_637/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_640/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[844, 75], [861, 153], [879, 266], [855, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the milk carton."}, {"from": "gpt", "value": "<point>[[584, 448], [535, 408], [465, 408], [436, 425]]</point>"}]}
{"id": 362, "image": ["rtx_frames_success_1/27_jaco_play#episode_645/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_649/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green cup."}, {"from": "gpt", "value": "<point>[[812, 145], [702, 185], [592, 246], [483, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the butter stick."}, {"from": "gpt", "value": "<point>[[957, 413], [887, 344], [775, 315], [653, 315], [613, 321]]</point>"}]}
{"id": 364, "image": ["rtx_frames_success_1/27_jaco_play#episode_652/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_652/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the green cup."}, {"from": "gpt", "value": "<point>[[850, 40], [754, 58], [584, 81], [471, 98], [387, 130], [208, 182], [142, 191]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the green cup in the oven."}, {"from": "gpt", "value": "<point>[[410, 98], [315, 118], [185, 130], [116, 165]]</point>"}]}
{"id": 366, "image": ["rtx_frames_success_1/27_jaco_play#episode_662/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_664/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the apple."}, {"from": "gpt", "value": "<point>[[566, 113], [561, 211], [566, 321], [578, 422], [584, 497], [566, 621]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the soap in the blue container."}, {"from": "gpt", "value": "<point>[[558, 448], [659, 393], [731, 390], [824, 434]]</point>"}]}
{"id": 368, "image": ["rtx_frames_success_1/27_jaco_play#episode_666/frame_0.png", "rtx_frames_success_1/27_jaco_play#episode_666/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the gray bowl."}, {"from": "gpt", "value": "<point>[[240, 72], [419, 69], [621, 95], [743, 185], [902, 327]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl in the blue container."}, {"from": "gpt", "value": "<point>[[685, 49], [792, 142], [873, 275], [922, 382]]</point>"}]}
{"id": 370, "image": ["rtx_frames_success_2/3_utokyo_xarm_bimanual_converted_externally_to_rlds#episode_9/frame_0.png", "rtx_frames_success_2/3_utokyo_xarm_bimanual_converted_externally_to_rlds#episode_13/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[314, 500], [322, 587], [337, 665], [356, 735]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[218, 487], [237, 555], [261, 619], [301, 689]]</point>"}]}
{"id": 372, "image": ["rtx_frames_success_2/3_utokyo_xarm_bimanual_converted_externally_to_rlds#episode_13/frame_15.png", "rtx_frames_success_2/3_utokyo_xarm_bimanual_converted_externally_to_rlds#episode_22/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grip the towel."}, {"from": "gpt", "value": "<point>[[316, 665], [265, 659], [214, 644], [174, 623]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[294, 434], [282, 521], [282, 621], [286, 725]]</point>"}]}
{"id": 374, "image": ["rtx_frames_success_2/3_utokyo_xarm_bimanual_converted_externally_to_rlds#episode_36/frame_0.png", "rtx_frames_success_2/3_utokyo_xarm_bimanual_converted_externally_to_rlds#episode_36/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grip the towel at two diagonal corners."}, {"from": "gpt", "value": "<point>[[309, 498], [328, 562], [347, 640], [362, 712]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to stretch the towel by moving the grippers apart."}, {"from": "gpt", "value": "<point>[[343, 701], [288, 691], [248, 674], [205, 648]]</point>"}]}
{"id": 376, "image": ["rtx_frames_success_2/3_utokyo_xarm_bimanual_converted_externally_to_rlds#episode_53/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_73/frame_0.png"], "height": [256, 224], "width": [256, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grip the towel."}, {"from": "gpt", "value": "<point>[[286, 432], [278, 540], [278, 646], [295, 754]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the butter container."}, {"from": "gpt", "value": "<point>[[650, 441], [701, 432], [752, 436], [805, 445]]</point>"}]}
{"id": 378, "image": ["rtx_frames_success_2/27_jaco_play#episode_73/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_83/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the butter container in the sink."}, {"from": "gpt", "value": "<point>[[812, 403], [867, 411], [898, 441], [915, 487]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the butter stick."}, {"from": "gpt", "value": "<point>[[646, 284], [695, 354], [725, 398], [754, 481]]</point>"}]}
{"id": 380, "image": ["rtx_frames_success_2/27_jaco_play#episode_83/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_85/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the target location."}, {"from": "gpt", "value": "<point>[[735, 451], [689, 498], [648, 553], [616, 614]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the orange."}, {"from": "gpt", "value": "<point>[[653, 341], [562, 428], [477, 513], [388, 623]]</point>"}]}
{"id": 382, "image": ["rtx_frames_success_2/27_jaco_play#episode_98/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_98/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the green cup."}, {"from": "gpt", "value": "<point>[[208, 170], [360, 155], [527, 150], [691, 148]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the green cup towards the dish rack."}, {"from": "gpt", "value": "<point>[[678, 104], [777, 152], [862, 212], [956, 277]]</point>"}]}
{"id": 384, "image": ["rtx_frames_success_2/27_jaco_play#episode_671/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_676/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cooked patty."}, {"from": "gpt", "value": "<point>[[157, 212], [312, 208], [449, 216], [606, 237]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the plate."}, {"from": "gpt", "value": "<point>[[951, 91], [841, 148], [722, 225], [625, 299]]</point>"}]}
{"id": 386, "image": ["rtx_frames_success_2/27_jaco_play#episode_679/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_679/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the green cup."}, {"from": "gpt", "value": "<point>[[653, 239], [562, 222], [460, 214], [362, 212]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the green cup in the oven."}, {"from": "gpt", "value": "<point>[[328, 172], [259, 188], [197, 220], [142, 261]]</point>"}]}
{"id": 388, "image": ["rtx_frames_success_2/27_jaco_play#episode_683/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_683/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the butter stick."}, {"from": "gpt", "value": "<point>[[545, 258], [585, 288], [619, 333], [631, 381]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the butter stick towards the bowl."}, {"from": "gpt", "value": "<point>[[661, 432], [661, 506], [669, 591], [684, 691]]</point>"}]}
{"id": 390, "image": ["rtx_frames_success_2/27_jaco_play#episode_686/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_690/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sponge."}, {"from": "gpt", "value": "<point>[[987, 311], [979, 356], [968, 402], [947, 443]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the gray bowl."}, {"from": "gpt", "value": "<point>[[873, 356], [748, 318], [598, 294], [445, 286]]</point>"}]}
{"id": 392, "image": ["rtx_frames_success_2/27_jaco_play#episode_690/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_692/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the gray bowl towards the oven."}, {"from": "gpt", "value": "<point>[[422, 229], [314, 231], [195, 248], [97, 267]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the dish rack."}, {"from": "gpt", "value": "<point>[[479, 348], [598, 314], [720, 273], [835, 212]]</point>"}]}
{"id": 394, "image": ["rtx_frames_success_2/27_jaco_play#episode_692/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_696/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the dish rack."}, {"from": "gpt", "value": "<point>[[837, 193], [875, 212], [920, 244], [949, 273]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the slice of bread on the toaster tray."}, {"from": "gpt", "value": "<point>[[150, 150], [225, 182], [292, 223], [350, 273]]</point>"}]}
{"id": 396, "image": ["rtx_frames_success_2/27_jaco_play#episode_696/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_706/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the slice of bread on the toaster tray."}, {"from": "gpt", "value": "<point>[[339, 286], [403, 343], [462, 417], [509, 485]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the soap."}, {"from": "gpt", "value": "<point>[[576, 98], [562, 184], [562, 271], [570, 356]]</point>"}]}
{"id": 398, "image": ["rtx_frames_success_2/27_jaco_play#episode_706/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_707/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the soap."}, {"from": "gpt", "value": "<point>[[587, 358], [576, 417], [580, 502], [589, 574]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the gray bowl."}, {"from": "gpt", "value": "<point>[[953, 184], [831, 184], [669, 193], [547, 208]]</point>"}]}
{"id": 400, "image": ["rtx_frames_success_2/27_jaco_play#episode_707/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_713/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the gray bowl."}, {"from": "gpt", "value": "<point>[[574, 161], [438, 191], [324, 223], [203, 265]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the yellow cup."}, {"from": "gpt", "value": "<point>[[562, 208], [502, 205], [438, 208], [383, 220]]</point>"}]}
{"id": 402, "image": ["rtx_frames_success_2/27_jaco_play#episode_713/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_718/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the yellow cup towards the oven."}, {"from": "gpt", "value": "<point>[[377, 182], [305, 195], [242, 223], [176, 265]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the black bowl."}, {"from": "gpt", "value": "<point>[[362, 398], [523, 409], [674, 428], [828, 453]]</point>"}]}
{"id": 404, "image": ["rtx_frames_success_2/27_jaco_play#episode_721/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_724/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[394, 299], [384, 350], [379, 407], [384, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the milk carton."}, {"from": "gpt", "value": "<point>[[146, 81], [309, 170], [436, 227], [604, 282]]</point>"}]}
{"id": 406, "image": ["rtx_frames_success_2/27_jaco_play#episode_740/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_754/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brown bowl."}, {"from": "gpt", "value": "<point>[[572, 271], [475, 239], [384, 218], [280, 205]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the green cup."}, {"from": "gpt", "value": "<point>[[843, 142], [820, 167], [805, 197], [790, 237]]</point>"}]}
{"id": 408, "image": ["rtx_frames_success_2/27_jaco_play#episode_758/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_758/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the butter."}, {"from": "gpt", "value": "<point>[[903, 335], [741, 273], [557, 206], [383, 161]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the butter."}, {"from": "gpt", "value": "<point>[[381, 110], [307, 152], [254, 193], [195, 246]]</point>"}]}
{"id": 410, "image": ["rtx_frames_success_2/27_jaco_play#episode_760/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_772/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the orange."}, {"from": "gpt", "value": "<point>[[384, 434], [445, 392], [513, 350], [580, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the soap."}, {"from": "gpt", "value": "<point>[[591, 93], [570, 233], [553, 390], [547, 532]]</point>"}]}
{"id": 412, "image": ["rtx_frames_success_2/27_jaco_play#episode_780/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_780/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the gray bowl."}, {"from": "gpt", "value": "<point>[[566, 205], [650, 193], [761, 191], [862, 210]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the gray bowl towards the sink."}, {"from": "gpt", "value": "<point>[[894, 216], [938, 284], [972, 352], [1000, 434]]</point>"}]}
{"id": 414, "image": ["rtx_frames_success_2/27_jaco_play#episode_782/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_786/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cardboard circle."}, {"from": "gpt", "value": "<point>[[892, 119], [820, 170], [729, 246], [657, 316]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the butter stick."}, {"from": "gpt", "value": "<point>[[462, 246], [473, 303], [479, 381], [475, 449]]</point>"}]}
{"id": 416, "image": ["rtx_frames_success_2/27_jaco_play#episode_802/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_806/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the orange."}, {"from": "gpt", "value": "<point>[[574, 38], [549, 146], [534, 269], [528, 375]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the apple."}, {"from": "gpt", "value": "<point>[[905, 345], [786, 324], [687, 318], [561, 322]]</point>"}]}
{"id": 418, "image": ["rtx_frames_success_2/27_jaco_play#episode_818/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_826/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the butter."}, {"from": "gpt", "value": "<point>[[795, 402], [739, 481], [691, 574], [665, 672]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sponge."}, {"from": "gpt", "value": "<point>[[777, 307], [676, 273], [597, 256], [494, 252]]</point>"}]}
{"id": 420, "image": ["rtx_frames_success_2/27_jaco_play#episode_826/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_829/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sponge."}, {"from": "gpt", "value": "<point>[[483, 182], [456, 250], [438, 320], [432, 396]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the burger patty from the plate."}, {"from": "gpt", "value": "<point>[[580, 261], [523, 322], [479, 384], [434, 458]]</point>"}]}
{"id": 422, "image": ["rtx_frames_success_2/27_jaco_play#episode_860/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_866/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green cup."}, {"from": "gpt", "value": "<point>[[441, 180], [386, 170], [331, 169], [265, 174]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the apple."}, {"from": "gpt", "value": "<point>[[352, 455], [381, 394], [417, 345], [455, 311]]</point>"}]}
{"id": 424, "image": ["rtx_frames_success_2/27_jaco_play#episode_873/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_873/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the gray bowl."}, {"from": "gpt", "value": "<point>[[53, 51], [121, 61], [193, 61], [278, 53]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the gray bowl."}, {"from": "gpt", "value": "<point>[[261, 136], [294, 169], [311, 203], [318, 244]]</point>"}]}
{"id": 426, "image": ["rtx_frames_success_2/27_jaco_play#episode_883/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_886/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the orange."}, {"from": "gpt", "value": "<point>[[612, 453], [597, 390], [598, 320], [608, 273]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the black bowl."}, {"from": "gpt", "value": "<point>[[117, 61], [314, 95], [508, 133], [684, 176]]</point>"}]}
{"id": 428, "image": ["rtx_frames_success_2/27_jaco_play#episode_886/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_888/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the black bowl."}, {"from": "gpt", "value": "<point>[[689, 152], [737, 218], [795, 326], [822, 398]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bread."}, {"from": "gpt", "value": "<point>[[585, 445], [502, 362], [422, 290], [339, 231]]</point>"}]}
{"id": 430, "image": ["rtx_frames_success_2/27_jaco_play#episode_888/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_894/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bread towards the oven."}, {"from": "gpt", "value": "<point>[[347, 188], [294, 178], [237, 186], [184, 203]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the butter."}, {"from": "gpt", "value": "<point>[[417, 394], [559, 388], [670, 394], [812, 409]]</point>"}]}
{"id": 432, "image": ["rtx_frames_success_2/27_jaco_play#episode_894/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_900/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the butter."}, {"from": "gpt", "value": "<point>[[805, 345], [835, 390], [854, 437], [866, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the black bowl."}, {"from": "gpt", "value": "<point>[[259, 176], [358, 176], [451, 186], [536, 203]]</point>"}]}
{"id": 434, "image": ["rtx_frames_success_2/27_jaco_play#episode_900/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_907/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bowl over the table."}, {"from": "gpt", "value": "<point>[[523, 170], [583, 229], [638, 299], [684, 375]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the milk carton."}, {"from": "gpt", "value": "<point>[[402, 269], [528, 225], [640, 201], [756, 178]]</point>"}]}
{"id": 436, "image": ["rtx_frames_success_2/27_jaco_play#episode_907/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_910/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the milk carton towards the gray bowl."}, {"from": "gpt", "value": "<point>[[706, 174], [792, 186], [875, 205], [943, 241]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bread."}, {"from": "gpt", "value": "<point>[[498, 159], [547, 231], [593, 311], [619, 394]]</point>"}]}
{"id": 438, "image": ["rtx_frames_success_2/27_jaco_play#episode_914/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_922/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower the apple towards the table."}, {"from": "gpt", "value": "<point>[[509, 337], [473, 383], [443, 441], [422, 508]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bread slice."}, {"from": "gpt", "value": "<point>[[669, 453], [545, 384], [430, 316], [271, 256]]</point>"}]}
{"id": 440, "image": ["rtx_frames_success_2/27_jaco_play#episode_922/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_929/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bread slice towards the toaster oven."}, {"from": "gpt", "value": "<point>[[307, 201], [256, 227], [218, 259], [172, 299]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the milk carton."}, {"from": "gpt", "value": "<point>[[600, 8], [545, 106], [491, 220], [449, 356]]</point>"}]}
{"id": 442, "image": ["rtx_frames_success_2/27_jaco_play#episode_929/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_932/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the milk carton."}, {"from": "gpt", "value": "<point>[[468, 282], [447, 341], [438, 390], [438, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the soap."}, {"from": "gpt", "value": "<point>[[934, 345], [833, 341], [750, 345], [636, 352]]</point>"}]}
{"id": 444, "image": ["rtx_frames_success_2/27_jaco_play#episode_932/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_936/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the block in front of a green cup."}, {"from": "gpt", "value": "<point>[[634, 341], [566, 386], [472, 458], [402, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the milk carton."}, {"from": "gpt", "value": "<point>[[235, 127], [468, 188], [676, 227], [888, 250]]</point>"}]}
{"id": 446, "image": ["rtx_frames_success_2/27_jaco_play#episode_936/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_944/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the milk carton."}, {"from": "gpt", "value": "<point>[[867, 233], [898, 303], [905, 345], [900, 411]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the carton."}, {"from": "gpt", "value": "<point>[[473, 492], [504, 415], [545, 348], [595, 290]]</point>"}]}
{"id": 448, "image": ["rtx_frames_success_2/27_jaco_play#episode_948/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_948/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow cup."}, {"from": "gpt", "value": "<point>[[170, 223], [284, 214], [379, 218], [492, 227]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow cup."}, {"from": "gpt", "value": "<point>[[479, 201], [504, 246], [519, 290], [527, 347]]</point>"}]}
{"id": 450, "image": ["rtx_frames_success_2/27_jaco_play#episode_952/frame_15.png", "rtx_frames_success_2/27_jaco_play#episode_954/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the steak towards the bowl."}, {"from": "gpt", "value": "<point>[[468, 350], [536, 326], [612, 303], [688, 290]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the burger patty."}, {"from": "gpt", "value": "<point>[[403, 439], [366, 379], [316, 328], [252, 294]]</point>"}]}
{"id": 452, "image": ["rtx_frames_success_2/27_jaco_play#episode_959/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_971/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the gray bowl."}, {"from": "gpt", "value": "<point>[[453, 413], [612, 337], [778, 277], [977, 214]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the dish rack."}, {"from": "gpt", "value": "<point>[[578, 241], [678, 223], [752, 220], [847, 225]]</point>"}]}
{"id": 454, "image": ["rtx_frames_success_2/27_jaco_play#episode_975/frame_0.png", "rtx_frames_success_2/27_jaco_play#episode_975/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bread."}, {"from": "gpt", "value": "<point>[[956, 367], [790, 326], [583, 290], [379, 273]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bread."}, {"from": "gpt", "value": "<point>[[362, 189], [292, 223], [231, 267], [182, 312]]</point>"}]}
{"id": 456, "image": ["rtx_frames_success_2/49_bridge#episode_1039/frame_0.png", "rtx_frames_success_2/49_bridge#episode_1040/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[415, 250], [430, 338], [438, 444], [438, 540]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[97, 263], [252, 391], [384, 510], [509, 646]]</point>"}]}
{"id": 458, "image": ["rtx_frames_success_2/49_bridge#episode_1057/frame_0.png", "rtx_frames_success_2/49_bridge#episode_1060/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the metal pot."}, {"from": "gpt", "value": "<point>[[400, 10], [411, 141], [415, 275], [405, 404]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spatula."}, {"from": "gpt", "value": "<point>[[358, 260], [472, 341], [587, 419], [729, 508]]</point>"}]}
{"id": 460, "image": ["rtx_frames_success_2/49_bridge#episode_1060/frame_15.png", "rtx_frames_success_2/49_bridge#episode_1071/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the bowl."}, {"from": "gpt", "value": "<point>[[566, 268], [464, 298], [366, 356], [275, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple towel."}, {"from": "gpt", "value": "<point>[[525, 5], [502, 116], [487, 260], [479, 381]]</point>"}]}
{"id": 462, "image": ["rtx_frames_success_2/49_bridge#episode_1080/frame_0.png", "rtx_frames_success_2/49_bridge#episode_1082/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the hot sauce bottle."}, {"from": "gpt", "value": "<point>[[725, 338], [699, 422], [676, 551], [659, 682]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[216, 242], [362, 328], [475, 424], [578, 543]]</point>"}]}
{"id": 464, "image": ["rtx_frames_success_2/49_bridge#episode_1082/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10002/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[551, 480], [475, 439], [409, 442], [347, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[278, 71], [407, 164], [549, 306], [657, 427]]</point>"}]}
{"id": 466, "image": ["rtx_frames_success_2/49_bridge#episode_10012/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10016/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[267, 146], [390, 336], [491, 528], [576, 730]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[670, 250], [612, 306], [566, 369], [525, 434]]</point>"}]}
{"id": 468, "image": ["rtx_frames_success_2/49_bridge#episode_10058/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10058/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[528, 366], [464, 333], [403, 313], [333, 293]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the sushi roll towards the bowl."}, {"from": "gpt", "value": "<point>[[335, 247], [483, 308], [610, 379], [737, 470]]</point>"}]}
{"id": 470, "image": ["rtx_frames_success_2/49_bridge#episode_10060/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10061/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue towel."}, {"from": "gpt", "value": "<point>[[400, 247], [320, 295], [250, 359], [174, 439]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[231, 76], [403, 212], [542, 328], [703, 487]]</point>"}]}
{"id": 472, "image": ["rtx_frames_success_2/49_bridge#episode_10063/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10063/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the pan."}, {"from": "gpt", "value": "<point>[[545, 409], [434, 404], [337, 414], [216, 434]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pan in front of the microwave."}, {"from": "gpt", "value": "<point>[[195, 419], [252, 379], [307, 323], [345, 263]]</point>"}]}
{"id": 474, "image": ["rtx_frames_success_2/49_bridge#episode_10090/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10092/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[188, 121], [258, 331], [347, 505], [453, 639]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[434, 465], [523, 457], [597, 434], [678, 394]]</point>"}]}
{"id": 476, "image": ["rtx_frames_success_2/49_bridge#episode_10092/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10103/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mushroom over the microwave."}, {"from": "gpt", "value": "<point>[[373, 93], [339, 141], [307, 192], [292, 242]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[229, 177], [326, 298], [411, 434], [517, 626]]</point>"}]}
{"id": 478, "image": ["rtx_frames_success_2/49_bridge#episode_10107/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10108/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[718, 253], [580, 348], [456, 475], [335, 611]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[456, 38], [426, 144], [405, 263], [392, 376]]</point>"}]}
{"id": 480, "image": ["rtx_frames_success_2/49_bridge#episode_10110/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10119/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[42, 126], [212, 258], [350, 374], [530, 540]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[282, 197], [242, 278], [201, 384], [172, 490]]</point>"}]}
{"id": 482, "image": ["rtx_frames_success_2/49_bridge#episode_10123/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10138/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[517, 207], [441, 237], [371, 288], [316, 348]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[328, 222], [473, 240], [597, 275], [752, 359]]</point>"}]}
{"id": 484, "image": ["rtx_frames_success_2/49_bridge#episode_10138/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10152/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the can towards the fork and spoon."}, {"from": "gpt", "value": "<point>[[672, 359], [604, 381], [545, 447], [502, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[612, 141], [572, 242], [536, 376], [508, 520]]</point>"}]}
{"id": 486, "image": ["rtx_frames_success_2/49_bridge#episode_10160/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10163/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[223, 207], [261, 321], [316, 424], [383, 513]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[873, 444], [725, 389], [572, 343], [377, 303]]</point>"}]}
{"id": 488, "image": ["rtx_frames_success_2/49_bridge#episode_10174/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10178/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the frog."}, {"from": "gpt", "value": "<point>[[466, 492], [551, 465], [629, 449], [710, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the green cloth."}, {"from": "gpt", "value": "<point>[[244, 260], [277, 381], [314, 477], [371, 586]]</point>"}]}
{"id": 490, "image": ["rtx_frames_success_2/49_bridge#episode_10179/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10179/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[373, 98], [314, 232], [261, 384], [231, 515]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[237, 515], [362, 566], [464, 626], [580, 712]]</point>"}]}
{"id": 492, "image": ["rtx_frames_success_2/49_bridge#episode_10187/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10191/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to maintain a stationary position."}, {"from": "gpt", "value": "<point>[[347, 212], [420, 273], [498, 351], [544, 424]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the saucepan."}, {"from": "gpt", "value": "<point>[[362, 136], [479, 280], [589, 462], [697, 677]]</point>"}]}
{"id": 494, "image": ["rtx_frames_success_2/49_bridge#episode_10197/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10204/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the strawberry towards the pot."}, {"from": "gpt", "value": "<point>[[341, 260], [415, 268], [464, 283], [530, 348]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the butter dish."}, {"from": "gpt", "value": "<point>[[250, 313], [295, 439], [354, 576], [413, 682]]</point>"}]}
{"id": 496, "image": ["rtx_frames_success_2/49_bridge#episode_10218/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10227/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the yellow cloth."}, {"from": "gpt", "value": "<point>[[369, 374], [388, 414], [428, 449], [466, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[483, 308], [434, 298], [394, 301], [352, 318]]</point>"}]}
{"id": 498, "image": ["rtx_frames_success_2/49_bridge#episode_10227/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10229/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move away from the container."}, {"from": "gpt", "value": "<point>[[347, 278], [403, 376], [455, 513], [492, 652]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[303, 73], [407, 157], [489, 237], [570, 348]]</point>"}]}
{"id": 500, "image": ["rtx_frames_success_2/49_bridge#episode_10236/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10238/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the cloth."}, {"from": "gpt", "value": "<point>[[142, 88], [354, 189], [576, 336], [750, 472]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal bowl."}, {"from": "gpt", "value": "<point>[[439, 303], [456, 419], [487, 538], [536, 652]]</point>"}]}
{"id": 502, "image": ["rtx_frames_success_2/49_bridge#episode_10248/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10248/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[57, 283], [155, 303], [237, 328], [326, 384]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place a sushi roll between a lid and a carrot."}, {"from": "gpt", "value": "<point>[[316, 333], [377, 414], [417, 503], [453, 611]]</point>"}]}
{"id": 504, "image": ["rtx_frames_success_2/49_bridge#episode_10253/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10292/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position above the pot."}, {"from": "gpt", "value": "<point>[[761, 437], [682, 518], [600, 596], [521, 684]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the eggplant."}, {"from": "gpt", "value": "<point>[[451, 177], [564, 308], [684, 487], [775, 662]]</point>"}]}
{"id": 506, "image": ["rtx_frames_success_2/49_bridge#episode_10326/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10326/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[432, 210], [419, 265], [409, 313], [409, 359]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spoon on the right of the table."}, {"from": "gpt", "value": "<point>[[383, 212], [509, 308], [629, 407], [758, 535]]</point>"}]}
{"id": 508, "image": ["rtx_frames_success_2/49_bridge#episode_10332/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10333/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot on the stove."}, {"from": "gpt", "value": "<point>[[305, 126], [428, 222], [551, 338], [663, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the stuffed toy."}, {"from": "gpt", "value": "<point>[[367, 13], [491, 134], [667, 326], [786, 480]]</point>"}]}
{"id": 510, "image": ["rtx_frames_success_2/49_bridge#episode_10336/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10343/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cucumber."}, {"from": "gpt", "value": "<point>[[303, 58], [373, 129], [438, 202], [492, 278]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the rag."}, {"from": "gpt", "value": "<point>[[244, 121], [362, 187], [472, 260], [578, 346]]</point>"}]}
{"id": 512, "image": ["rtx_frames_success_2/49_bridge#episode_10348/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10350/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[383, 109], [377, 207], [358, 278], [328, 381]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula on the plate."}, {"from": "gpt", "value": "<point>[[773, 399], [697, 399], [627, 409], [555, 427]]</point>"}]}
{"id": 514, "image": ["rtx_frames_success_2/49_bridge#episode_10355/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10355/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[150, 124], [239, 366], [352, 601], [483, 808]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the cutting board."}, {"from": "gpt", "value": "<point>[[472, 758], [407, 735], [341, 684], [286, 624]]</point>"}]}
{"id": 516, "image": ["rtx_frames_success_2/49_bridge#episode_10359/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10359/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[318, 348], [392, 351], [458, 364], [519, 386]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the metal pot on the towel."}, {"from": "gpt", "value": "<point>[[479, 164], [419, 268], [366, 396], [326, 530]]</point>"}]}
{"id": 518, "image": ["rtx_frames_success_2/49_bridge#episode_10360/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10364/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the pot."}, {"from": "gpt", "value": "<point>[[559, 285], [420, 316], [271, 371], [161, 434]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli inside the cardboard box."}, {"from": "gpt", "value": "<point>[[491, 558], [394, 553], [277, 566], [178, 591]]</point>"}]}
{"id": 520, "image": ["rtx_frames_success_2/49_bridge#episode_10369/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10377/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pot on the right corner of the table."}, {"from": "gpt", "value": "<point>[[229, 126], [347, 240], [489, 444], [597, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the spatula."}, {"from": "gpt", "value": "<point>[[432, 429], [489, 389], [542, 371], [610, 376]]</point>"}]}
{"id": 522, "image": ["rtx_frames_success_2/49_bridge#episode_10378/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10381/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[375, 187], [320, 258], [277, 351], [244, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[273, 162], [436, 232], [589, 318], [722, 424]]</point>"}]}
{"id": 524, "image": ["rtx_frames_success_2/49_bridge#episode_10389/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10391/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[331, 136], [252, 230], [189, 331], [125, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[479, 348], [585, 369], [657, 399], [752, 442]]</point>"}]}
{"id": 526, "image": ["rtx_frames_success_2/49_bridge#episode_10391/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10398/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the left burner."}, {"from": "gpt", "value": "<point>[[706, 298], [606, 313], [515, 316], [419, 308]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue spoon."}, {"from": "gpt", "value": "<point>[[498, 169], [398, 247], [290, 348], [197, 457]]</point>"}]}
{"id": 528, "image": ["rtx_frames_success_2/49_bridge#episode_10404/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10404/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[667, 333], [604, 394], [562, 444], [515, 505]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[555, 465], [583, 399], [616, 348], [665, 308]]</point>"}]}
{"id": 530, "image": ["rtx_frames_success_2/49_bridge#episode_10406/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10406/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the object from the bowl."}, {"from": "gpt", "value": "<point>[[699, 174], [669, 323], [644, 530], [631, 687]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the object to the left of the cloth."}, {"from": "gpt", "value": "<point>[[638, 407], [578, 351], [523, 323], [453, 311]]</point>"}]}
{"id": 532, "image": ["rtx_frames_success_2/49_bridge#episode_10410/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10454/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[612, 136], [659, 199], [701, 265], [723, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[441, 25], [360, 136], [267, 295], [199, 434]]</point>"}]}
{"id": 534, "image": ["rtx_frames_success_2/49_bridge#episode_10472/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10473/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spice bottle."}, {"from": "gpt", "value": "<point>[[540, 283], [483, 308], [430, 351], [383, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[388, 154], [494, 346], [574, 515], [657, 740]]</point>"}]}
{"id": 536, "image": ["rtx_frames_success_2/49_bridge#episode_10474/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10474/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the vessel."}, {"from": "gpt", "value": "<point>[[189, 270], [275, 258], [362, 268], [449, 298]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the vessel down."}, {"from": "gpt", "value": "<point>[[434, 167], [494, 313], [578, 530], [633, 712]]</point>"}]}
{"id": 538, "image": ["rtx_frames_success_2/49_bridge#episode_10480/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10492/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the scissors."}, {"from": "gpt", "value": "<point>[[81, 91], [267, 260], [487, 513], [625, 727]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cardboard fence."}, {"from": "gpt", "value": "<point>[[256, 114], [403, 245], [549, 399], [676, 548]]</point>"}]}
{"id": 540, "image": ["rtx_frames_success_2/49_bridge#episode_10500/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10500/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the mushroom."}, {"from": "gpt", "value": "<point>[[295, 13], [288, 83], [290, 169], [297, 242]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom over the bowl."}, {"from": "gpt", "value": "<point>[[333, 66], [360, 222], [381, 429], [392, 604]]</point>"}]}
{"id": 542, "image": ["rtx_frames_success_2/49_bridge#episode_10507/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10515/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bunny."}, {"from": "gpt", "value": "<point>[[312, 275], [456, 336], [547, 391], [653, 470]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pan."}, {"from": "gpt", "value": "<point>[[324, 101], [331, 265], [354, 449], [390, 611]]</point>"}]}
{"id": 544, "image": ["rtx_frames_success_2/49_bridge#episode_10523/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10528/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[400, 210], [475, 278], [547, 354], [600, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[263, 81], [407, 197], [562, 336], [699, 475]]</point>"}]}
{"id": 546, "image": ["rtx_frames_success_2/49_bridge#episode_10543/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10543/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object in the container."}, {"from": "gpt", "value": "<point>[[415, 172], [354, 222], [309, 285], [269, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place a sushi roll to the right of the towel."}, {"from": "gpt", "value": "<point>[[218, 232], [360, 295], [500, 381], [604, 480]]</point>"}]}
{"id": 548, "image": ["rtx_frames_success_2/49_bridge#episode_10549/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10562/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[509, 258], [555, 369], [593, 503], [623, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[354, 91], [354, 215], [371, 369], [396, 530]]</point>"}]}
{"id": 550, "image": ["rtx_frames_success_2/49_bridge#episode_10577/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10584/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower the gripper towards the can."}, {"from": "gpt", "value": "<point>[[504, 343], [468, 414], [409, 500], [339, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[396, 313], [462, 295], [528, 263], [608, 217]]</point>"}]}
{"id": 552, "image": ["rtx_frames_success_2/49_bridge#episode_10588/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10623/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the towel."}, {"from": "gpt", "value": "<point>[[394, 104], [481, 268], [564, 477], [636, 689]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the stuffed duck towards the pan."}, {"from": "gpt", "value": "<point>[[258, 475], [314, 404], [390, 326], [460, 270]]</point>"}]}
{"id": 554, "image": ["rtx_frames_success_2/49_bridge#episode_10631/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10640/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[9, 270], [205, 333], [411, 424], [617, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the rag."}, {"from": "gpt", "value": "<point>[[528, 394], [453, 389], [362, 409], [280, 447]]</point>"}]}
{"id": 556, "image": ["rtx_frames_success_2/49_bridge#episode_10644/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10650/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow pepper."}, {"from": "gpt", "value": "<point>[[464, 227], [460, 333], [473, 457], [506, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom towards the cans."}, {"from": "gpt", "value": "<point>[[259, 159], [403, 232], [555, 313], [731, 422]]</point>"}]}
{"id": 558, "image": ["rtx_frames_success_2/49_bridge#episode_10655/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10667/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position over the pan containing the broccoli."}, {"from": "gpt", "value": "<point>[[392, 217], [491, 283], [602, 364], [688, 455]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position over the sushi roll in the pot."}, {"from": "gpt", "value": "<point>[[706, 66], [667, 220], [608, 462], [572, 664]]</point>"}]}
{"id": 560, "image": ["rtx_frames_success_2/49_bridge#episode_10677/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10693/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[509, 263], [589, 384], [657, 503], [733, 662]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[362, 247], [489, 301], [625, 374], [759, 470]]</point>"}]}
{"id": 562, "image": ["rtx_frames_success_2/49_bridge#episode_10693/frame_15.png", "rtx_frames_success_2/49_bridge#episode_10702/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the bottom right corner of the table."}, {"from": "gpt", "value": "<point>[[705, 467], [657, 556], [616, 679], [591, 798]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[627, 336], [581, 427], [547, 515], [515, 619]]</point>"}]}
{"id": 564, "image": ["rtx_frames_success_2/49_bridge#episode_10720/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10727/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[513, 215], [496, 285], [470, 356], [434, 422]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[396, 207], [384, 346], [388, 449], [398, 596]]</point>"}]}
{"id": 566, "image": ["rtx_frames_success_2/49_bridge#episode_10740/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10742/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the rag."}, {"from": "gpt", "value": "<point>[[513, 197], [398, 270], [286, 354], [176, 447]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[737, 152], [686, 280], [634, 495], [604, 662]]</point>"}]}
{"id": 568, "image": ["rtx_frames_success_2/49_bridge#episode_10743/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10753/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana in the colander."}, {"from": "gpt", "value": "<point>[[684, 273], [572, 391], [475, 508], [373, 654]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple cloth."}, {"from": "gpt", "value": "<point>[[498, 235], [473, 366], [475, 480], [498, 611]]</point>"}]}
{"id": 570, "image": ["rtx_frames_success_2/49_bridge#episode_10763/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10765/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli in the pot."}, {"from": "gpt", "value": "<point>[[330, 162], [464, 197], [564, 250], [659, 316]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the beet in the pot."}, {"from": "gpt", "value": "<point>[[513, 263], [481, 318], [458, 384], [441, 460]]</point>"}]}
{"id": 572, "image": ["rtx_frames_success_2/49_bridge#episode_10784/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10792/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[502, 169], [420, 298], [339, 455], [278, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move a sushi roll to the right of a pot."}, {"from": "gpt", "value": "<point>[[335, 250], [403, 293], [447, 361], [477, 439]]</point>"}]}
{"id": 574, "image": ["rtx_frames_success_2/49_bridge#episode_10827/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10838/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[623, 247], [540, 295], [475, 343], [403, 404]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[339, 449], [470, 447], [562, 442], [712, 422]]</point>"}]}
{"id": 576, "image": ["rtx_frames_success_2/49_bridge#episode_10843/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10852/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the broccoli."}, {"from": "gpt", "value": "<point>[[464, 116], [466, 359], [479, 561], [496, 753]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[801, 298], [661, 298], [515, 306], [354, 333]]</point>"}]}
{"id": 578, "image": ["rtx_frames_success_2/49_bridge#episode_10853/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10859/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato in the bowl."}, {"from": "gpt", "value": "<point>[[331, 207], [350, 318], [360, 429], [366, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom towards the pot."}, {"from": "gpt", "value": "<point>[[532, 326], [445, 247], [350, 202], [269, 189]]</point>"}]}
{"id": 580, "image": ["rtx_frames_success_2/49_bridge#episode_10863/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10871/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sausage."}, {"from": "gpt", "value": "<point>[[311, 227], [362, 348], [407, 482], [436, 619]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow brush."}, {"from": "gpt", "value": "<point>[[250, 265], [354, 293], [438, 328], [525, 386]]</point>"}]}
{"id": 582, "image": ["rtx_frames_success_2/49_bridge#episode_10873/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10888/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[441, 162], [481, 308], [506, 442], [528, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the eggplant."}, {"from": "gpt", "value": "<point>[[458, 172], [386, 265], [305, 384], [235, 510]]</point>"}]}
{"id": 584, "image": ["rtx_frames_success_2/49_bridge#episode_10891/frame_0.png", "rtx_frames_success_2/49_bridge#episode_10892/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[547, 598], [453, 485], [337, 336], [259, 217]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[182, 298], [358, 328], [496, 361], [633, 419]]</point>"}]}
{"id": 586, "image": ["rtx_frames_success_3/49_bridge#episode_119/frame_0.png", "rtx_frames_success_3/49_bridge#episode_121/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[278, 236], [426, 317], [556, 398], [625, 494], [715, 653]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[695, 413], [565, 371], [443, 336], [382, 340]]</point>"}]}
{"id": 588, "image": ["rtx_frames_success_3/49_bridge#episode_121/frame_15.png", "rtx_frames_success_3/49_bridge#episode_1100/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[365, 290], [400, 425], [434, 541], [472, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the plate with the carrot on it."}, {"from": "gpt", "value": "<point>[[507, 208], [605, 293], [698, 386], [773, 537]]</point>"}]}
{"id": 590, "image": ["rtx_frames_success_3/49_bridge#episode_1103/frame_0.png", "rtx_frames_success_3/49_bridge#episode_1118/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pepper."}, {"from": "gpt", "value": "<point>[[220, 69], [336, 166], [423, 266], [515, 413], [605, 637]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blueberries."}, {"from": "gpt", "value": "<point>[[498, 263], [669, 382], [756, 483], [799, 537]]</point>"}]}
{"id": 592, "image": ["rtx_frames_success_3/49_bridge#episode_1130/frame_0.png", "rtx_frames_success_3/49_bridge#episode_1136/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fruit in the bowl."}, {"from": "gpt", "value": "<point>[[420, 147], [498, 282], [559, 413], [608, 494], [686, 637]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[892, 224], [782, 201], [643, 197], [539, 220], [417, 259], [342, 340]]</point>"}]}
{"id": 594, "image": ["rtx_frames_success_3/49_bridge#episode_1143/frame_0.png", "rtx_frames_success_3/49_bridge#episode_1160/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the pot."}, {"from": "gpt", "value": "<point>[[379, 232], [330, 263], [287, 309], [258, 363]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[660, 309], [498, 351], [397, 398], [284, 475]]</point>"}]}
{"id": 596, "image": ["rtx_frames_success_3/49_bridge#episode_1160/frame_15.png", "rtx_frames_success_3/49_bridge#episode_1163/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the beet."}, {"from": "gpt", "value": "<point>[[290, 440], [426, 409], [524, 413], [608, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[605, 39], [544, 131], [492, 232], [405, 332]]</point>"}]}
{"id": 598, "image": ["rtx_frames_success_3/49_bridge#episode_1179/frame_0.png", "rtx_frames_success_3/49_bridge#episode_1192/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[420, 174], [481, 243], [542, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[133, 394], [258, 440], [376, 506], [527, 645]]</point>"}]}
{"id": 600, "image": ["rtx_frames_success_3/49_bridge#episode_1194/frame_0.png", "rtx_frames_success_3/49_bridge#episode_1201/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fish-shaped object in the pot."}, {"from": "gpt", "value": "<point>[[200, 278], [298, 340], [376, 382], [426, 440], [559, 618]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the lime bottle towards the yellow napkin."}, {"from": "gpt", "value": "<point>[[625, 193], [654, 259], [683, 317], [704, 371]]</point>"}]}
{"id": 602, "image": ["rtx_frames_success_3/49_bridge#episode_1204/frame_0.png", "rtx_frames_success_3/49_bridge#episode_1220/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[313, 112], [345, 216], [388, 332], [431, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the bottom left corner of the counter."}, {"from": "gpt", "value": "<point>[[278, 232], [197, 309], [162, 390]]</point>"}]}
{"id": 604, "image": ["rtx_frames_success_3/49_bridge#episode_10896/frame_0.png", "rtx_frames_success_3/49_bridge#episode_10908/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[295, 208], [347, 313], [414, 440], [484, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato in the bowl."}, {"from": "gpt", "value": "<point>[[266, 347], [391, 297], [475, 305], [562, 328], [605, 359]]</point>"}]}
{"id": 606, "image": ["rtx_frames_success_3/49_bridge#episode_10911/frame_0.png", "rtx_frames_success_3/49_bridge#episode_10929/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[446, 193], [530, 232], [568, 290], [634, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl on the surface."}, {"from": "gpt", "value": "<point>[[321, 166], [405, 212], [466, 309], [521, 440], [539, 548]]</point>"}]}
{"id": 608, "image": ["rtx_frames_success_3/49_bridge#episode_10956/frame_0.png", "rtx_frames_success_3/49_bridge#episode_10962/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[301, 247], [429, 259], [533, 317], [591, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[443, 301], [394, 340], [281, 409], [203, 475]]</point>"}]}
{"id": 610, "image": ["rtx_frames_success_3/49_bridge#episode_10976/frame_0.png", "rtx_frames_success_3/49_bridge#episode_10984/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan containing broccoli."}, {"from": "gpt", "value": "<point>[[165, 479], [281, 413], [400, 394], [486, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[365, 212], [408, 320], [466, 456], [533, 676]]</point>"}]}
{"id": 612, "image": ["rtx_frames_success_3/49_bridge#episode_10988/frame_0.png", "rtx_frames_success_3/49_bridge#episode_10988/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach into the laundry machine."}, {"from": "gpt", "value": "<point>[[333, 205], [446, 197], [544, 197], [637, 205]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the clothing out of the machine."}, {"from": "gpt", "value": "<point>[[689, 185], [628, 232], [573, 297], [489, 398], [420, 517]]</point>"}]}
{"id": 614, "image": ["rtx_frames_success_3/49_bridge#episode_10997/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11002/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the vessel."}, {"from": "gpt", "value": "<point>[[709, 212], [582, 205], [507, 236], [385, 270], [304, 336], [264, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pear in the bowl."}, {"from": "gpt", "value": "<point>[[336, 139], [350, 236], [350, 328], [347, 378]]</point>"}]}
{"id": 616, "image": ["rtx_frames_success_3/49_bridge#episode_11004/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11030/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the microwave door closed."}, {"from": "gpt", "value": "<point>[[319, 189], [449, 143], [524, 135], [608, 124]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi container."}, {"from": "gpt", "value": "<point>[[515, 363], [431, 409], [330, 436], [287, 467]]</point>"}]}
{"id": 618, "image": ["rtx_frames_success_3/49_bridge#episode_11033/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11035/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[440, 19], [449, 127], [449, 270], [458, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red pepper."}, {"from": "gpt", "value": "<point>[[843, 139], [854, 259], [846, 390], [811, 568]]</point>"}]}
{"id": 620, "image": ["rtx_frames_success_3/49_bridge#episode_11039/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11041/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[403, 220], [449, 293], [518, 398], [565, 548], [623, 772]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[353, 274], [437, 398], [481, 479], [527, 560], [573, 633]]</point>"}]}
{"id": 622, "image": ["rtx_frames_success_3/49_bridge#episode_11066/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11066/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the cauliflower head."}, {"from": "gpt", "value": "<point>[[854, 351], [837, 456], [817, 575], [805, 695]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot with the cauliflower to the stove."}, {"from": "gpt", "value": "<point>[[709, 104], [709, 201], [701, 282], [657, 274], [568, 274], [510, 266], [440, 266]]</point>"}]}
{"id": 624, "image": ["rtx_frames_success_3/49_bridge#episode_11078/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11078/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[330, 320], [469, 266], [539, 270], [634, 344]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spatula towards the orange cloth."}, {"from": "gpt", "value": "<point>[[594, 332], [489, 293], [431, 274], [362, 270], [287, 270]]</point>"}]}
{"id": 626, "image": ["rtx_frames_success_3/49_bridge#episode_11080/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11081/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot over the table."}, {"from": "gpt", "value": "<point>[[78, 444], [168, 421], [313, 448], [429, 506], [533, 595], [599, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[261, 602], [394, 587], [501, 598], [672, 645]]</point>"}]}
{"id": 628, "image": ["rtx_frames_success_3/49_bridge#episode_11084/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11113/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pineapple."}, {"from": "gpt", "value": "<point>[[371, 243], [446, 232], [510, 239], [625, 278]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[423, 181], [304, 293], [246, 394], [165, 525]]</point>"}]}
{"id": 630, "image": ["rtx_frames_success_3/49_bridge#episode_11148/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11149/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the basil container."}, {"from": "gpt", "value": "<point>[[206, 174], [391, 232], [492, 313], [614, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[295, 687], [376, 606], [446, 587], [539, 583]]</point>"}]}
{"id": 632, "image": ["rtx_frames_success_3/49_bridge#episode_11180/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11180/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[388, 328], [307, 355], [240, 398], [200, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the cloth."}, {"from": "gpt", "value": "<point>[[347, 386], [527, 409], [672, 444], [782, 483]]</point>"}]}
{"id": 634, "image": ["rtx_frames_success_3/49_bridge#episode_11213/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11224/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green vegetable."}, {"from": "gpt", "value": "<point>[[347, 293], [359, 425], [376, 525], [397, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the stuffed duck."}, {"from": "gpt", "value": "<point>[[261, 62], [261, 201], [261, 320], [275, 432]]</point>"}]}
{"id": 636, "image": ["rtx_frames_success_3/49_bridge#episode_11227/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11227/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[281, 212], [310, 309], [330, 405], [333, 486]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl on the right corner of the table."}, {"from": "gpt", "value": "<point>[[605, 263], [675, 456], [707, 568], [750, 707]]</point>"}]}
{"id": 638, "image": ["rtx_frames_success_3/49_bridge#episode_11233/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11237/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[290, 197], [330, 239], [388, 301], [417, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[151, 104], [217, 170], [290, 255], [359, 324]]</point>"}]}
{"id": 640, "image": ["rtx_frames_success_3/49_bridge#episode_11259/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11266/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the sushi roll towards the orange cloth."}, {"from": "gpt", "value": "<point>[[359, 255], [339, 340], [330, 436], [319, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the blue block."}, {"from": "gpt", "value": "<point>[[266, 239], [371, 417], [469, 568], [611, 749]]</point>"}]}
{"id": 642, "image": ["rtx_frames_success_3/49_bridge#episode_11272/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11275/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[345, 290], [449, 375], [550, 483], [654, 660]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the carrot towards the plate."}, {"from": "gpt", "value": "<point>[[426, 340], [518, 440], [539, 506], [550, 564]]</point>"}]}
{"id": 644, "image": ["rtx_frames_success_3/49_bridge#episode_11281/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11286/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the hot sauce bottle."}, {"from": "gpt", "value": "<point>[[458, 405], [394, 463], [356, 525], [319, 625]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[382, 89], [489, 154], [620, 224], [785, 305]]</point>"}]}
{"id": 646, "image": ["rtx_frames_success_3/49_bridge#episode_11286/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11295/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to open the microwave door."}, {"from": "gpt", "value": "<point>[[736, 309], [698, 390], [660, 448], [597, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the fork."}, {"from": "gpt", "value": "<point>[[264, 127], [295, 205], [350, 324], [382, 421], [408, 514]]</point>"}]}
{"id": 648, "image": ["rtx_frames_success_3/49_bridge#episode_11295/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11323/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the fork towards the cloth."}, {"from": "gpt", "value": "<point>[[556, 351], [628, 452], [660, 544], [681, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[527, 263], [405, 297], [266, 363], [211, 517]]</point>"}]}
{"id": 650, "image": ["rtx_frames_success_3/49_bridge#episode_11323/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11330/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the spoon on the table."}, {"from": "gpt", "value": "<point>[[232, 390], [385, 429], [605, 490], [724, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[429, 201], [530, 220], [623, 239], [727, 263]]</point>"}]}
{"id": 652, "image": ["rtx_frames_success_3/49_bridge#episode_11332/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11334/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[162, 344], [382, 305], [527, 340], [628, 378], [698, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the fridge door."}, {"from": "gpt", "value": "<point>[[750, 494], [631, 390], [484, 324], [295, 270], [159, 274], [101, 297], [64, 320]]</point>"}]}
{"id": 654, "image": ["rtx_frames_success_3/49_bridge#episode_11343/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11375/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[217, 347], [434, 351], [568, 367], [678, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[759, 212], [698, 347], [597, 502], [507, 633]]</point>"}]}
{"id": 656, "image": ["rtx_frames_success_3/49_bridge#episode_11390/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11392/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to turn on the stove."}, {"from": "gpt", "value": "<point>[[304, 247], [327, 317], [347, 375], [374, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[712, 205], [634, 201], [495, 208], [371, 236], [229, 282]]</point>"}]}
{"id": 658, "image": ["rtx_frames_success_3/49_bridge#episode_11408/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11411/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[394, 201], [443, 301], [492, 429], [533, 687]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[295, 185], [411, 239], [556, 301], [669, 355], [793, 425]]</point>"}]}
{"id": 660, "image": ["rtx_frames_success_3/49_bridge#episode_11421/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11428/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the saucepan."}, {"from": "gpt", "value": "<point>[[266, 104], [278, 236], [298, 363], [298, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot in the pan."}, {"from": "gpt", "value": "<point>[[429, 197], [452, 324], [486, 471], [518, 645]]</point>"}]}
{"id": 662, "image": ["rtx_frames_success_3/49_bridge#episode_11448/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11482/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red spoon."}, {"from": "gpt", "value": "<point>[[168, 220], [252, 320], [336, 417], [417, 529], [539, 707]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to open the microwave door."}, {"from": "gpt", "value": "<point>[[290, 100], [266, 201], [255, 266], [249, 371]]</point>"}]}
{"id": 664, "image": ["rtx_frames_success_3/49_bridge#episode_11512/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11512/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[550, 85], [481, 205], [458, 274], [440, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pan on the green towel."}, {"from": "gpt", "value": "<point>[[466, 174], [550, 232], [620, 305], [707, 398]]</point>"}]}
{"id": 666, "image": ["rtx_frames_success_3/49_bridge#episode_11520/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11523/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cloth."}, {"from": "gpt", "value": "<point>[[434, 89], [539, 201], [599, 293], [657, 421], [692, 552], [707, 625]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[475, 189], [437, 320], [423, 390], [350, 510]]</point>"}]}
{"id": 668, "image": ["rtx_frames_success_3/49_bridge#episode_11535/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11535/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[570, 89], [637, 131], [724, 205], [785, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom over the napkin."}, {"from": "gpt", "value": "<point>[[588, 46], [513, 58], [417, 97], [347, 174]]</point>"}]}
{"id": 670, "image": ["rtx_frames_success_3/49_bridge#episode_11547/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11556/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[356, 274], [391, 371], [417, 479], [426, 544]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[533, 247], [460, 263], [385, 301]]</point>"}]}
{"id": 672, "image": ["rtx_frames_success_3/49_bridge#episode_11564/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11580/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the fridge handle."}, {"from": "gpt", "value": "<point>[[327, 351], [423, 425], [515, 506], [611, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[458, 205], [486, 320], [504, 471], [521, 656]]</point>"}]}
{"id": 674, "image": ["rtx_frames_success_3/49_bridge#episode_11582/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11605/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon above the table."}, {"from": "gpt", "value": "<point>[[197, 301], [313, 332], [455, 432], [518, 510], [553, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grip the vessel."}, {"from": "gpt", "value": "<point>[[336, 162], [403, 344], [458, 479], [515, 575]]</point>"}]}
{"id": 676, "image": ["rtx_frames_success_3/49_bridge#episode_11637/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11645/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[214, 336], [336, 320], [417, 313], [544, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[403, 205], [460, 282], [550, 398], [614, 494]]</point>"}]}
{"id": 678, "image": ["rtx_frames_success_3/49_bridge#episode_11646/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11657/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the yellow capsicum."}, {"from": "gpt", "value": "<point>[[472, 216], [539, 347], [591, 479], [663, 660]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[333, 676], [347, 359], [292, 89]]</point>"}]}
{"id": 680, "image": ["rtx_frames_success_3/49_bridge#episode_11665/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11667/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[237, 278], [368, 336], [452, 382], [542, 444], [625, 517], [709, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the grapes."}, {"from": "gpt", "value": "<point>[[350, 205], [544, 286], [637, 340], [738, 452], [796, 564], [840, 668]]</point>"}]}
{"id": 682, "image": ["rtx_frames_success_3/49_bridge#episode_11682/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11712/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[281, 62], [330, 139], [397, 263], [452, 355], [536, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[107, 301], [240, 251], [313, 205]]</point>"}]}
{"id": 684, "image": ["rtx_frames_success_3/49_bridge#episode_11714/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11734/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[171, 81], [217, 193], [278, 336], [368, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[75, 232], [197, 251], [333, 309], [437, 355], [513, 402], [669, 490]]</point>"}]}
{"id": 686, "image": ["rtx_frames_success_3/49_bridge#episode_11739/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11742/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the broccoli."}, {"from": "gpt", "value": "<point>[[599, 58], [472, 120], [411, 208], [368, 297]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[707, 483], [617, 444], [518, 471], [417, 517], [347, 587]]</point>"}]}
{"id": 688, "image": ["rtx_frames_success_3/49_bridge#episode_11766/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11773/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[429, 170], [582, 220], [654, 255], [747, 309]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[440, 120], [510, 201], [568, 282], [623, 359], [672, 440], [724, 541]]</point>"}]}
{"id": 690, "image": ["rtx_frames_success_3/49_bridge#episode_11776/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11788/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[414, 124], [353, 185], [290, 259], [232, 324], [177, 398]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[573, 320], [492, 270], [374, 243], [281, 266], [229, 309]]</point>"}]}
{"id": 692, "image": ["rtx_frames_success_3/49_bridge#episode_11805/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11806/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to raise away from the plate and sink."}, {"from": "gpt", "value": "<point>[[481, 440], [527, 429], [594, 479], [634, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the eggplant."}, {"from": "gpt", "value": "<point>[[327, 108], [400, 154], [431, 197], [478, 293]]</point>"}]}
{"id": 694, "image": ["rtx_frames_success_3/49_bridge#episode_11829/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11829/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow sponge in the bowl."}, {"from": "gpt", "value": "<point>[[463, 124], [365, 232], [275, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the sponge away from the bowl."}, {"from": "gpt", "value": "<point>[[319, 93], [443, 154], [565, 232], [683, 320], [791, 413]]</point>"}]}
{"id": 696, "image": ["rtx_frames_success_3/49_bridge#episode_11852/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11852/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the towel."}, {"from": "gpt", "value": "<point>[[683, 394], [582, 328], [489, 286], [376, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the towel."}, {"from": "gpt", "value": "<point>[[391, 251], [434, 378], [452, 494], [469, 571]]</point>"}]}
{"id": 698, "image": ["rtx_frames_success_3/49_bridge#episode_11855/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11859/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[672, 440], [539, 371], [397, 386], [217, 409]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[484, 382], [510, 459], [527, 533], [542, 649]]</point>"}]}
{"id": 700, "image": ["rtx_frames_success_3/49_bridge#episode_11868/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11869/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull open the oven door."}, {"from": "gpt", "value": "<point>[[492, 637], [559, 722], [588, 768], [663, 888]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the strawberry on the right corner of the table."}, {"from": "gpt", "value": "<point>[[339, 236], [469, 378], [550, 548], [623, 745]]</point>"}]}
{"id": 702, "image": ["rtx_frames_success_3/49_bridge#episode_11897/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11897/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[446, 305], [356, 266], [275, 293]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spatula towards the front edge of the table."}, {"from": "gpt", "value": "<point>[[321, 239], [405, 382], [458, 506], [524, 672]]</point>"}]}
{"id": 704, "image": ["rtx_frames_success_3/49_bridge#episode_11916/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11922/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the beet in the pot."}, {"from": "gpt", "value": "<point>[[414, 255], [472, 359], [501, 475], [518, 583]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cheese wedge."}, {"from": "gpt", "value": "<point>[[336, 170], [475, 201], [565, 270], [689, 413]]</point>"}]}
{"id": 706, "image": ["rtx_frames_success_3/49_bridge#episode_11927/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11936/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[182, 340], [264, 317], [333, 309], [434, 297]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the meat."}, {"from": "gpt", "value": "<point>[[405, 178], [356, 290], [266, 537]]</point>"}]}
{"id": 708, "image": ["rtx_frames_success_3/49_bridge#episode_11940/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11958/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[510, 185], [501, 309], [495, 394], [498, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[313, 85], [458, 220], [513, 340], [565, 444]]</point>"}]}
{"id": 710, "image": ["rtx_frames_success_3/49_bridge#episode_11963/frame_15.png", "rtx_frames_success_3/49_bridge#episode_11964/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bowl towards the drumstick."}, {"from": "gpt", "value": "<point>[[431, 139], [492, 274], [556, 479], [588, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the vegetable."}, {"from": "gpt", "value": "<point>[[249, 120], [371, 147], [501, 201], [585, 255], [657, 363]]</point>"}]}
{"id": 712, "image": ["rtx_frames_success_3/49_bridge#episode_11969/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11969/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[174, 216], [278, 282], [394, 359], [599, 517], [698, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[712, 587], [634, 529], [550, 490], [455, 459], [391, 436]]</point>"}]}
{"id": 714, "image": ["rtx_frames_success_3/49_bridge#episode_11982/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11996/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[394, 73], [449, 151], [492, 224], [544, 301], [573, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position above the chicken wing in the pot."}, {"from": "gpt", "value": "<point>[[374, 255], [388, 340], [394, 429], [411, 556]]</point>"}]}
{"id": 716, "image": ["rtx_frames_success_3/49_bridge#episode_11997/frame_0.png", "rtx_frames_success_3/49_bridge#episode_11998/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object in the pot."}, {"from": "gpt", "value": "<point>[[261, 236], [356, 309], [469, 417], [631, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the refrigerator door handle."}, {"from": "gpt", "value": "<point>[[307, 212], [417, 255], [521, 297], [672, 340], [724, 355]]</point>"}]}
{"id": 718, "image": ["rtx_frames_success_3/49_bridge#episode_12054/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12055/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[423, 371], [313, 378], [240, 413], [206, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[553, 367], [437, 367], [362, 413], [287, 471], [237, 544]]</point>"}]}
{"id": 720, "image": ["rtx_frames_success_3/49_bridge#episode_12060/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12060/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana in the pot."}, {"from": "gpt", "value": "<point>[[145, 185], [310, 224], [437, 259], [536, 297], [652, 363]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the banana to the left of a pot."}, {"from": "gpt", "value": "<point>[[654, 239], [585, 243], [475, 266], [379, 305]]</point>"}]}
{"id": 722, "image": ["rtx_frames_success_3/49_bridge#episode_12091/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12093/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green pepper."}, {"from": "gpt", "value": "<point>[[692, 232], [617, 178], [562, 162], [452, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the stuffed duck."}, {"from": "gpt", "value": "<point>[[646, 54], [501, 93], [420, 193], [342, 332], [330, 394]]</point>"}]}
{"id": 724, "image": ["rtx_frames_success_3/49_bridge#episode_12097/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12097/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[206, 320], [290, 270], [353, 263], [429, 266]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the bottom right corner."}, {"from": "gpt", "value": "<point>[[440, 151], [547, 305], [628, 463], [666, 583]]</point>"}]}
{"id": 726, "image": ["rtx_frames_success_3/49_bridge#episode_12099/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12103/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the small pan."}, {"from": "gpt", "value": "<point>[[255, 317], [356, 293], [458, 301], [597, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[762, 251], [692, 317], [652, 359], [608, 398]]</point>"}]}
{"id": 728, "image": ["rtx_frames_success_3/49_bridge#episode_12126/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12144/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[116, 73], [165, 224], [226, 371], [269, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[556, 413], [481, 278], [394, 236], [278, 247]]</point>"}]}
{"id": 730, "image": ["rtx_frames_success_3/49_bridge#episode_12149/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12149/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[313, 382], [440, 359], [582, 386], [721, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the left of the bowl."}, {"from": "gpt", "value": "<point>[[620, 185], [423, 193], [376, 216], [301, 251]]</point>"}]}
{"id": 732, "image": ["rtx_frames_success_3/49_bridge#episode_12158/frame_15.png", "rtx_frames_success_3/49_bridge#episode_12183/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pan on the stove."}, {"from": "gpt", "value": "<point>[[408, 112], [371, 197], [336, 290], [321, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the scissors."}, {"from": "gpt", "value": "<point>[[90, 170], [237, 255], [391, 328], [544, 494]]</point>"}]}
{"id": 734, "image": ["rtx_frames_success_3/49_bridge#episode_12200/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12200/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the sushi from a bowl."}, {"from": "gpt", "value": "<point>[[582, 208], [515, 286], [431, 417], [379, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the sushi on the table."}, {"from": "gpt", "value": "<point>[[556, 185], [646, 328], [675, 375], [698, 421]]</point>"}]}
{"id": 736, "image": ["rtx_frames_success_3/49_bridge#episode_12225/frame_15.png", "rtx_frames_success_3/49_bridge#episode_12235/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mushroom over the table."}, {"from": "gpt", "value": "<point>[[55, 174], [122, 351], [197, 483], [304, 595]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pot on the stove."}, {"from": "gpt", "value": "<point>[[640, 216], [562, 324], [544, 402], [524, 525]]</point>"}]}
{"id": 738, "image": ["rtx_frames_success_3/49_bridge#episode_12238/frame_15.png", "rtx_frames_success_3/49_bridge#episode_12240/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the vessel."}, {"from": "gpt", "value": "<point>[[220, 185], [339, 193], [429, 282], [495, 355], [542, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cucumber."}, {"from": "gpt", "value": "<point>[[666, 212], [539, 290], [458, 363], [359, 490], [327, 571]]</point>"}]}
{"id": 740, "image": ["rtx_frames_success_3/49_bridge#episode_12241/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12254/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula in the sink."}, {"from": "gpt", "value": "<point>[[750, 429], [724, 514], [689, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the pan."}, {"from": "gpt", "value": "<point>[[463, 475], [559, 459], [669, 506], [721, 525]]</point>"}]}
{"id": 742, "image": ["rtx_frames_success_3/49_bridge#episode_12255/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12257/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[298, 116], [333, 247], [350, 359], [379, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[594, 286], [547, 293], [486, 317], [446, 344]]</point>"}]}
{"id": 744, "image": ["rtx_frames_success_3/49_bridge#episode_12262/frame_0.png", "rtx_frames_success_3/49_bridge#episode_12272/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[368, 77], [455, 263], [527, 402], [599, 583], [623, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the vessel."}, {"from": "gpt", "value": "<point>[[379, 243], [440, 347], [504, 475], [539, 541]]</point>"}]}
{"id": 746, "image": ["rtx_frames_success_4/49_bridge#episode_1244/frame_0.png", "rtx_frames_success_4/49_bridge#episode_1265/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[468, 244], [433, 300], [419, 347], [440, 423]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[637, 333], [511, 291], [387, 305], [338, 357]]</point>"}]}
{"id": 748, "image": ["rtx_frames_success_4/49_bridge#episode_1265/frame_15.png", "rtx_frames_success_4/49_bridge#episode_1268/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the napkin over the pot."}, {"from": "gpt", "value": "<point>[[423, 394], [518, 371], [616, 385], [673, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the beet in the pot."}, {"from": "gpt", "value": "<point>[[824, 460], [736, 432], [658, 427], [581, 488]]</point>"}]}
{"id": 750, "image": ["rtx_frames_success_4/49_bridge#episode_1295/frame_0.png", "rtx_frames_success_4/49_bridge#episode_1325/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sauce pot."}, {"from": "gpt", "value": "<point>[[285, 216], [299, 305], [292, 385], [250, 460]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[356, 385], [426, 437], [504, 474], [616, 446]]</point>"}]}
{"id": 752, "image": ["rtx_frames_success_4/49_bridge#episode_1349/frame_0.png", "rtx_frames_success_4/49_bridge#episode_1349/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[363, 300], [482, 371], [556, 474], [599, 700]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the left edge of the counter."}, {"from": "gpt", "value": "<point>[[620, 676], [521, 554], [384, 488], [222, 460]]</point>"}]}
{"id": 754, "image": ["rtx_frames_success_4/49_bridge#episode_1355/frame_0.png", "rtx_frames_success_4/49_bridge#episode_1355/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[440, 174], [454, 249], [461, 319], [433, 427]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the brush towards the napkin."}, {"from": "gpt", "value": "<point>[[352, 366], [465, 404], [574, 469], [687, 718]]</point>"}]}
{"id": 756, "image": ["rtx_frames_success_4/49_bridge#episode_1364/frame_0.png", "rtx_frames_success_4/49_bridge#episode_1365/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the bowl."}, {"from": "gpt", "value": "<point>[[718, 239], [546, 239], [444, 315], [415, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[542, 146], [408, 155], [356, 202], [303, 268]]</point>"}]}
{"id": 758, "image": ["rtx_frames_success_4/49_bridge#episode_12286/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12286/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the red object."}, {"from": "gpt", "value": "<point>[[468, 413], [356, 333], [236, 315], [162, 385]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the red pepper to the right of the towel."}, {"from": "gpt", "value": "<point>[[303, 249], [479, 244], [585, 286], [732, 451]]</point>"}]}
{"id": 760, "image": ["rtx_frames_success_4/49_bridge#episode_12294/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12317/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the bowl."}, {"from": "gpt", "value": "<point>[[511, 272], [482, 329], [482, 413], [518, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[401, 268], [264, 291], [194, 362], [173, 446]]</point>"}]}
{"id": 762, "image": ["rtx_frames_success_4/49_bridge#episode_12317/frame_15.png", "rtx_frames_success_4/49_bridge#episode_12348/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the potato towards the pot."}, {"from": "gpt", "value": "<point>[[194, 545], [264, 488], [327, 484], [444, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice container."}, {"from": "gpt", "value": "<point>[[602, 399], [486, 385], [405, 394], [349, 432]]</point>"}]}
{"id": 764, "image": ["rtx_frames_success_4/49_bridge#episode_12353/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12371/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[437, 347], [493, 338], [546, 366], [616, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pepper in the pot."}, {"from": "gpt", "value": "<point>[[412, 207], [482, 178], [553, 211], [630, 333]]</point>"}]}
{"id": 766, "image": ["rtx_frames_success_4/49_bridge#episode_12372/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12377/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[433, 329], [535, 418], [606, 535], [602, 704]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue object."}, {"from": "gpt", "value": "<point>[[539, 197], [405, 197], [292, 258], [208, 423]]</point>"}]}
{"id": 768, "image": ["rtx_frames_success_4/49_bridge#episode_12379/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12414/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[447, 272], [475, 362], [518, 385], [581, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[320, 141], [313, 315], [356, 432], [525, 596]]</point>"}]}
{"id": 770, "image": ["rtx_frames_success_4/49_bridge#episode_12416/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12431/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[894, 554], [820, 559], [775, 601], [771, 685]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[229, 277], [204, 324], [208, 376], [232, 432]]</point>"}]}
{"id": 772, "image": ["rtx_frames_success_4/49_bridge#episode_12444/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12447/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the orange object."}, {"from": "gpt", "value": "<point>[[458, 61], [391, 164], [373, 296], [391, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[468, 211], [433, 329], [440, 451], [496, 549]]</point>"}]}
{"id": 774, "image": ["rtx_frames_success_4/49_bridge#episode_12457/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12466/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red pepper."}, {"from": "gpt", "value": "<point>[[595, 423], [468, 357], [324, 329], [201, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[391, 122], [472, 263], [482, 423], [468, 714]]</point>"}]}
{"id": 776, "image": ["rtx_frames_success_4/49_bridge#episode_12488/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12488/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[363, 352], [468, 432], [532, 488], [609, 737]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the bottom left corner."}, {"from": "gpt", "value": "<point>[[282, 310], [222, 357], [197, 418], [201, 474]]</point>"}]}
{"id": 778, "image": ["rtx_frames_success_4/49_bridge#episode_12492/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12497/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[271, 150], [387, 254], [454, 357], [496, 507]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue object."}, {"from": "gpt", "value": "<point>[[521, 272], [588, 347], [644, 474], [673, 728]]</point>"}]}
{"id": 780, "image": ["rtx_frames_success_4/49_bridge#episode_12497/frame_15.png", "rtx_frames_success_4/49_bridge#episode_12499/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the blue object towards the purple cloth."}, {"from": "gpt", "value": "<point>[[394, 282], [317, 282], [261, 315], [222, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[204, 216], [296, 282], [356, 366], [384, 526]]</point>"}]}
{"id": 782, "image": ["rtx_frames_success_4/49_bridge#episode_12502/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12513/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the measuring cup."}, {"from": "gpt", "value": "<point>[[482, 42], [447, 103], [419, 174], [398, 272]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the towel."}, {"from": "gpt", "value": "<point>[[180, 122], [250, 211], [299, 310], [306, 460]]</point>"}]}
{"id": 784, "image": ["rtx_frames_success_4/49_bridge#episode_12533/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12551/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[401, 352], [468, 324], [511, 315], [542, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[465, 85], [546, 254], [570, 390], [567, 545]]</point>"}]}
{"id": 786, "image": ["rtx_frames_success_4/49_bridge#episode_12558/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12558/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[680, 221], [708, 272], [722, 366], [704, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom towards the microwave."}, {"from": "gpt", "value": "<point>[[641, 296], [570, 254], [486, 225], [405, 239]]</point>"}]}
{"id": 788, "image": ["rtx_frames_success_4/49_bridge#episode_12559/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12569/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up clothes from a basket."}, {"from": "gpt", "value": "<point>[[535, 408], [616, 469], [651, 545], [704, 695]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spatula."}, {"from": "gpt", "value": "<point>[[366, 174], [486, 272], [574, 413], [676, 685]]</point>"}]}
{"id": 790, "image": ["rtx_frames_success_4/49_bridge#episode_12569/frame_15.png", "rtx_frames_success_4/49_bridge#episode_12575/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the cloth."}, {"from": "gpt", "value": "<point>[[641, 667], [585, 535], [507, 423], [408, 324], [616, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[461, 263], [542, 333], [577, 404], [606, 545]]</point>"}]}
{"id": 792, "image": ["rtx_frames_success_4/49_bridge#episode_12606/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12614/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet handle."}, {"from": "gpt", "value": "<point>[[384, 526], [475, 535], [563, 573], [662, 676]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green towel."}, {"from": "gpt", "value": "<point>[[176, 385], [275, 352], [349, 343], [444, 366]]</point>"}]}
{"id": 794, "image": ["rtx_frames_success_4/49_bridge#episode_12623/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12623/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[588, 404], [511, 324], [447, 272], [387, 272]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the orange towards the purple rag."}, {"from": "gpt", "value": "<point>[[468, 268], [511, 376], [528, 479], [528, 620]]</point>"}]}
{"id": 796, "image": ["rtx_frames_success_4/49_bridge#episode_12628/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12633/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the spatula."}, {"from": "gpt", "value": "<point>[[574, 160], [528, 319], [546, 451], [602, 723]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[261, 300], [380, 272], [486, 286], [609, 333]]</point>"}]}
{"id": 798, "image": ["rtx_frames_success_4/49_bridge#episode_12649/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12656/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[243, 315], [345, 300], [472, 296], [687, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[363, 268], [437, 347], [489, 451], [496, 592]]</point>"}]}
{"id": 800, "image": ["rtx_frames_success_4/49_bridge#episode_12667/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12669/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[669, 169], [475, 211], [342, 329], [222, 516]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula on the plate."}, {"from": "gpt", "value": "<point>[[391, 362], [461, 380], [504, 404], [539, 451]]</point>"}]}
{"id": 802, "image": ["rtx_frames_success_4/49_bridge#episode_12671/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12678/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the small metal pan."}, {"from": "gpt", "value": "<point>[[250, 235], [338, 230], [412, 239], [546, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[412, 347], [451, 394], [468, 469], [489, 592]]</point>"}]}
{"id": 804, "image": ["rtx_frames_success_4/49_bridge#episode_12686/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12687/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the cheese."}, {"from": "gpt", "value": "<point>[[609, 484], [493, 488], [426, 526], [401, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green spatula."}, {"from": "gpt", "value": "<point>[[218, 197], [458, 300], [581, 455], [665, 695]]</point>"}]}
{"id": 806, "image": ["rtx_frames_success_4/49_bridge#episode_12707/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12713/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry in the bowl."}, {"from": "gpt", "value": "<point>[[613, 385], [662, 432], [665, 484], [665, 554]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[461, 296], [525, 362], [539, 441], [507, 568]]</point>"}]}
{"id": 808, "image": ["rtx_frames_success_4/49_bridge#episode_12721/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12736/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the eggplant."}, {"from": "gpt", "value": "<point>[[299, 164], [275, 235], [254, 343], [285, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[363, 244], [489, 263], [585, 282], [764, 465]]</point>"}]}
{"id": 810, "image": ["rtx_frames_success_4/49_bridge#episode_12756/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12766/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[423, 254], [384, 286], [345, 352], [338, 437]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[521, 131], [574, 211], [609, 319], [602, 535]]</point>"}]}
{"id": 812, "image": ["rtx_frames_success_4/49_bridge#episode_12778/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12780/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula on the plate."}, {"from": "gpt", "value": "<point>[[739, 296], [690, 376], [676, 465], [669, 606]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[246, 160], [306, 225], [352, 286], [412, 465]]</point>"}]}
{"id": 814, "image": ["rtx_frames_success_4/49_bridge#episode_12782/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12821/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[359, 70], [398, 188], [412, 324], [419, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[521, 131], [482, 202], [475, 300], [504, 399]]</point>"}]}
{"id": 816, "image": ["rtx_frames_success_4/49_bridge#episode_12831/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12835/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[669, 376], [609, 329], [525, 272], [384, 268]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[349, 235], [468, 347], [570, 516], [644, 704], [359, 282]]</point>"}]}
{"id": 818, "image": ["rtx_frames_success_4/49_bridge#episode_12835/frame_15.png", "rtx_frames_success_4/49_bridge#episode_12864/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon over the pot."}, {"from": "gpt", "value": "<point>[[637, 676], [514, 526], [387, 451], [183, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[419, 221], [486, 282], [532, 310], [560, 427]]</point>"}]}
{"id": 820, "image": ["rtx_frames_success_4/49_bridge#episode_12867/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12880/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[229, 324], [324, 277], [415, 291], [574, 380]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the cloth."}, {"from": "gpt", "value": "<point>[[398, 300], [370, 366], [377, 460], [423, 573]]</point>"}]}
{"id": 822, "image": ["rtx_frames_success_4/49_bridge#episode_12893/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12906/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue spoon."}, {"from": "gpt", "value": "<point>[[430, 169], [539, 282], [585, 399], [655, 662]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[500, 254], [623, 268], [690, 310], [810, 446]]</point>"}]}
{"id": 824, "image": ["rtx_frames_success_4/49_bridge#episode_12913/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12944/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange vessel."}, {"from": "gpt", "value": "<point>[[482, 127], [588, 272], [620, 446], [602, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[496, 131], [542, 282], [532, 404], [504, 596]]</point>"}]}
{"id": 826, "image": ["rtx_frames_success_4/49_bridge#episode_12950/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12960/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[451, 235], [518, 239], [609, 258], [711, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[380, 296], [278, 277], [204, 319], [169, 380]]</point>"}]}
{"id": 828, "image": ["rtx_frames_success_4/49_bridge#episode_12991/frame_0.png", "rtx_frames_success_4/49_bridge#episode_12993/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[271, 507], [430, 432], [567, 460], [683, 559]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[669, 286], [722, 329], [743, 366], [743, 432]]</point>"}]}
{"id": 830, "image": ["rtx_frames_success_4/49_bridge#episode_12993/frame_15.png", "rtx_frames_success_4/49_bridge#episode_13001/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the object across the surface."}, {"from": "gpt", "value": "<point>[[662, 423], [683, 516], [676, 624], [662, 737]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[398, 277], [433, 347], [437, 423], [419, 563]]</point>"}]}
{"id": 832, "image": ["rtx_frames_success_4/49_bridge#episode_13003/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13009/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[489, 244], [486, 300], [496, 352], [546, 362]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[532, 99], [595, 188], [606, 263], [641, 446]]</point>"}]}
{"id": 834, "image": ["rtx_frames_success_4/49_bridge#episode_13009/frame_15.png", "rtx_frames_success_4/49_bridge#episode_13019/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mushroom towards the pot."}, {"from": "gpt", "value": "<point>[[697, 559], [648, 427], [563, 362], [423, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[384, 319], [430, 366], [433, 427], [426, 507]]</point>"}]}
{"id": 836, "image": ["rtx_frames_success_4/49_bridge#episode_13023/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13031/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[549, 136], [482, 155], [433, 202], [405, 291]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[306, 244], [387, 300], [415, 413], [419, 606]]</point>"}]}
{"id": 838, "image": ["rtx_frames_success_4/49_bridge#episode_13057/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13064/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the stuffed animal."}, {"from": "gpt", "value": "<point>[[835, 554], [704, 423], [546, 380], [320, 418]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cardboard fence."}, {"from": "gpt", "value": "<point>[[405, 244], [306, 197], [215, 207], [109, 305]]</point>"}]}
{"id": 840, "image": ["rtx_frames_success_4/49_bridge#episode_13064/frame_15.png", "rtx_frames_success_4/49_bridge#episode_13072/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cardboard fence."}, {"from": "gpt", "value": "<point>[[127, 366], [246, 376], [324, 418], [401, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[352, 131], [271, 207], [229, 272], [194, 413]]</point>"}]}
{"id": 842, "image": ["rtx_frames_success_4/49_bridge#episode_13093/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13100/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[454, 291], [577, 366], [658, 469], [680, 667]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot on the stove top."}, {"from": "gpt", "value": "<point>[[342, 296], [444, 338], [521, 404], [570, 516]]</point>"}]}
{"id": 844, "image": ["rtx_frames_success_4/49_bridge#episode_13114/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13115/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven door."}, {"from": "gpt", "value": "<point>[[187, 737], [349, 737], [479, 770], [546, 892]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[201, 305], [335, 310], [444, 343], [577, 451]]</point>"}]}
{"id": 846, "image": ["rtx_frames_success_4/49_bridge#episode_13115/frame_15.png", "rtx_frames_success_4/49_bridge#episode_13149/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife towards the pot."}, {"from": "gpt", "value": "<point>[[535, 418], [430, 423], [303, 437], [250, 507]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[704, 211], [546, 188], [391, 221], [264, 305]]</point>"}]}
{"id": 848, "image": ["rtx_frames_success_4/49_bridge#episode_13165/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13167/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[310, 244], [472, 282], [588, 371], [644, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[567, 146], [412, 188], [363, 268], [303, 413]]</point>"}]}
{"id": 850, "image": ["rtx_frames_success_4/49_bridge#episode_13175/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13202/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over a mushroom toy."}, {"from": "gpt", "value": "<point>[[412, 531], [313, 432], [222, 404], [137, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[616, 268], [651, 357], [655, 404], [641, 502]]</point>"}]}
{"id": 852, "image": ["rtx_frames_success_4/49_bridge#episode_13209/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13211/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[370, 117], [444, 169], [472, 244], [472, 329]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[271, 211], [204, 286], [187, 357], [201, 408]]</point>"}]}
{"id": 854, "image": ["rtx_frames_success_4/49_bridge#episode_13211/frame_15.png", "rtx_frames_success_4/49_bridge#episode_13212/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the baster."}, {"from": "gpt", "value": "<point>[[278, 493], [320, 366], [394, 300], [475, 329]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[201, 272], [394, 324], [567, 418], [715, 667]]</point>"}]}
{"id": 856, "image": ["rtx_frames_success_4/49_bridge#episode_13223/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13246/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[592, 296], [556, 413], [563, 498], [609, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[218, 146], [465, 235], [630, 362], [722, 502]]</point>"}]}
{"id": 858, "image": ["rtx_frames_success_4/49_bridge#episode_13246/frame_15.png", "rtx_frames_success_4/49_bridge#episode_13247/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the left of the cloth."}, {"from": "gpt", "value": "<point>[[644, 286], [472, 291], [356, 333], [236, 423]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[387, 117], [489, 230], [535, 390], [599, 662]]</point>"}]}
{"id": 860, "image": ["rtx_frames_success_4/49_bridge#episode_13249/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13278/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[331, 202], [398, 291], [412, 376], [391, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[518, 131], [419, 244], [394, 404], [384, 624]]</point>"}]}
{"id": 862, "image": ["rtx_frames_success_4/49_bridge#episode_13283/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13295/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[180, 286], [335, 333], [433, 423], [546, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[299, 338], [363, 441], [363, 484], [349, 563]]</point>"}]}
{"id": 864, "image": ["rtx_frames_success_4/49_bridge#episode_13300/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13309/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[70, 310], [306, 385], [475, 512], [581, 746]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[496, 235], [454, 315], [451, 385], [461, 469]]</point>"}]}
{"id": 866, "image": ["rtx_frames_success_4/49_bridge#episode_13326/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13326/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[563, 282], [433, 291], [356, 366], [306, 441]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot towards the spoon."}, {"from": "gpt", "value": "<point>[[324, 404], [472, 441], [549, 479], [599, 620]]</point>"}]}
{"id": 868, "image": ["rtx_frames_success_4/49_bridge#episode_13344/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13366/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[130, 178], [201, 197], [257, 235], [282, 329]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[525, 211], [542, 305], [560, 408], [542, 512]]</point>"}]}
{"id": 870, "image": ["rtx_frames_success_4/49_bridge#episode_13368/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13384/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[475, 131], [563, 272], [613, 404], [644, 615]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[137, 249], [292, 235], [373, 254], [447, 338]]</point>"}]}
{"id": 872, "image": ["rtx_frames_success_4/49_bridge#episode_13424/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13428/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[169, 357], [282, 310], [391, 333], [556, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[236, 244], [426, 423], [511, 601], [588, 864]]</point>"}]}
{"id": 874, "image": ["rtx_frames_success_4/49_bridge#episode_13431/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13442/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cucumber in the cup."}, {"from": "gpt", "value": "<point>[[806, 418], [644, 460], [549, 516], [426, 653]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[285, 277], [405, 254], [468, 272], [567, 338]]</point>"}]}
{"id": 876, "image": ["rtx_frames_success_4/49_bridge#episode_13452/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13469/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container with sushi."}, {"from": "gpt", "value": "<point>[[264, 216], [401, 347], [440, 479], [496, 714]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[352, 258], [454, 329], [546, 451], [623, 662]]</point>"}]}
{"id": 878, "image": ["rtx_frames_success_4/49_bridge#episode_13476/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13489/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[282, 239], [387, 244], [489, 319], [609, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the beet in the sink."}, {"from": "gpt", "value": "<point>[[658, 343], [563, 362], [507, 404], [454, 502]]</point>"}]}
{"id": 880, "image": ["rtx_frames_success_4/49_bridge#episode_13497/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13505/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[563, 282], [609, 366], [623, 493], [651, 700]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[468, 239], [560, 357], [567, 399], [567, 465]]</point>"}]}
{"id": 882, "image": ["rtx_frames_success_4/49_bridge#episode_13516/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13521/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green vegetable."}, {"from": "gpt", "value": "<point>[[535, 258], [468, 235], [412, 268], [384, 291]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot containing the beet."}, {"from": "gpt", "value": "<point>[[599, 319], [482, 329], [405, 394], [391, 469]]</point>"}]}
{"id": 884, "image": ["rtx_frames_success_4/49_bridge#episode_13535/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13545/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot and plate."}, {"from": "gpt", "value": "<point>[[423, 535], [507, 469], [577, 451], [683, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[225, 319], [313, 404], [412, 465], [574, 498]]</point>"}]}
{"id": 886, "image": ["rtx_frames_success_4/49_bridge#episode_13560/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13561/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[676, 272], [567, 296], [489, 366], [472, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[588, 338], [507, 277], [405, 258], [261, 263]]</point>"}]}
{"id": 888, "image": ["rtx_frames_success_4/49_bridge#episode_13568/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13568/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the green bell pepper."}, {"from": "gpt", "value": "<point>[[630, 615], [574, 563], [507, 554], [401, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the green bell pepper on the stove top."}, {"from": "gpt", "value": "<point>[[345, 573], [296, 512], [222, 474], [134, 479]]</point>"}]}
{"id": 890, "image": ["rtx_frames_success_4/49_bridge#episode_13589/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13596/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[690, 399], [563, 385], [461, 418], [349, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the pot."}, {"from": "gpt", "value": "<point>[[504, 535], [468, 488], [405, 484], [342, 488]]</point>"}]}
{"id": 892, "image": ["rtx_frames_success_4/49_bridge#episode_13597/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13599/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[415, 300], [496, 432], [539, 516], [563, 643]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[454, 300], [384, 347], [342, 413], [342, 596]]</point>"}]}
{"id": 894, "image": ["rtx_frames_success_4/49_bridge#episode_13615/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13635/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[627, 282], [468, 338], [377, 394], [299, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[651, 390], [447, 376], [327, 423], [201, 484]]</point>"}]}
{"id": 896, "image": ["rtx_frames_success_4/49_bridge#episode_13635/frame_15.png", "rtx_frames_success_4/49_bridge#episode_13637/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the towel."}, {"from": "gpt", "value": "<point>[[415, 418], [535, 451], [655, 493], [715, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[458, 122], [338, 188], [275, 286], [239, 479]]</point>"}]}
{"id": 898, "image": ["rtx_frames_success_4/49_bridge#episode_13646/frame_0.png", "rtx_frames_success_4/49_bridge#episode_13658/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[486, 146], [595, 263], [676, 380], [750, 676]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[535, 272], [623, 329], [658, 404], [690, 516]]</point>"}]}
{"id": 900, "image": ["rtx_frames_success_4/49_bridge#episode_13658/frame_15.png", "rtx_frames_success_4/49_bridge#episode_13689/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the fork to the left of the cloth."}, {"from": "gpt", "value": "<point>[[739, 469], [648, 385], [514, 362], [366, 385]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[504, 423], [518, 451], [546, 479], [577, 451]]</point>"}]}
{"id": 902, "image": ["rtx_frames_success_5/49_bridge#episode_142/frame_0.png", "rtx_frames_success_5/49_bridge#episode_149/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[540, 432], [452, 321], [476, 225]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi in the bowl."}, {"from": "gpt", "value": "<point>[[258, 380], [391, 440], [488, 384]]</point>"}]}
{"id": 904, "image": ["rtx_frames_success_5/49_bridge#episode_1407/frame_0.png", "rtx_frames_success_5/49_bridge#episode_1409/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red spoon."}, {"from": "gpt", "value": "<point>[[258, 284], [657, 584], [726, 698]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the apple."}, {"from": "gpt", "value": "<point>[[530, 560], [341, 388], [238, 414]]</point>"}]}
{"id": 906, "image": ["rtx_frames_success_5/49_bridge#episode_1411/frame_0.png", "rtx_frames_success_5/49_bridge#episode_1413/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[274, 247], [205, 380], [283, 528]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[463, 347], [640, 421], [717, 502]]</point>"}]}
{"id": 908, "image": ["rtx_frames_success_5/49_bridge#episode_1457/frame_0.png", "rtx_frames_success_5/49_bridge#episode_1459/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[474, 451], [283, 421], [235, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the sushi roll."}, {"from": "gpt", "value": "<point>[[269, 295], [385, 247], [435, 340]]</point>"}]}
{"id": 910, "image": ["rtx_frames_success_5/49_bridge#episode_1459/frame_15.png", "rtx_frames_success_5/49_bridge#episode_1464/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the sushi roll over the bowl."}, {"from": "gpt", "value": "<point>[[576, 126], [529, 329], [620, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl containing corn."}, {"from": "gpt", "value": "<point>[[535, 307], [554, 484], [634, 572]]</point>"}]}
{"id": 912, "image": ["rtx_frames_success_5/49_bridge#episode_1470/frame_0.png", "rtx_frames_success_5/49_bridge#episode_1483/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[260, 63], [493, 332], [582, 295]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[587, 543], [407, 646], [374, 606]]</point>"}]}
{"id": 914, "image": ["rtx_frames_success_5/49_bridge#episode_1489/frame_15.png", "rtx_frames_success_5/49_bridge#episode_1491/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bowl towards the cloth."}, {"from": "gpt", "value": "<point>[[247, 384], [313, 524], [435, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple cloth."}, {"from": "gpt", "value": "<point>[[452, 225], [582, 299], [651, 395]]</point>"}]}
{"id": 916, "image": ["rtx_frames_success_5/49_bridge#episode_1500/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13696/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[482, 288], [648, 646], [737, 702]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bowl towards the towel."}, {"from": "gpt", "value": "<point>[[560, 351], [543, 635], [632, 702]]</point>"}]}
{"id": 918, "image": ["rtx_frames_success_5/49_bridge#episode_13710/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13716/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the pot."}, {"from": "gpt", "value": "<point>[[587, 178], [421, 405], [429, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the steel pan."}, {"from": "gpt", "value": "<point>[[399, 151], [465, 325], [496, 436]]</point>"}]}
{"id": 920, "image": ["rtx_frames_success_5/49_bridge#episode_13716/frame_15.png", "rtx_frames_success_5/49_bridge#episode_13728/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the steel pan to the left corner of the table."}, {"from": "gpt", "value": "<point>[[377, 247], [216, 392], [161, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[515, 403], [687, 425], [837, 469]]</point>"}]}
{"id": 922, "image": ["rtx_frames_success_5/49_bridge#episode_13728/frame_15.png", "rtx_frames_success_5/49_bridge#episode_13729/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[740, 488], [277, 447], [197, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi on the plate."}, {"from": "gpt", "value": "<point>[[554, 192], [731, 284], [806, 406]]</point>"}]}
{"id": 924, "image": ["rtx_frames_success_5/49_bridge#episode_13752/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13779/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green towel."}, {"from": "gpt", "value": "<point>[[584, 188], [319, 314], [263, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[526, 66], [404, 380], [380, 476]]</point>"}]}
{"id": 926, "image": ["rtx_frames_success_5/49_bridge#episode_13779/frame_15.png", "rtx_frames_success_5/49_bridge#episode_13789/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife towards the right side of the board."}, {"from": "gpt", "value": "<point>[[380, 458], [701, 558], [789, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[438, 222], [319, 414], [283, 499]]</point>"}]}
{"id": 928, "image": ["rtx_frames_success_5/49_bridge#episode_13811/frame_15.png", "rtx_frames_success_5/49_bridge#episode_13812/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the cans and red lid."}, {"from": "gpt", "value": "<point>[[416, 240], [568, 366], [659, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the food item."}, {"from": "gpt", "value": "<point>[[504, 199], [640, 403], [643, 510]]</point>"}]}
{"id": 930, "image": ["rtx_frames_success_5/49_bridge#episode_13812/frame_15.png", "rtx_frames_success_5/49_bridge#episode_13815/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the food item towards the pot."}, {"from": "gpt", "value": "<point>[[338, 355], [252, 355], [199, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[535, 111], [454, 506], [482, 646]]</point>"}]}
{"id": 932, "image": ["rtx_frames_success_5/49_bridge#episode_13821/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13825/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[446, 103], [512, 281], [524, 362]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the round brush."}, {"from": "gpt", "value": "<point>[[607, 236], [726, 606], [640, 717]]</point>"}]}
{"id": 934, "image": ["rtx_frames_success_5/49_bridge#episode_13836/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13844/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[604, 259], [507, 620], [568, 742]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[144, 229], [388, 561], [432, 669]]</point>"}]}
{"id": 936, "image": ["rtx_frames_success_5/49_bridge#episode_13859/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13859/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot in the pot."}, {"from": "gpt", "value": "<point>[[310, 266], [496, 299], [562, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the carrot to the pot to the right corner of the table."}, {"from": "gpt", "value": "<point>[[485, 236], [247, 343], [166, 414]]</point>"}]}
{"id": 938, "image": ["rtx_frames_success_5/49_bridge#episode_13864/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13868/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[235, 336], [618, 259], [737, 355]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[393, 336], [291, 480], [296, 591]]</point>"}]}
{"id": 940, "image": ["rtx_frames_success_5/49_bridge#episode_13886/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13896/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cup on the plate."}, {"from": "gpt", "value": "<point>[[213, 362], [404, 262], [474, 266]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[493, 288], [388, 303], [316, 369]]</point>"}]}
{"id": 942, "image": ["rtx_frames_success_5/49_bridge#episode_13898/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13906/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[382, 251], [645, 565], [712, 687]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bowl towards the stove."}, {"from": "gpt", "value": "<point>[[452, 332], [233, 380], [144, 480]]</point>"}]}
{"id": 944, "image": ["rtx_frames_success_5/49_bridge#episode_13925/frame_15.png", "rtx_frames_success_5/49_bridge#episode_13930/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the rag."}, {"from": "gpt", "value": "<point>[[416, 177], [393, 554], [499, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[327, 347], [213, 480], [139, 469]]</point>"}]}
{"id": 946, "image": ["rtx_frames_success_5/49_bridge#episode_13930/frame_15.png", "rtx_frames_success_5/49_bridge#episode_13933/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the orange cloth."}, {"from": "gpt", "value": "<point>[[163, 484], [532, 403], [623, 458]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi on the plate."}, {"from": "gpt", "value": "<point>[[355, 476], [524, 351], [609, 414]]</point>"}]}
{"id": 948, "image": ["rtx_frames_success_5/49_bridge#episode_13944/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13949/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[263, 63], [643, 657], [731, 672]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[277, 329], [457, 414], [518, 506]]</point>"}]}
{"id": 950, "image": ["rtx_frames_success_5/49_bridge#episode_13951/frame_15.png", "rtx_frames_success_5/49_bridge#episode_13957/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the toy hammer towards the pot."}, {"from": "gpt", "value": "<point>[[504, 288], [413, 270], [377, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[274, 144], [601, 502], [693, 462]]</point>"}]}
{"id": 952, "image": ["rtx_frames_success_5/49_bridge#episode_13958/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13959/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom toy in the bowl."}, {"from": "gpt", "value": "<point>[[269, 414], [404, 657], [548, 657]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the ball towards the orange cloth."}, {"from": "gpt", "value": "<point>[[731, 458], [662, 687], [576, 709]]</point>"}]}
{"id": 954, "image": ["rtx_frames_success_5/49_bridge#episode_13972/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13980/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the steel pan."}, {"from": "gpt", "value": "<point>[[543, 70], [607, 384], [596, 595]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[241, 247], [524, 635], [590, 657]]</point>"}]}
{"id": 956, "image": ["rtx_frames_success_5/49_bridge#episode_13984/frame_0.png", "rtx_frames_success_5/49_bridge#episode_13996/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli in the cardboard box."}, {"from": "gpt", "value": "<point>[[562, 307], [460, 606], [479, 731]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the food."}, {"from": "gpt", "value": "<point>[[429, 251], [607, 236], [695, 399]]</point>"}]}
{"id": 958, "image": ["rtx_frames_success_5/49_bridge#episode_14000/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14008/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the strawberry."}, {"from": "gpt", "value": "<point>[[501, 236], [493, 343], [568, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[233, 255], [548, 510], [604, 639]]</point>"}]}
{"id": 960, "image": ["rtx_frames_success_5/49_bridge#episode_14012/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14012/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[183, 314], [241, 340], [321, 218]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[382, 255], [385, 584], [496, 753]]</point>"}]}
{"id": 962, "image": ["rtx_frames_success_5/49_bridge#episode_14019/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14027/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the blueberries in the sink."}, {"from": "gpt", "value": "<point>[[396, 259], [427, 399], [385, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon in the sink."}, {"from": "gpt", "value": "<point>[[543, 259], [457, 343], [393, 347]]</point>"}]}
{"id": 964, "image": ["rtx_frames_success_5/49_bridge#episode_14028/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14037/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[507, 432], [388, 443], [410, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the slice of sushi in the metal bowl."}, {"from": "gpt", "value": "<point>[[668, 266], [590, 606], [643, 720]]</point>"}]}
{"id": 966, "image": ["rtx_frames_success_5/49_bridge#episode_14037/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14044/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the sushi slice towards the green cloth."}, {"from": "gpt", "value": "<point>[[554, 373], [266, 414], [180, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[438, 432], [313, 406], [288, 332]]</point>"}]}
{"id": 968, "image": ["rtx_frames_success_5/49_bridge#episode_14044/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14089/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[274, 351], [479, 451], [623, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[418, 92], [512, 635], [593, 717]]</point>"}]}
{"id": 970, "image": ["rtx_frames_success_5/49_bridge#episode_14101/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14101/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the can."}, {"from": "gpt", "value": "<point>[[357, 155], [291, 373], [324, 491]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the can."}, {"from": "gpt", "value": "<point>[[435, 473], [598, 513], [676, 443]]</point>"}]}
{"id": 972, "image": ["rtx_frames_success_5/49_bridge#episode_14108/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14111/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange pot."}, {"from": "gpt", "value": "<point>[[543, 539], [391, 414], [396, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[490, 214], [521, 362], [488, 476]]</point>"}]}
{"id": 974, "image": ["rtx_frames_success_5/49_bridge#episode_14116/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14123/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal bowl."}, {"from": "gpt", "value": "<point>[[388, 292], [565, 447], [593, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula on the plate."}, {"from": "gpt", "value": "<point>[[587, 358], [557, 440], [587, 491]]</point>"}]}
{"id": 976, "image": ["rtx_frames_success_5/49_bridge#episode_14130/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14131/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[396, 351], [565, 598], [596, 765]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[482, 233], [579, 595], [662, 669]]</point>"}]}
{"id": 978, "image": ["rtx_frames_success_5/49_bridge#episode_14140/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14151/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[335, 255], [565, 713], [654, 720]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bread in the pot."}, {"from": "gpt", "value": "<point>[[584, 347], [645, 561], [731, 572]]</point>"}]}
{"id": 980, "image": ["rtx_frames_success_5/49_bridge#episode_14173/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14184/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the object towards the edge of the surface."}, {"from": "gpt", "value": "<point>[[557, 277], [679, 358], [767, 476]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the blue brush."}, {"from": "gpt", "value": "<point>[[283, 222], [211, 410], [125, 443]]</point>"}]}
{"id": 982, "image": ["rtx_frames_success_5/49_bridge#episode_14184/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14197/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the blue brush behind the pot."}, {"from": "gpt", "value": "<point>[[161, 388], [510, 414], [648, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[482, 517], [332, 624], [222, 580]]</point>"}]}
{"id": 984, "image": ["rtx_frames_success_5/49_bridge#episode_14212/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14239/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to release the spoon near the canned goods."}, {"from": "gpt", "value": "<point>[[371, 362], [568, 484], [662, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the yellow pepper towards the green towel."}, {"from": "gpt", "value": "<point>[[413, 347], [490, 624], [579, 720]]</point>"}]}
{"id": 986, "image": ["rtx_frames_success_5/49_bridge#episode_14242/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14244/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot in the bowl."}, {"from": "gpt", "value": "<point>[[479, 218], [493, 510], [546, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[294, 321], [294, 473], [247, 491]]</point>"}]}
{"id": 988, "image": ["rtx_frames_success_5/49_bridge#episode_14268/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14268/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the piece of food."}, {"from": "gpt", "value": "<point>[[654, 462], [454, 462], [382, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the food item to a new location."}, {"from": "gpt", "value": "<point>[[360, 403], [501, 580], [632, 620]]</point>"}]}
{"id": 990, "image": ["rtx_frames_success_5/49_bridge#episode_14286/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14300/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[343, 321], [252, 484], [169, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[474, 340], [393, 536], [319, 550]]</point>"}]}
{"id": 992, "image": ["rtx_frames_success_5/49_bridge#episode_14333/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14339/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[389, 226], [476, 417], [548, 443]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[540, 244], [346, 399], [305, 480]]</point>"}]}
{"id": 994, "image": ["rtx_frames_success_5/49_bridge#episode_14339/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14353/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pot on front of the banana."}, {"from": "gpt", "value": "<point>[[382, 447], [485, 528], [532, 669]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[554, 488], [343, 587], [263, 513]]</point>"}]}
{"id": 996, "image": ["rtx_frames_success_5/49_bridge#episode_14353/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14358/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the strawberry to the back right corner of the table."}, {"from": "gpt", "value": "<point>[[252, 495], [596, 428], [729, 473]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[388, 144], [623, 595], [712, 687]]</point>"}]}
{"id": 998, "image": ["rtx_frames_success_5/49_bridge#episode_14358/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14371/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot to the left edge of the table."}, {"from": "gpt", "value": "<point>[[457, 329], [211, 484], [91, 476]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green towel."}, {"from": "gpt", "value": "<point>[[551, 85], [399, 451], [283, 495]]</point>"}]}
{"id": 1000, "image": ["rtx_frames_success_5/49_bridge#episode_14381/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14381/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[197, 303], [407, 277], [446, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the bottom right corner."}, {"from": "gpt", "value": "<point>[[457, 336], [446, 598], [499, 750]]</point>"}]}
{"id": 1002, "image": ["rtx_frames_success_5/49_bridge#episode_14385/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14387/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[432, 192], [263, 384], [266, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the yellow pepper in the pot."}, {"from": "gpt", "value": "<point>[[615, 292], [435, 476], [485, 646]]</point>"}]}
{"id": 1004, "image": ["rtx_frames_success_5/49_bridge#episode_14392/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14394/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the vessel."}, {"from": "gpt", "value": "<point>[[443, 284], [285, 580], [222, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue spoon."}, {"from": "gpt", "value": "<point>[[235, 325], [321, 277], [388, 347]]</point>"}]}
{"id": 1006, "image": ["rtx_frames_success_5/49_bridge#episode_14394/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14403/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the green cloth."}, {"from": "gpt", "value": "<point>[[391, 321], [560, 657], [679, 705]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[382, 155], [155, 392], [100, 476]]</point>"}]}
{"id": 1008, "image": ["rtx_frames_success_5/49_bridge#episode_14411/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14428/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[407, 388], [543, 709], [679, 717]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[382, 192], [537, 447], [637, 454]]</point>"}]}
{"id": 1010, "image": ["rtx_frames_success_5/49_bridge#episode_14428/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14432/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the corn towards the bowl."}, {"from": "gpt", "value": "<point>[[648, 465], [460, 428], [385, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[427, 70], [510, 362], [499, 488]]</point>"}]}
{"id": 1012, "image": ["rtx_frames_success_5/49_bridge#episode_14434/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14443/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pepper shaker."}, {"from": "gpt", "value": "<point>[[645, 358], [496, 329], [416, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot towards the left edge of the table."}, {"from": "gpt", "value": "<point>[[463, 247], [194, 414], [61, 484]]</point>"}]}
{"id": 1014, "image": ["rtx_frames_success_5/49_bridge#episode_14447/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14448/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower over the bowl."}, {"from": "gpt", "value": "<point>[[543, 321], [629, 591], [715, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[446, 211], [501, 458], [565, 524]]</point>"}]}
{"id": 1016, "image": ["rtx_frames_success_5/49_bridge#episode_14451/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14451/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[501, 199], [601, 425], [693, 458]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the right bottom corner of the table."}, {"from": "gpt", "value": "<point>[[712, 384], [665, 669], [723, 805]]</point>"}]}
{"id": 1018, "image": ["rtx_frames_success_5/49_bridge#episode_14472/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14478/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position above the pot."}, {"from": "gpt", "value": "<point>[[532, 236], [468, 558], [526, 635]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the capsicum."}, {"from": "gpt", "value": "<point>[[476, 410], [296, 384], [258, 292]]</point>"}]}
{"id": 1020, "image": ["rtx_frames_success_5/49_bridge#episode_14478/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14484/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the capsicum towards the cloth."}, {"from": "gpt", "value": "<point>[[269, 174], [416, 550], [504, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow cloth."}, {"from": "gpt", "value": "<point>[[396, 188], [249, 351], [186, 476]]</point>"}]}
{"id": 1022, "image": ["rtx_frames_success_5/49_bridge#episode_14493/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14513/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[515, 126], [360, 336], [341, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice container."}, {"from": "gpt", "value": "<point>[[374, 292], [399, 628], [479, 691]]</point>"}]}
{"id": 1024, "image": ["rtx_frames_success_5/49_bridge#episode_14522/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14540/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[515, 554], [269, 358], [227, 403]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[551, 222], [543, 332], [490, 417]]</point>"}]}
{"id": 1026, "image": ["rtx_frames_success_5/49_bridge#episode_14541/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14541/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue scoop."}, {"from": "gpt", "value": "<point>[[258, 170], [371, 262], [402, 355]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the blue scoop towards the metal pot."}, {"from": "gpt", "value": "<point>[[391, 229], [205, 469], [116, 473]]</point>"}]}
{"id": 1028, "image": ["rtx_frames_success_5/49_bridge#episode_14550/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14577/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[504, 244], [593, 414], [623, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple cloth."}, {"from": "gpt", "value": "<point>[[468, 222], [557, 281], [634, 392]]</point>"}]}
{"id": 1030, "image": ["rtx_frames_success_5/49_bridge#episode_14580/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14583/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[368, 181], [172, 351], [119, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot lid."}, {"from": "gpt", "value": "<point>[[510, 170], [396, 247], [335, 188]]</point>"}]}
{"id": 1032, "image": ["rtx_frames_success_5/49_bridge#episode_14597/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14597/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the bowl."}, {"from": "gpt", "value": "<point>[[518, 399], [490, 584], [546, 669]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bowl towards the burner."}, {"from": "gpt", "value": "<point>[[485, 521], [280, 362], [285, 292]]</point>"}]}
{"id": 1034, "image": ["rtx_frames_success_5/49_bridge#episode_14626/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14643/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[324, 329], [349, 532], [424, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bowl."}, {"from": "gpt", "value": "<point>[[476, 395], [391, 491], [313, 458]]</point>"}]}
{"id": 1036, "image": ["rtx_frames_success_5/49_bridge#episode_14644/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14655/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[510, 403], [468, 643], [557, 728]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the cloth."}, {"from": "gpt", "value": "<point>[[524, 103], [474, 495], [407, 606]]</point>"}]}
{"id": 1038, "image": ["rtx_frames_success_5/49_bridge#episode_14666/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14666/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[568, 321], [258, 321], [216, 218]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the microwave door shut."}, {"from": "gpt", "value": "<point>[[399, 244], [709, 188], [762, 4]]</point>"}]}
{"id": 1040, "image": ["rtx_frames_success_5/49_bridge#episode_14687/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14696/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cone."}, {"from": "gpt", "value": "<point>[[548, 211], [224, 369], [136, 443]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[421, 81], [402, 458], [476, 558]]</point>"}]}
{"id": 1042, "image": ["rtx_frames_success_5/49_bridge#episode_14726/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14766/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lemon."}, {"from": "gpt", "value": "<point>[[256, 359], [558, 352], [659, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the knife."}, {"from": "gpt", "value": "<point>[[396, 155], [496, 620], [562, 650]]</point>"}]}
{"id": 1044, "image": ["rtx_frames_success_5/49_bridge#episode_14795/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14798/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[252, 211], [424, 480], [529, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[241, 436], [582, 816], [706, 772]]</point>"}]}
{"id": 1046, "image": ["rtx_frames_success_5/49_bridge#episode_14798/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14810/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to slide the can towards the pan."}, {"from": "gpt", "value": "<point>[[404, 550], [202, 517], [144, 395]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the light green object."}, {"from": "gpt", "value": "<point>[[305, 229], [338, 347], [388, 325]]</point>"}]}
{"id": 1048, "image": ["rtx_frames_success_5/49_bridge#episode_14825/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14827/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[554, 214], [640, 650], [609, 742]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[357, 181], [341, 410], [291, 432]]</point>"}]}
{"id": 1050, "image": ["rtx_frames_success_5/49_bridge#episode_14854/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14854/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[454, 373], [551, 646], [645, 720]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the green towel."}, {"from": "gpt", "value": "<point>[[368, 410], [252, 510], [166, 473]]</point>"}]}
{"id": 1052, "image": ["rtx_frames_success_5/49_bridge#episode_14857/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14865/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[404, 211], [460, 539], [402, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the yellow cube from the container."}, {"from": "gpt", "value": "<point>[[277, 440], [175, 414], [211, 488]]</point>"}]}
{"id": 1054, "image": ["rtx_frames_success_5/49_bridge#episode_14865/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14870/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the yellow cube over the countertop."}, {"from": "gpt", "value": "<point>[[235, 421], [355, 384], [468, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[543, 126], [598, 646], [665, 698]]</point>"}]}
{"id": 1056, "image": ["rtx_frames_success_5/49_bridge#episode_14870/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14871/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon to the left."}, {"from": "gpt", "value": "<point>[[391, 118], [260, 406], [183, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[141, 384], [449, 713], [593, 698]]</point>"}]}
{"id": 1058, "image": ["rtx_frames_success_5/49_bridge#episode_14872/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14873/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan towards the pickle."}, {"from": "gpt", "value": "<point>[[429, 292], [435, 536], [488, 680]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[316, 214], [163, 469], [188, 561]]</point>"}]}
{"id": 1060, "image": ["rtx_frames_success_5/49_bridge#episode_14873/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14876/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the sauce cans."}, {"from": "gpt", "value": "<point>[[158, 558], [526, 473], [596, 395]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[416, 129], [665, 329], [709, 392]]</point>"}]}
{"id": 1062, "image": ["rtx_frames_success_5/49_bridge#episode_14882/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14901/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the green towel."}, {"from": "gpt", "value": "<point>[[529, 100], [429, 440], [321, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[421, 332], [435, 524], [488, 591]]</point>"}]}
{"id": 1064, "image": ["rtx_frames_success_5/49_bridge#episode_14903/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14903/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[374, 174], [645, 369], [715, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the mushroom."}, {"from": "gpt", "value": "<point>[[554, 188], [352, 558], [241, 598]]</point>"}]}
{"id": 1066, "image": ["rtx_frames_success_5/49_bridge#episode_14904/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14909/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[507, 259], [454, 550], [535, 635]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[418, 85], [291, 436], [235, 428]]</point>"}]}
{"id": 1068, "image": ["rtx_frames_success_5/49_bridge#episode_14909/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14913/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon to the bottom right of the table."}, {"from": "gpt", "value": "<point>[[393, 199], [623, 665], [715, 753]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spice container."}, {"from": "gpt", "value": "<point>[[454, 343], [222, 388], [205, 299]]</point>"}]}
{"id": 1070, "image": ["rtx_frames_success_5/49_bridge#episode_14918/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14918/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the fork."}, {"from": "gpt", "value": "<point>[[474, 369], [557, 624], [659, 665]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the fork on the yellow towel."}, {"from": "gpt", "value": "<point>[[521, 421], [274, 488], [247, 432]]</point>"}]}
{"id": 1072, "image": ["rtx_frames_success_5/49_bridge#episode_14920/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14940/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the cheese to the left of the pot."}, {"from": "gpt", "value": "<point>[[418, 255], [324, 299], [283, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[499, 151], [432, 307], [338, 262]]</point>"}]}
{"id": 1074, "image": ["rtx_frames_success_5/49_bridge#episode_14940/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14941/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[321, 292], [255, 366], [310, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red pepper."}, {"from": "gpt", "value": "<point>[[352, 225], [532, 576], [629, 606]]</point>"}]}
{"id": 1076, "image": ["rtx_frames_success_5/49_bridge#episode_14957/frame_0.png", "rtx_frames_success_5/49_bridge#episode_14957/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red spoon."}, {"from": "gpt", "value": "<point>[[413, 170], [285, 392], [213, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon over the wok."}, {"from": "gpt", "value": "<point>[[258, 199], [537, 598], [657, 657]]</point>"}]}
{"id": 1078, "image": ["rtx_frames_success_5/49_bridge#episode_14960/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14964/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the drumstick over the table."}, {"from": "gpt", "value": "<point>[[327, 270], [643, 388], [715, 513]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach into the laundry machine."}, {"from": "gpt", "value": "<point>[[53, 410], [114, 488], [100, 643]]</point>"}]}
{"id": 1080, "image": ["rtx_frames_success_5/49_bridge#episode_14964/frame_15.png", "rtx_frames_success_5/49_bridge#episode_14966/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the clothing out of the machine."}, {"from": "gpt", "value": "<point>[[213, 465], [291, 591], [269, 665]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[338, 181], [446, 484], [404, 569]]</point>"}]}
{"id": 1082, "image": ["rtx_frames_success_5/49_bridge#episode_14986/frame_0.png", "rtx_frames_success_5/49_bridge#episode_15003/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[496, 151], [399, 432], [391, 528]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the plate."}, {"from": "gpt", "value": "<point>[[729, 510], [615, 720], [490, 709]]</point>"}]}
{"id": 1084, "image": ["rtx_frames_success_5/49_bridge#episode_15010/frame_0.png", "rtx_frames_success_5/49_bridge#episode_15015/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the towel."}, {"from": "gpt", "value": "<point>[[385, 218], [501, 484], [449, 628]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[626, 303], [596, 602], [518, 654]]</point>"}]}
{"id": 1086, "image": ["rtx_frames_success_5/49_bridge#episode_15020/frame_15.png", "rtx_frames_success_5/49_bridge#episode_15033/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the napkin."}, {"from": "gpt", "value": "<point>[[615, 369], [380, 550], [343, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[363, 694], [490, 236], [565, 177]]</point>"}]}
{"id": 1088, "image": ["rtx_frames_success_5/49_bridge#episode_15038/frame_0.png", "rtx_frames_success_5/49_bridge#episode_15049/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[720, 380], [720, 742], [657, 809]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[493, 211], [510, 440], [562, 521]]</point>"}]}
{"id": 1090, "image": ["rtx_frames_success_5/49_bridge#episode_15049/frame_15.png", "rtx_frames_success_5/49_bridge#episode_15050/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the bottom right edge of the table."}, {"from": "gpt", "value": "<point>[[704, 495], [690, 650], [596, 805]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach into the laundry machine."}, {"from": "gpt", "value": "<point>[[219, 284], [396, 222], [476, 288]]</point>"}]}
{"id": 1092, "image": ["rtx_frames_success_5/49_bridge#episode_15057/frame_0.png", "rtx_frames_success_5/49_bridge#episode_15061/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab a piece of clothing from the machine."}, {"from": "gpt", "value": "<point>[[266, 635], [416, 465], [421, 358]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[668, 222], [258, 273], [172, 332]]</point>"}]}
{"id": 1094, "image": ["rtx_frames_success_5/49_bridge#episode_15065/frame_0.png", "rtx_frames_success_5/49_bridge#episode_15070/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the yellow towel."}, {"from": "gpt", "value": "<point>[[482, 273], [598, 388], [637, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the eggplant towards the spatula."}, {"from": "gpt", "value": "<point>[[357, 233], [310, 436], [380, 521]]</point>"}]}
{"id": 1096, "image": ["rtx_frames_success_5/49_bridge#episode_15072/frame_15.png", "rtx_frames_success_6/49_bridge#episode_165/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan towards the stove."}, {"from": "gpt", "value": "<point>[[532, 403], [554, 595], [632, 702]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[602, 298], [605, 411], [602, 514], [590, 634]]</point>"}]}
{"id": 1098, "image": ["rtx_frames_success_6/49_bridge#episode_1519/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1525/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the item."}, {"from": "gpt", "value": "<point>[[239, 201], [235, 256], [250, 313], [274, 373]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[447, 170], [376, 228], [314, 301], [239, 411]]</point>"}]}
{"id": 1100, "image": ["rtx_frames_success_6/49_bridge#episode_1525/frame_15.png", "rtx_frames_success_6/49_bridge#episode_1529/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush towards the towel."}, {"from": "gpt", "value": "<point>[[314, 366], [432, 278], [577, 298], [617, 446]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[184, 185], [258, 206], [320, 253], [372, 321]]</point>"}]}
{"id": 1102, "image": ["rtx_frames_success_6/49_bridge#episode_1537/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1555/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position above the green towel."}, {"from": "gpt", "value": "<point>[[688, 40], [707, 198], [712, 363], [701, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[150, 208], [372, 243], [581, 338], [727, 486]]</point>"}]}
{"id": 1104, "image": ["rtx_frames_success_6/49_bridge#episode_1555/frame_15.png", "rtx_frames_success_6/49_bridge#episode_1570/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[724, 446], [701, 561], [671, 662], [630, 777], [624, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot containing the broccoli."}, {"from": "gpt", "value": "<point>[[538, 576], [573, 529], [622, 486], [667, 461]]</point>"}]}
{"id": 1106, "image": ["rtx_frames_success_6/49_bridge#episode_1576/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1578/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the rag."}, {"from": "gpt", "value": "<point>[[220, 223], [406, 321], [545, 419], [675, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[269, 150], [367, 221], [466, 311], [541, 404]]</point>"}]}
{"id": 1108, "image": ["rtx_frames_success_6/49_bridge#episode_1582/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1586/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[585, 383], [564, 481], [560, 624], [568, 717]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the eggplant."}, {"from": "gpt", "value": "<point>[[350, 135], [461, 263], [583, 466], [677, 662]]</point>"}]}
{"id": 1110, "image": ["rtx_frames_success_6/49_bridge#episode_1588/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1589/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the blue pen."}, {"from": "gpt", "value": "<point>[[222, 168], [417, 301], [647, 561], [788, 802]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the small metal bowl."}, {"from": "gpt", "value": "<point>[[231, 383], [342, 424], [417, 476], [485, 561]]</point>"}]}
{"id": 1112, "image": ["rtx_frames_success_6/49_bridge#episode_1602/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1629/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[662, 306], [577, 396], [476, 481], [363, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[173, 138], [252, 180], [329, 241], [391, 296]]</point>"}]}
{"id": 1114, "image": ["rtx_frames_success_6/49_bridge#episode_1630/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1630/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[380, 333], [318, 353], [250, 401], [199, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the brush between a sushi rool and a pot."}, {"from": "gpt", "value": "<point>[[186, 378], [295, 409], [402, 421], [513, 419]]</point>"}]}
{"id": 1116, "image": ["rtx_frames_success_6/49_bridge#episode_1631/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1631/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the lemon."}, {"from": "gpt", "value": "<point>[[180, 125], [382, 298], [635, 551], [789, 737]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hold the lemon."}, {"from": "gpt", "value": "<point>[[791, 654], [694, 476], [609, 388], [477, 343]]</point>"}]}
{"id": 1118, "image": ["rtx_frames_success_6/49_bridge#episode_1632/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1633/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red spatula."}, {"from": "gpt", "value": "<point>[[391, 153], [348, 308], [310, 424], [246, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the sponge."}, {"from": "gpt", "value": "<point>[[395, 381], [449, 426], [530, 479], [602, 506]]</point>"}]}
{"id": 1120, "image": ["rtx_frames_success_6/49_bridge#episode_1641/frame_0.png", "rtx_frames_success_6/49_bridge#episode_1644/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[335, 150], [442, 246], [515, 333], [581, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[242, 226], [423, 271], [568, 316], [733, 396]]</point>"}]}
{"id": 1122, "image": ["rtx_frames_success_6/49_bridge#episode_15106/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15119/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[496, 281], [404, 281], [312, 296], [224, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[320, 188], [320, 246], [338, 296], [372, 321]]</point>"}]}
{"id": 1124, "image": ["rtx_frames_success_6/49_bridge#episode_15121/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15140/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet handle."}, {"from": "gpt", "value": "<point>[[825, 536], [801, 619], [744, 687], [686, 724]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[453, 125], [293, 183], [152, 266], [17, 401]]</point>"}]}
{"id": 1126, "image": ["rtx_frames_success_6/49_bridge#episode_15140/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15150/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to knock the metal pot over."}, {"from": "gpt", "value": "<point>[[36, 371], [128, 278], [201, 308], [180, 346]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red pepper."}, {"from": "gpt", "value": "<point>[[162, 198], [286, 298], [442, 356], [588, 393]]</point>"}]}
{"id": 1128, "image": ["rtx_frames_success_6/49_bridge#episode_15162/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15163/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[470, 258], [434, 286], [387, 296], [342, 283]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the stuffed animal."}, {"from": "gpt", "value": "<point>[[669, 201], [714, 286], [726, 346], [729, 426]]</point>"}]}
{"id": 1130, "image": ["rtx_frames_success_6/49_bridge#episode_15179/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15205/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[335, 170], [391, 258], [432, 356], [474, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow towel."}, {"from": "gpt", "value": "<point>[[427, 0], [530, 113], [652, 301], [737, 476]]</point>"}]}
{"id": 1132, "image": ["rtx_frames_success_6/49_bridge#episode_15216/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15242/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[440, 160], [530, 218], [626, 326], [695, 431]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[316, 25], [244, 113], [150, 321], [98, 486]]</point>"}]}
{"id": 1134, "image": ["rtx_frames_success_6/49_bridge#episode_15242/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15246/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[92, 446], [212, 409], [338, 406], [442, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the egg."}, {"from": "gpt", "value": "<point>[[639, 271], [519, 281], [383, 323], [254, 386]]</point>"}]}
{"id": 1136, "image": ["rtx_frames_success_6/49_bridge#episode_15246/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15283/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the egg into the pot."}, {"from": "gpt", "value": "<point>[[273, 391], [306, 336], [344, 296], [408, 276]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spice bottle."}, {"from": "gpt", "value": "<point>[[237, 301], [357, 386], [455, 539], [511, 689]]</point>"}]}
{"id": 1138, "image": ["rtx_frames_success_6/49_bridge#episode_15290/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15306/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green towel."}, {"from": "gpt", "value": "<point>[[295, 261], [303, 298], [299, 336], [280, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[714, 266], [635, 311], [528, 326], [408, 308]]</point>"}]}
{"id": 1140, "image": ["rtx_frames_success_6/49_bridge#episode_15315/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15321/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[583, 125], [462, 223], [357, 343], [252, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[615, 434], [492, 298], [368, 288], [291, 424]]</point>"}]}
{"id": 1142, "image": ["rtx_frames_success_6/49_bridge#episode_15321/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15325/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the oven door closed."}, {"from": "gpt", "value": "<point>[[709, 140], [633, 193], [617, 273], [695, 368], [763, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom."}, {"from": "gpt", "value": "<point>[[344, 258], [367, 353], [378, 459], [376, 564]]</point>"}]}
{"id": 1144, "image": ["rtx_frames_success_6/49_bridge#episode_15340/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15340/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the scoop."}, {"from": "gpt", "value": "<point>[[203, 291], [355, 368], [496, 539], [583, 722]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the scoop towards the bread."}, {"from": "gpt", "value": "<point>[[583, 657], [496, 569], [427, 436], [383, 318]]</point>"}]}
{"id": 1146, "image": ["rtx_frames_success_6/49_bridge#episode_15344/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15348/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[746, 454], [630, 459], [523, 494], [417, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula in the sink."}, {"from": "gpt", "value": "<point>[[718, 346], [660, 366], [626, 386], [575, 449]]</point>"}]}
{"id": 1148, "image": ["rtx_frames_success_6/49_bridge#episode_15352/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15377/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[485, 175], [481, 358], [496, 544], [523, 712]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue fork."}, {"from": "gpt", "value": "<point>[[331, 150], [414, 303], [515, 561], [564, 729]]</point>"}]}
{"id": 1150, "image": ["rtx_frames_success_6/49_bridge#episode_15377/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15380/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the blue fork over the pan."}, {"from": "gpt", "value": "<point>[[581, 591], [470, 544], [310, 491], [201, 479], [199, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[342, 170], [372, 231], [387, 291], [387, 356]]</point>"}]}
{"id": 1152, "image": ["rtx_frames_success_6/49_bridge#episode_15380/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15388/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[397, 308], [492, 313], [596, 346], [680, 406]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the tomato."}, {"from": "gpt", "value": "<point>[[434, 221], [436, 268], [425, 318], [383, 203]]</point>"}]}
{"id": 1154, "image": ["rtx_frames_success_6/49_bridge#episode_15394/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15398/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot on the plate."}, {"from": "gpt", "value": "<point>[[440, 211], [470, 348], [494, 576], [496, 757]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the hot sauce bottle."}, {"from": "gpt", "value": "<point>[[333, 328], [470, 301], [598, 306], [748, 338]]</point>"}]}
{"id": 1156, "image": ["rtx_frames_success_6/49_bridge#episode_15426/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15428/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[709, 281], [579, 316], [472, 393], [400, 486]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[147, 3], [242, 95], [333, 276], [378, 406]]</point>"}]}
{"id": 1158, "image": ["rtx_frames_success_6/49_bridge#episode_15448/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15456/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the sushi."}, {"from": "gpt", "value": "<point>[[502, 343], [483, 301], [414, 293], [372, 326]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[53, 424], [135, 416], [220, 393], [276, 333]]</point>"}]}
{"id": 1160, "image": ["rtx_frames_success_6/49_bridge#episode_15458/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15460/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[335, 135], [305, 288], [289, 409], [289, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the mushroom."}, {"from": "gpt", "value": "<point>[[276, 140], [400, 273], [481, 409], [549, 596]]</point>"}]}
{"id": 1162, "image": ["rtx_frames_success_6/49_bridge#episode_15466/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15466/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[218, 168], [359, 248], [536, 371], [650, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to slide the pan closer to the sushi roll."}, {"from": "gpt", "value": "<point>[[652, 361], [555, 318], [427, 301], [344, 308], [327, 241]]</point>"}]}
{"id": 1164, "image": ["rtx_frames_success_6/49_bridge#episode_15485/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15536/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[263, 236], [363, 273], [483, 361], [517, 414]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[532, 351], [483, 421], [432, 524], [406, 629]]</point>"}]}
{"id": 1166, "image": ["rtx_frames_success_6/49_bridge#episode_15536/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15547/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth towards the bottom right corner."}, {"from": "gpt", "value": "<point>[[393, 576], [489, 637], [586, 689], [684, 722], [729, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the hot sauce bottle."}, {"from": "gpt", "value": "<point>[[214, 291], [372, 346], [506, 439], [652, 576]]</point>"}]}
{"id": 1168, "image": ["rtx_frames_success_6/49_bridge#episode_15550/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15578/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cloth."}, {"from": "gpt", "value": "<point>[[357, 228], [462, 323], [558, 454], [630, 589]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[233, 311], [342, 333], [449, 446], [483, 561]]</point>"}]}
{"id": 1170, "image": ["rtx_frames_success_6/49_bridge#episode_15586/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15590/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan containing broccoli."}, {"from": "gpt", "value": "<point>[[280, 303], [380, 308], [492, 343], [568, 393]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the microwave handle."}, {"from": "gpt", "value": "<point>[[566, 5], [643, 65], [692, 140], [731, 286]]</point>"}]}
{"id": 1172, "image": ["rtx_frames_success_6/49_bridge#episode_15621/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15666/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[105, 243], [301, 383], [502, 556], [673, 747]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[310, 85], [382, 316], [419, 494], [440, 707]]</point>"}]}
{"id": 1174, "image": ["rtx_frames_success_6/49_bridge#episode_15677/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15681/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cup on the plate."}, {"from": "gpt", "value": "<point>[[286, 133], [348, 221], [421, 361], [472, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange and green item."}, {"from": "gpt", "value": "<point>[[297, 63], [374, 241], [447, 446], [492, 622]]</point>"}]}
{"id": 1176, "image": ["rtx_frames_success_6/49_bridge#episode_15686/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15693/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[241, 281], [289, 236], [338, 155], [370, 93]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spatula towards the orange cloth."}, {"from": "gpt", "value": "<point>[[459, 313], [444, 441], [447, 574], [457, 737]]</point>"}]}
{"id": 1178, "image": ["rtx_frames_success_6/49_bridge#episode_15725/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15736/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan containing broccoli."}, {"from": "gpt", "value": "<point>[[383, 246], [444, 444], [479, 622], [513, 832]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[549, 363], [462, 343], [385, 336], [291, 346]]</point>"}]}
{"id": 1180, "image": ["rtx_frames_success_6/49_bridge#episode_15736/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15746/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[297, 303], [353, 368], [393, 444], [421, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position above the pot."}, {"from": "gpt", "value": "<point>[[271, 165], [278, 251], [299, 326], [329, 391]]</point>"}]}
{"id": 1182, "image": ["rtx_frames_success_6/49_bridge#episode_15747/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15756/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[374, 456], [308, 469], [250, 459], [205, 431]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot in the pot."}, {"from": "gpt", "value": "<point>[[558, 236], [502, 338], [451, 486], [402, 654]]</point>"}]}
{"id": 1184, "image": ["rtx_frames_success_6/49_bridge#episode_15783/frame_15.png", "rtx_frames_success_6/49_bridge#episode_15797/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush over the middle left of the table."}, {"from": "gpt", "value": "<point>[[297, 211], [237, 286], [186, 381], [152, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[485, 261], [459, 401], [445, 509], [440, 622]]</point>"}]}
{"id": 1186, "image": ["rtx_frames_success_6/49_bridge#episode_15812/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15817/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[380, 221], [430, 281], [466, 361], [492, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the capsicum in the pan."}, {"from": "gpt", "value": "<point>[[506, 105], [609, 271], [699, 436], [756, 576]]</point>"}]}
{"id": 1188, "image": ["rtx_frames_success_6/49_bridge#episode_15821/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15880/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the banana."}, {"from": "gpt", "value": "<point>[[541, 211], [468, 263], [391, 351], [335, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[707, 411], [624, 411], [570, 398], [506, 371]]</point>"}]}
{"id": 1190, "image": ["rtx_frames_success_6/49_bridge#episode_15910/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15922/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[192, 243], [338, 331], [489, 436], [622, 566]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[271, 10], [355, 153], [445, 373], [530, 591]]</point>"}]}
{"id": 1192, "image": ["rtx_frames_success_6/49_bridge#episode_15926/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15926/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the mushroom."}, {"from": "gpt", "value": "<point>[[519, 271], [536, 341], [551, 421], [556, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom to the left of the spoon."}, {"from": "gpt", "value": "<point>[[588, 434], [513, 363], [440, 306], [336, 241]]</point>"}]}
{"id": 1194, "image": ["rtx_frames_success_6/49_bridge#episode_15935/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15935/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the bowl containing corn."}, {"from": "gpt", "value": "<point>[[628, 481], [491, 454], [344, 436], [195, 441]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to release the bowl."}, {"from": "gpt", "value": "<point>[[190, 404], [335, 396], [483, 416], [628, 461]]</point>"}]}
{"id": 1196, "image": ["rtx_frames_success_6/49_bridge#episode_15959/frame_0.png", "rtx_frames_success_6/49_bridge#episode_15968/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[175, 153], [250, 283], [310, 431], [363, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[605, 296], [523, 381], [442, 514], [387, 614]]</point>"}]}
{"id": 1198, "image": ["rtx_frames_success_6/49_bridge#episode_15968/frame_15.png", "rtx_frames_success_6/49_bridge#episode_16006/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife towards the pot."}, {"from": "gpt", "value": "<point>[[380, 496], [461, 576], [526, 682], [575, 805]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow rag."}, {"from": "gpt", "value": "<point>[[551, 376], [444, 398], [323, 434], [218, 471]]</point>"}]}
{"id": 1200, "image": ["rtx_frames_success_6/49_bridge#episode_16014/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16065/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position over the object in the pot."}, {"from": "gpt", "value": "<point>[[541, 311], [487, 313], [432, 301], [382, 271]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow cloth."}, {"from": "gpt", "value": "<point>[[523, 120], [536, 248], [534, 381], [523, 516]]</point>"}]}
{"id": 1202, "image": ["rtx_frames_success_6/49_bridge#episode_16080/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16082/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[564, 288], [459, 291], [355, 311], [263, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[282, 140], [218, 231], [165, 331], [128, 436]]</point>"}]}
{"id": 1204, "image": ["rtx_frames_success_6/49_bridge#episode_16101/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16101/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the pot."}, {"from": "gpt", "value": "<point>[[568, 496], [481, 464], [397, 446], [305, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot towards the spoon."}, {"from": "gpt", "value": "<point>[[284, 356], [415, 456], [508, 539], [622, 657]]</point>"}]}
{"id": 1206, "image": ["rtx_frames_success_6/49_bridge#episode_16103/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16110/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container with carrots."}, {"from": "gpt", "value": "<point>[[652, 431], [508, 501], [367, 586], [248, 674]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[391, 83], [398, 296], [436, 509], [481, 697]]</point>"}]}
{"id": 1208, "image": ["rtx_frames_success_6/49_bridge#episode_16158/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16158/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green spatula."}, {"from": "gpt", "value": "<point>[[184, 195], [342, 263], [519, 361], [645, 461]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spatula towards the sushi roll."}, {"from": "gpt", "value": "<point>[[521, 268], [400, 386], [280, 489], [164, 576]]</point>"}]}
{"id": 1210, "image": ["rtx_frames_success_6/49_bridge#episode_16180/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16180/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[179, 273], [291, 286], [376, 303], [472, 331]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot towards the edge of the table."}, {"from": "gpt", "value": "<point>[[502, 125], [609, 258], [684, 368], [767, 496]]</point>"}]}
{"id": 1212, "image": ["rtx_frames_success_6/49_bridge#episode_16181/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16203/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[442, 155], [372, 216], [327, 278], [263, 361]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drop the vegetable into the bowl."}, {"from": "gpt", "value": "<point>[[402, 188], [481, 251], [543, 328], [583, 414]]</point>"}]}
{"id": 1214, "image": ["rtx_frames_success_6/49_bridge#episode_16221/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16221/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[226, 216], [203, 283], [192, 368], [188, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the mushroom behind the pot."}, {"from": "gpt", "value": "<point>[[199, 371], [346, 378], [492, 396], [639, 436]]</point>"}]}
{"id": 1216, "image": ["rtx_frames_success_6/49_bridge#episode_16230/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16283/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[758, 83], [652, 145], [555, 216], [447, 303]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[258, 446], [209, 416], [150, 398], [88, 391]]</point>"}]}
{"id": 1218, "image": ["rtx_frames_success_6/49_bridge#episode_16283/frame_15.png", "rtx_frames_success_6/49_bridge#episode_16286/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pot over the broccoli."}, {"from": "gpt", "value": "<point>[[55, 351], [120, 398], [182, 449], [233, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[759, 130], [641, 293], [558, 439], [461, 639]]</point>"}]}
{"id": 1220, "image": ["rtx_frames_success_6/49_bridge#episode_16287/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16304/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the blueberries in the sink."}, {"from": "gpt", "value": "<point>[[372, 188], [459, 251], [530, 313], [603, 386]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the pot."}, {"from": "gpt", "value": "<point>[[492, 98], [470, 216], [464, 313], [466, 444]]</point>"}]}
{"id": 1222, "image": ["rtx_frames_success_6/49_bridge#episode_16330/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16345/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[489, 180], [380, 221], [273, 281], [158, 353]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[383, 5], [376, 140], [387, 296], [421, 456]]</point>"}]}
{"id": 1224, "image": ["rtx_frames_success_6/49_bridge#episode_16350/frame_15.png", "rtx_frames_success_6/49_bridge#episode_16352/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan towards the cans and cloth."}, {"from": "gpt", "value": "<point>[[506, 271], [596, 331], [673, 409], [754, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange towel."}, {"from": "gpt", "value": "<point>[[180, 100], [359, 206], [524, 321], [680, 439]]</point>"}]}
{"id": 1226, "image": ["rtx_frames_success_6/49_bridge#episode_16356/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16366/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[352, 3], [389, 128], [451, 258], [523, 378]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the salmon."}, {"from": "gpt", "value": "<point>[[111, 123], [327, 211], [521, 308], [709, 424]]</point>"}]}
{"id": 1228, "image": ["rtx_frames_success_6/49_bridge#episode_16367/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16370/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[259, 138], [237, 221], [218, 336], [212, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[509, 293], [400, 326], [295, 368], [182, 421]]</point>"}]}
{"id": 1230, "image": ["rtx_frames_success_6/49_bridge#episode_16388/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16407/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[479, 251], [440, 301], [378, 351], [316, 381]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the pot from the stove."}, {"from": "gpt", "value": "<point>[[553, 391], [617, 414], [695, 424], [758, 416]]</point>"}]}
{"id": 1232, "image": ["rtx_frames_success_6/49_bridge#episode_16407/frame_15.png", "rtx_frames_success_6/49_bridge#episode_16433/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the towel."}, {"from": "gpt", "value": "<point>[[436, 246], [404, 333], [363, 419], [310, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the brush."}, {"from": "gpt", "value": "<point>[[515, 135], [464, 195], [408, 246], [331, 291]]</point>"}]}
{"id": 1234, "image": ["rtx_frames_success_6/49_bridge#episode_16433/frame_15.png", "rtx_frames_success_6/49_bridge#episode_16438/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush to the right."}, {"from": "gpt", "value": "<point>[[395, 85], [508, 183], [626, 313], [712, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple cloth."}, {"from": "gpt", "value": "<point>[[645, 263], [643, 366], [637, 501], [622, 612]]</point>"}]}
{"id": 1236, "image": ["rtx_frames_success_6/49_bridge#episode_16454/frame_0.png", "rtx_frames_success_6/49_bridge#episode_16459/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the chicken leg."}, {"from": "gpt", "value": "<point>[[312, 328], [414, 353], [485, 381], [555, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the scissors."}, {"from": "gpt", "value": "<point>[[64, 135], [209, 251], [301, 348], [406, 486]]</point>"}]}
{"id": 1238, "image": ["rtx_frames_success_7/49_bridge#episode_171/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1659/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[206, 107], [320, 213], [489, 303], [695, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[320, 118], [379, 225], [434, 320], [463, 472]]</point>"}]}
{"id": 1240, "image": ["rtx_frames_success_7/49_bridge#episode_1661/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1663/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the curved blue tool."}, {"from": "gpt", "value": "<point>[[771, 275], [725, 433], [657, 584], [577, 725]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[565, 253], [438, 298], [350, 320], [215, 404]]</point>"}]}
{"id": 1242, "image": ["rtx_frames_success_7/49_bridge#episode_1702/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1709/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[426, 28], [426, 180], [426, 320], [417, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the potato."}, {"from": "gpt", "value": "<point>[[459, 96], [455, 253], [455, 365], [430, 528]]</point>"}]}
{"id": 1244, "image": ["rtx_frames_success_7/49_bridge#episode_1711/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1730/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[93, 73], [228, 264], [362, 404], [489, 528]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spatula."}, {"from": "gpt", "value": "<point>[[624, 320], [476, 337], [362, 326], [253, 326]]</point>"}]}
{"id": 1246, "image": ["rtx_frames_success_7/49_bridge#episode_1730/frame_15.png", "rtx_frames_success_7/49_bridge#episode_1732/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drop the spatula into the sink."}, {"from": "gpt", "value": "<point>[[240, 298], [295, 382], [346, 461], [417, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana in the colander."}, {"from": "gpt", "value": "<point>[[362, 410], [295, 388], [249, 360], [177, 315]]</point>"}]}
{"id": 1248, "image": ["rtx_frames_success_7/49_bridge#episode_1742/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1756/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the clothes."}, {"from": "gpt", "value": "<point>[[367, 404], [472, 528], [560, 624], [649, 753]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[312, 230], [413, 303], [522, 399], [632, 500]]</point>"}]}
{"id": 1250, "image": ["rtx_frames_success_7/49_bridge#episode_1766/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1766/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the plate."}, {"from": "gpt", "value": "<point>[[194, 152], [316, 191], [421, 208], [581, 236]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the plate under the cup."}, {"from": "gpt", "value": "<point>[[636, 253], [573, 309], [506, 331], [426, 360]]</point>"}]}
{"id": 1252, "image": ["rtx_frames_success_7/49_bridge#episode_1770/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1770/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the white block."}, {"from": "gpt", "value": "<point>[[362, 264], [362, 376], [303, 388], [228, 472]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the sushi roll to the right of the spatula."}, {"from": "gpt", "value": "<point>[[308, 326], [409, 365], [485, 410], [573, 410]]</point>"}]}
{"id": 1254, "image": ["rtx_frames_success_7/49_bridge#episode_1773/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1780/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[400, 62], [392, 140], [358, 197], [308, 247]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bowl."}, {"from": "gpt", "value": "<point>[[362, 140], [442, 258], [480, 354], [510, 494]]</point>"}]}
{"id": 1256, "image": ["rtx_frames_success_7/49_bridge#episode_1784/frame_0.png", "rtx_frames_success_7/49_bridge#episode_1792/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[442, 567], [396, 433], [350, 287], [295, 129]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red fork."}, {"from": "gpt", "value": "<point>[[303, 388], [232, 275], [202, 169], [126, 28]]</point>"}]}
{"id": 1258, "image": ["rtx_frames_success_7/49_bridge#episode_16539/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16540/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[451, 208], [548, 258], [636, 303], [775, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the measuring spoon towards the towel."}, {"from": "gpt", "value": "<point>[[729, 376], [691, 478], [666, 567], [666, 685]]</point>"}]}
{"id": 1260, "image": ["rtx_frames_success_7/49_bridge#episode_16541/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16550/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[421, 225], [371, 275], [295, 320], [223, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[265, 146], [371, 225], [489, 343], [624, 601]]</point>"}]}
{"id": 1262, "image": ["rtx_frames_success_7/49_bridge#episode_16556/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16558/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave door."}, {"from": "gpt", "value": "<point>[[640, 315], [535, 247], [430, 174], [265, 84]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[396, 208], [438, 331], [497, 494], [619, 730]]</point>"}]}
{"id": 1264, "image": ["rtx_frames_success_7/49_bridge#episode_16569/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16571/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[699, 213], [590, 287], [468, 371], [228, 528]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the pot with the lid on."}, {"from": "gpt", "value": "<point>[[253, 73], [278, 157], [287, 230], [287, 320]]</point>"}]}
{"id": 1266, "image": ["rtx_frames_success_7/49_bridge#episode_16582/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16634/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower over the towel."}, {"from": "gpt", "value": "<point>[[742, 124], [695, 275], [645, 444], [590, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red object in the pot."}, {"from": "gpt", "value": "<point>[[611, 39], [598, 174], [548, 348], [518, 545]]</point>"}]}
{"id": 1268, "image": ["rtx_frames_success_7/49_bridge#episode_16636/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16640/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[253, 281], [404, 382], [527, 421], [649, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot towards the spoon and mushroom."}, {"from": "gpt", "value": "<point>[[750, 292], [598, 360], [463, 433], [291, 522]]</point>"}]}
{"id": 1270, "image": ["rtx_frames_success_7/49_bridge#episode_16670/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16686/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange pot."}, {"from": "gpt", "value": "<point>[[105, 230], [303, 303], [480, 354], [674, 365]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green cloth."}, {"from": "gpt", "value": "<point>[[754, 39], [590, 152], [396, 320], [228, 478]]</point>"}]}
{"id": 1272, "image": ["rtx_frames_success_7/49_bridge#episode_16698/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16698/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the brush."}, {"from": "gpt", "value": "<point>[[472, 258], [421, 298], [371, 320], [303, 331]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the brush."}, {"from": "gpt", "value": "<point>[[438, 140], [438, 258], [480, 421], [493, 612]]</point>"}]}
{"id": 1274, "image": ["rtx_frames_success_7/49_bridge#episode_16729/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16730/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the strawberry."}, {"from": "gpt", "value": "<point>[[417, 303], [367, 343], [299, 360], [232, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[548, 230], [586, 399], [615, 590], [581, 843]]</point>"}]}
{"id": 1276, "image": ["rtx_frames_success_7/49_bridge#episode_16739/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16753/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[552, 410], [451, 449], [320, 438], [211, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon in the sink."}, {"from": "gpt", "value": "<point>[[463, 73], [472, 174], [472, 275], [468, 410]]</point>"}]}
{"id": 1278, "image": ["rtx_frames_success_7/49_bridge#episode_16757/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16762/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[202, 253], [303, 343], [421, 478], [518, 669]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the tomato in the pan."}, {"from": "gpt", "value": "<point>[[544, 230], [506, 337], [455, 438], [421, 556]]</point>"}]}
{"id": 1280, "image": ["rtx_frames_success_7/49_bridge#episode_16772/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16780/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[228, 258], [274, 326], [312, 388], [341, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to close the cabinet door."}, {"from": "gpt", "value": "<point>[[379, 494], [472, 494], [552, 478], [657, 466]]</point>"}]}
{"id": 1282, "image": ["rtx_frames_success_7/49_bridge#episode_16782/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16794/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green object."}, {"from": "gpt", "value": "<point>[[746, 107], [716, 202], [683, 287], [632, 449]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the almond container."}, {"from": "gpt", "value": "<point>[[362, 275], [438, 343], [506, 438], [577, 573]]</point>"}]}
{"id": 1284, "image": ["rtx_frames_success_7/49_bridge#episode_16824/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16825/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spatula."}, {"from": "gpt", "value": "<point>[[670, 230], [624, 298], [573, 354], [497, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[371, 96], [371, 236], [346, 343], [312, 478]]</point>"}]}
{"id": 1286, "image": ["rtx_frames_success_7/49_bridge#episode_16825/frame_15.png", "rtx_frames_success_7/49_bridge#episode_16850/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife towards the pot."}, {"from": "gpt", "value": "<point>[[303, 410], [392, 388], [447, 320], [514, 287]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[497, 51], [455, 163], [421, 258], [400, 365]]</point>"}]}
{"id": 1288, "image": ["rtx_frames_success_7/49_bridge#episode_16852/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16852/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[565, 500], [438, 506], [316, 511], [126, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move away from the stove top."}, {"from": "gpt", "value": "<point>[[169, 421], [312, 466], [438, 522], [569, 567]]</point>"}]}
{"id": 1290, "image": ["rtx_frames_success_7/49_bridge#episode_16854/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16860/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ice cream scoop."}, {"from": "gpt", "value": "<point>[[392, 84], [472, 202], [577, 331], [662, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[455, 270], [522, 320], [603, 365], [699, 444]]</point>"}]}
{"id": 1292, "image": ["rtx_frames_success_7/49_bridge#episode_16868/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16868/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[388, 287], [463, 354], [565, 466], [712, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bowl to the desired location."}, {"from": "gpt", "value": "<point>[[708, 545], [552, 494], [430, 410], [282, 326]]</point>"}]}
{"id": 1294, "image": ["rtx_frames_success_7/49_bridge#episode_16873/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16879/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[126, 67], [253, 191], [392, 298], [573, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[430, 163], [362, 230], [278, 303], [143, 466]]</point>"}]}
{"id": 1296, "image": ["rtx_frames_success_7/49_bridge#episode_16879/frame_15.png", "rtx_frames_success_7/49_bridge#episode_16897/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[160, 309], [278, 326], [388, 343], [493, 331]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[506, 236], [569, 303], [640, 343], [695, 421]]</point>"}]}
{"id": 1298, "image": ["rtx_frames_success_7/49_bridge#episode_16898/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16899/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[531, 118], [506, 270], [480, 388], [447, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the orange toy towards the vessel."}, {"from": "gpt", "value": "<point>[[518, 208], [531, 331], [573, 483], [619, 736]]</point>"}]}
{"id": 1300, "image": ["rtx_frames_success_7/49_bridge#episode_16905/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16907/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow towel."}, {"from": "gpt", "value": "<point>[[303, 281], [438, 298], [590, 337], [754, 393]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[341, 320], [434, 444], [581, 596], [771, 787]]</point>"}]}
{"id": 1302, "image": ["rtx_frames_success_7/49_bridge#episode_16908/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16908/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[215, 399], [287, 388], [341, 376], [400, 354]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cloth to the left of the spoon."}, {"from": "gpt", "value": "<point>[[388, 343], [379, 455], [337, 539], [282, 612]]</point>"}]}
{"id": 1304, "image": ["rtx_frames_success_7/49_bridge#episode_16919/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16929/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[581, 112], [506, 208], [447, 309], [375, 365]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cloth."}, {"from": "gpt", "value": "<point>[[371, 96], [320, 208], [274, 298], [219, 433]]</point>"}]}
{"id": 1306, "image": ["rtx_frames_success_7/49_bridge#episode_16939/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16956/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[147, 376], [228, 382], [282, 393], [337, 433]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[362, 84], [489, 197], [590, 309], [683, 494]]</point>"}]}
{"id": 1308, "image": ["rtx_frames_success_7/49_bridge#episode_16956/frame_15.png", "rtx_frames_success_7/49_bridge#episode_16981/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the center of the table."}, {"from": "gpt", "value": "<point>[[674, 315], [544, 320], [421, 365], [303, 433]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the food item in the container."}, {"from": "gpt", "value": "<point>[[438, 17], [539, 174], [628, 343], [695, 579]]</point>"}]}
{"id": 1310, "image": ["rtx_frames_success_7/49_bridge#episode_16992/frame_0.png", "rtx_frames_success_7/49_bridge#episode_16998/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green pepper."}, {"from": "gpt", "value": "<point>[[581, 163], [497, 287], [400, 337], [223, 427]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the measuring cup."}, {"from": "gpt", "value": "<point>[[472, 163], [451, 275], [413, 399], [354, 506]]</point>"}]}
{"id": 1312, "image": ["rtx_frames_success_7/49_bridge#episode_17005/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17007/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green object."}, {"from": "gpt", "value": "<point>[[388, 152], [535, 242], [615, 343], [721, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[261, 337], [388, 365], [480, 343], [573, 376]]</point>"}]}
{"id": 1314, "image": ["rtx_frames_success_7/49_bridge#episode_17010/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17013/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[244, 298], [375, 382], [480, 438], [607, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[126, 292], [236, 309], [350, 360], [438, 365]]</point>"}]}
{"id": 1316, "image": ["rtx_frames_success_7/49_bridge#episode_17018/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17032/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl containing the corn."}, {"from": "gpt", "value": "<point>[[362, 34], [362, 174], [367, 337], [362, 528]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the fork."}, {"from": "gpt", "value": "<point>[[598, 140], [573, 388], [573, 528], [565, 713]]</point>"}]}
{"id": 1318, "image": ["rtx_frames_success_7/49_bridge#episode_17036/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17042/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[169, 287], [295, 404], [421, 489], [518, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up a plate."}, {"from": "gpt", "value": "<point>[[371, 163], [421, 236], [514, 315], [611, 365]]</point>"}]}
{"id": 1320, "image": ["rtx_frames_success_7/49_bridge#episode_17049/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17053/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[514, 208], [527, 320], [565, 461], [721, 702]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[754, 140], [754, 281], [729, 388], [657, 478]]</point>"}]}
{"id": 1322, "image": ["rtx_frames_success_7/49_bridge#episode_17060/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17066/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi roll in the bowl."}, {"from": "gpt", "value": "<point>[[202, 275], [346, 331], [497, 404], [628, 478]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[223, 118], [303, 163], [388, 298], [426, 528]]</point>"}]}
{"id": 1324, "image": ["rtx_frames_success_7/49_bridge#episode_17081/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17110/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the fork."}, {"from": "gpt", "value": "<point>[[362, 135], [396, 253], [442, 410], [468, 556]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the stuffed animal."}, {"from": "gpt", "value": "<point>[[547, 35], [618, 224], [701, 429], [732, 647]]</point>"}]}
{"id": 1326, "image": ["rtx_frames_success_7/49_bridge#episode_17110/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17117/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the stuffed animal in the pot."}, {"from": "gpt", "value": "<point>[[732, 412], [649, 371], [574, 335], [543, 400]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the yellow pepper."}, {"from": "gpt", "value": "<point>[[202, 416], [371, 483], [501, 567], [666, 680]]</point>"}]}
{"id": 1328, "image": ["rtx_frames_success_7/49_bridge#episode_17119/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17122/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[324, 129], [316, 242], [312, 343], [278, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green cloth."}, {"from": "gpt", "value": "<point>[[152, 174], [308, 247], [493, 410], [539, 646]]</point>"}]}
{"id": 1330, "image": ["rtx_frames_success_7/49_bridge#episode_17124/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17126/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[371, 124], [312, 208], [240, 309], [164, 461]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[430, 118], [346, 213], [253, 298], [173, 354]]</point>"}]}
{"id": 1332, "image": ["rtx_frames_success_7/49_bridge#episode_17143/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17146/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spatula in the sink."}, {"from": "gpt", "value": "<point>[[687, 292], [662, 376], [628, 433], [535, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the pot."}, {"from": "gpt", "value": "<point>[[662, 73], [531, 202], [421, 287], [270, 410]]</point>"}]}
{"id": 1334, "image": ["rtx_frames_success_7/49_bridge#episode_17146/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17151/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the cheese."}, {"from": "gpt", "value": "<point>[[392, 51], [371, 174], [329, 247], [274, 393]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the eggplant."}, {"from": "gpt", "value": "<point>[[333, 174], [354, 292], [383, 410], [404, 590]]</point>"}]}
{"id": 1336, "image": ["rtx_frames_success_7/49_bridge#episode_17158/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17158/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[261, 112], [383, 213], [506, 298], [674, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl under the napkin."}, {"from": "gpt", "value": "<point>[[560, 421], [459, 472], [354, 500], [211, 517]]</point>"}]}
{"id": 1338, "image": ["rtx_frames_success_7/49_bridge#episode_17164/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17164/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[754, 315], [632, 365], [480, 404], [278, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cloth towards the pot."}, {"from": "gpt", "value": "<point>[[346, 371], [463, 399], [611, 416], [750, 433]]</point>"}]}
{"id": 1340, "image": ["rtx_frames_success_7/49_bridge#episode_17185/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17189/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the towel."}, {"from": "gpt", "value": "<point>[[337, 112], [430, 219], [506, 331], [611, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[101, 230], [244, 253], [350, 275], [472, 287]]</point>"}]}
{"id": 1342, "image": ["rtx_frames_success_7/49_bridge#episode_17196/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17210/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the item in the bowl."}, {"from": "gpt", "value": "<point>[[409, 96], [371, 230], [295, 309], [236, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[194, 253], [320, 360], [455, 522], [581, 725]]</point>"}]}
{"id": 1344, "image": ["rtx_frames_success_7/49_bridge#episode_17210/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17222/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the towel."}, {"from": "gpt", "value": "<point>[[379, 337], [329, 320], [291, 287], [240, 258]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[885, 124], [855, 208], [843, 281], [792, 382]]</point>"}]}
{"id": 1346, "image": ["rtx_frames_success_7/49_bridge#episode_17222/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17226/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pot on the stove."}, {"from": "gpt", "value": "<point>[[388, 51], [400, 219], [400, 337], [413, 511]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[565, 112], [539, 230], [480, 365], [417, 522]]</point>"}]}
{"id": 1348, "image": ["rtx_frames_success_7/49_bridge#episode_17240/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17243/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[354, 208], [303, 320], [236, 371], [181, 478]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[493, 17], [489, 174], [451, 298], [438, 416]]</point>"}]}
{"id": 1350, "image": ["rtx_frames_success_7/49_bridge#episode_17245/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17247/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the stuffed animal."}, {"from": "gpt", "value": "<point>[[455, 169], [573, 275], [670, 309], [729, 449]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the vessel."}, {"from": "gpt", "value": "<point>[[417, 6], [506, 140], [577, 253], [603, 382]]</point>"}]}
{"id": 1352, "image": ["rtx_frames_success_7/49_bridge#episode_17256/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17281/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pan."}, {"from": "gpt", "value": "<point>[[426, 17], [535, 140], [615, 247], [670, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[649, 500], [430, 416], [274, 309], [114, 242]]</point>"}]}
{"id": 1354, "image": ["rtx_frames_success_7/49_bridge#episode_17288/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17289/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[308, 410], [463, 433], [573, 461], [674, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[287, 107], [253, 213], [198, 275], [131, 421]]</point>"}]}
{"id": 1356, "image": ["rtx_frames_success_7/49_bridge#episode_17295/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17315/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the bottom left corner of the table."}, {"from": "gpt", "value": "<point>[[615, 67], [426, 185], [278, 264], [131, 337]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the potato in the pot."}, {"from": "gpt", "value": "<point>[[417, 247], [346, 315], [287, 410], [287, 489]]</point>"}]}
{"id": 1358, "image": ["rtx_frames_success_7/49_bridge#episode_17324/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17325/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the scoop."}, {"from": "gpt", "value": "<point>[[160, 62], [177, 197], [177, 343], [169, 522]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[413, 197], [522, 343], [590, 489], [645, 624]]</point>"}]}
{"id": 1360, "image": ["rtx_frames_success_7/49_bridge#episode_17334/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17340/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato in the pan."}, {"from": "gpt", "value": "<point>[[421, 56], [485, 140], [514, 208], [535, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the rag."}, {"from": "gpt", "value": "<point>[[506, 287], [611, 399], [699, 472], [775, 551]]</point>"}]}
{"id": 1362, "image": ["rtx_frames_success_7/49_bridge#episode_17366/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17367/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[308, 174], [413, 287], [455, 371], [497, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spoon on the table."}, {"from": "gpt", "value": "<point>[[388, 107], [447, 230], [480, 376], [463, 534]]</point>"}]}
{"id": 1364, "image": ["rtx_frames_success_7/49_bridge#episode_17375/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17375/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[497, 34], [527, 219], [514, 393], [514, 590]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to knock over the metal pot."}, {"from": "gpt", "value": "<point>[[518, 416], [400, 399], [354, 337], [219, 292]]</point>"}]}
{"id": 1366, "image": ["rtx_frames_success_7/49_bridge#episode_17383/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17383/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[261, 67], [379, 197], [497, 298], [649, 348]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the mushroom."}, {"from": "gpt", "value": "<point>[[493, 152], [404, 185], [346, 242], [333, 331]]</point>"}]}
{"id": 1368, "image": ["rtx_frames_success_7/49_bridge#episode_17387/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17391/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[417, 197], [539, 365], [598, 534], [611, 730]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the towel."}, {"from": "gpt", "value": "<point>[[447, 0], [447, 118], [447, 275], [413, 506]]</point>"}]}
{"id": 1370, "image": ["rtx_frames_success_7/49_bridge#episode_17393/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17398/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lime."}, {"from": "gpt", "value": "<point>[[80, 84], [316, 208], [518, 287], [674, 438]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[421, 444], [522, 478], [569, 494], [636, 528]]</point>"}]}
{"id": 1372, "image": ["rtx_frames_success_7/49_bridge#episode_17417/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17422/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[413, 320], [548, 303], [657, 309], [737, 275]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[261, 202], [337, 253], [430, 281], [539, 320]]</point>"}]}
{"id": 1374, "image": ["rtx_frames_success_7/49_bridge#episode_17428/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17453/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[404, 6], [451, 174], [560, 275], [581, 478]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[632, 208], [645, 337], [649, 449], [649, 590]]</point>"}]}
{"id": 1376, "image": ["rtx_frames_success_7/49_bridge#episode_17457/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17491/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pickle."}, {"from": "gpt", "value": "<point>[[489, 73], [489, 213], [472, 287], [421, 360]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[228, 157], [350, 298], [463, 472], [556, 843]]</point>"}]}
{"id": 1378, "image": ["rtx_frames_success_7/49_bridge#episode_17494/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17496/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the rag."}, {"from": "gpt", "value": "<point>[[493, 84], [548, 270], [598, 472], [640, 691]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pan containing broccoli."}, {"from": "gpt", "value": "<point>[[539, 185], [628, 292], [683, 416], [674, 590]]</point>"}]}
{"id": 1380, "image": ["rtx_frames_success_7/49_bridge#episode_17503/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17516/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the croissant towards the lower right corner near the spoon."}, {"from": "gpt", "value": "<point>[[421, 197], [531, 410], [573, 562], [657, 747]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position above the napkin."}, {"from": "gpt", "value": "<point>[[581, 281], [581, 404], [581, 522], [586, 663]]</point>"}]}
{"id": 1382, "image": ["rtx_frames_success_7/49_bridge#episode_17530/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17542/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[190, 62], [270, 118], [375, 213], [459, 320]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup on the plate."}, {"from": "gpt", "value": "<point>[[160, 101], [236, 163], [295, 242], [341, 320]]</point>"}]}
{"id": 1384, "image": ["rtx_frames_success_7/49_bridge#episode_17542/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17548/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the cup on the front left side of the stove."}, {"from": "gpt", "value": "<point>[[270, 157], [232, 225], [198, 309], [118, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[434, 320], [371, 275], [308, 264], [223, 281]]</point>"}]}
{"id": 1386, "image": ["rtx_frames_success_7/49_bridge#episode_17551/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17555/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[539, 124], [518, 242], [451, 444], [346, 713]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[413, 140], [497, 202], [590, 287], [691, 399]]</point>"}]}
{"id": 1388, "image": ["rtx_frames_success_7/49_bridge#episode_17559/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17562/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green cloth."}, {"from": "gpt", "value": "<point>[[308, 62], [409, 163], [455, 348], [590, 640]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[135, 309], [282, 360], [480, 399], [725, 410]]</point>"}]}
{"id": 1390, "image": ["rtx_frames_success_7/49_bridge#episode_17562/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17565/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the cloth."}, {"from": "gpt", "value": "<point>[[619, 399], [493, 444], [362, 466], [194, 478]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the chicken leg piece."}, {"from": "gpt", "value": "<point>[[489, 376], [615, 483], [733, 584], [860, 669]]</point>"}]}
{"id": 1392, "image": ["rtx_frames_success_7/49_bridge#episode_17569/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17572/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan containing the broccoli."}, {"from": "gpt", "value": "<point>[[552, 292], [556, 449], [565, 669], [539, 843]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[522, 264], [472, 365], [417, 455], [312, 579]]</point>"}]}
{"id": 1394, "image": ["rtx_frames_success_7/49_bridge#episode_17584/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17593/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[143, 230], [219, 242], [270, 258], [324, 331]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the towel towards the center of the front edge of the table."}, {"from": "gpt", "value": "<point>[[581, 270], [581, 421], [531, 562], [438, 691]]</point>"}]}
{"id": 1396, "image": ["rtx_frames_success_7/49_bridge#episode_17617/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17617/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[624, 39], [729, 140], [813, 213], [838, 298]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the cauliflower."}, {"from": "gpt", "value": "<point>[[851, 135], [851, 270], [826, 466], [792, 702]]</point>"}]}
{"id": 1398, "image": ["rtx_frames_success_7/49_bridge#episode_17620/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17634/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[303, 219], [476, 275], [577, 365], [683, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the chicken wing."}, {"from": "gpt", "value": "<point>[[413, 253], [346, 275], [257, 326], [177, 455]]</point>"}]}
{"id": 1400, "image": ["rtx_frames_success_7/49_bridge#episode_17642/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17645/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[573, 51], [577, 197], [556, 331], [518, 500]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[413, 275], [371, 275], [324, 275], [270, 287]]</point>"}]}
{"id": 1402, "image": ["rtx_frames_success_7/49_bridge#episode_17647/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17655/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the white vegetable in the pot."}, {"from": "gpt", "value": "<point>[[548, 163], [548, 225], [539, 309], [522, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[164, 90], [278, 275], [367, 433], [442, 669]]</point>"}]}
{"id": 1404, "image": ["rtx_frames_success_7/49_bridge#episode_17660/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17688/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[754, 112], [754, 253], [779, 348], [817, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[421, 213], [362, 320], [303, 376], [198, 449]]</point>"}]}
{"id": 1406, "image": ["rtx_frames_success_7/49_bridge#episode_17694/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17694/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift a small metal pot."}, {"from": "gpt", "value": "<point>[[430, 399], [341, 416], [244, 399], [122, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to release the pot."}, {"from": "gpt", "value": "<point>[[59, 382], [244, 433], [413, 438], [590, 371]]</point>"}]}
{"id": 1408, "image": ["rtx_frames_success_7/49_bridge#episode_17718/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17718/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the cauliflower."}, {"from": "gpt", "value": "<point>[[426, 275], [426, 365], [430, 433], [434, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the cauliflower in the pot."}, {"from": "gpt", "value": "<point>[[472, 79], [548, 107], [640, 202], [767, 275]]</point>"}]}
{"id": 1410, "image": ["rtx_frames_success_7/49_bridge#episode_17733/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17733/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[573, 309], [489, 343], [379, 331], [236, 320]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[261, 354], [371, 393], [455, 438], [539, 483]]</point>"}]}
{"id": 1412, "image": ["rtx_frames_success_7/49_bridge#episode_17736/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17743/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower the gripper towards the blue pad."}, {"from": "gpt", "value": "<point>[[598, 242], [514, 309], [413, 343], [282, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[569, 843], [628, 781], [640, 680], [649, 562]]</point>"}]}
{"id": 1414, "image": ["rtx_frames_success_7/49_bridge#episode_17745/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17745/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cleaning brush."}, {"from": "gpt", "value": "<point>[[649, 6], [649, 51], [556, 163], [489, 230], [350, 354]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cleaning brush to the left side of the countertop."}, {"from": "gpt", "value": "<point>[[295, 28], [337, 309], [350, 444], [371, 618]]</point>"}]}
{"id": 1416, "image": ["rtx_frames_success_7/49_bridge#episode_17753/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17755/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[581, 331], [611, 522], [628, 618], [628, 764]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green bell pepper."}, {"from": "gpt", "value": "<point>[[556, 163], [518, 242], [468, 264], [421, 320]]</point>"}]}
{"id": 1418, "image": ["rtx_frames_success_7/49_bridge#episode_17777/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17783/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[581, 669], [463, 438], [388, 242], [337, 56]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move above a cloth."}, {"from": "gpt", "value": "<point>[[687, 376], [704, 522], [674, 674], [674, 753]]</point>"}]}
{"id": 1420, "image": ["rtx_frames_success_7/49_bridge#episode_17784/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17794/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[742, 444], [539, 371], [329, 365], [135, 433]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[615, 118], [581, 219], [539, 309], [480, 416]]</point>"}]}
{"id": 1422, "image": ["rtx_frames_success_7/49_bridge#episode_17795/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17797/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green object."}, {"from": "gpt", "value": "<point>[[750, 584], [792, 444], [826, 287], [872, 73]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[270, 371], [396, 371], [539, 371], [657, 315]]</point>"}]}
{"id": 1424, "image": ["rtx_frames_success_7/49_bridge#episode_17797/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17809/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth towards the pan."}, {"from": "gpt", "value": "<point>[[358, 275], [489, 320], [611, 343], [725, 354]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[678, 410], [577, 287], [463, 174], [324, 39]]</point>"}]}
{"id": 1426, "image": ["rtx_frames_success_7/49_bridge#episode_17815/frame_15.png", "rtx_frames_success_7/49_bridge#episode_17840/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the blue spoon to the right of the red spoon."}, {"from": "gpt", "value": "<point>[[346, 197], [489, 292], [662, 365], [801, 500]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the can of mushrooms on the front right of the table."}, {"from": "gpt", "value": "<point>[[691, 725], [649, 539], [581, 388], [514, 225]]</point>"}]}
{"id": 1428, "image": ["rtx_frames_success_7/49_bridge#episode_17845/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17852/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[447, 433], [295, 388], [194, 320], [84, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the sushi towards the plate."}, {"from": "gpt", "value": "<point>[[628, 551], [573, 444], [506, 281], [472, 112]]</point>"}]}
{"id": 1430, "image": ["rtx_frames_success_7/49_bridge#episode_17868/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17868/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cheese."}, {"from": "gpt", "value": "<point>[[615, 579], [522, 410], [426, 253], [337, 112]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cheese towards the pan."}, {"from": "gpt", "value": "<point>[[615, 399], [514, 365], [404, 348], [337, 354]]</point>"}]}
{"id": 1432, "image": ["rtx_frames_success_7/49_bridge#episode_17869/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17876/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[463, 820], [497, 624], [514, 404], [480, 135]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place a carrot in the pot."}, {"from": "gpt", "value": "<point>[[615, 489], [531, 388], [413, 287], [312, 185]]</point>"}]}
{"id": 1434, "image": ["rtx_frames_success_7/49_bridge#episode_17877/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17888/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[59, 545], [215, 410], [316, 287], [383, 51]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[362, 404], [480, 399], [607, 388], [716, 365]]</point>"}]}
{"id": 1436, "image": ["rtx_frames_success_7/49_bridge#episode_17894/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17897/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the capsicum."}, {"from": "gpt", "value": "<point>[[691, 691], [624, 511], [565, 360], [489, 146]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move pot away from the jar."}, {"from": "gpt", "value": "<point>[[497, 433], [421, 376], [350, 298], [228, 219]]</point>"}]}
{"id": 1438, "image": ["rtx_frames_success_7/49_bridge#episode_17909/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17929/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the blue cloth."}, {"from": "gpt", "value": "<point>[[640, 646], [594, 489], [565, 360], [463, 169]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pan containing broccoli."}, {"from": "gpt", "value": "<point>[[544, 444], [413, 416], [282, 360], [135, 264]]</point>"}]}
{"id": 1440, "image": ["rtx_frames_success_7/49_bridge#episode_17931/frame_0.png", "rtx_frames_success_7/49_bridge#episode_17934/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[746, 258], [666, 174], [603, 96], [522, 22]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull out the red cloth from the machine."}, {"from": "gpt", "value": "<point>[[312, 579], [421, 511], [497, 360], [565, 180]]</point>"}]}
{"id": 1442, "image": ["rtx_frames_success_7/49_bridge#episode_17937/frame_0.png", "rtx_frames_success_8/49_bridge#episode_182/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[282, 427], [404, 410], [501, 354], [577, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the mushroom."}, {"from": "gpt", "value": "<point>[[468, 225], [368, 151], [288, 196]]</point>"}]}
{"id": 1444, "image": ["rtx_frames_success_8/49_bridge#episode_182/frame_15.png", "rtx_frames_success_8/49_bridge#episode_1812/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mushroom towards the cans."}, {"from": "gpt", "value": "<point>[[324, 70], [615, 251], [726, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[388, 299], [524, 288], [593, 380]]</point>"}]}
{"id": 1446, "image": ["rtx_frames_success_8/49_bridge#episode_1818/frame_0.png", "rtx_frames_success_8/49_bridge#episode_1838/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[329, 162], [579, 611], [592, 705]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[446, 214], [543, 443], [501, 554]]</point>"}]}
{"id": 1448, "image": ["rtx_frames_success_8/49_bridge#episode_1849/frame_0.png", "rtx_frames_success_8/49_bridge#episode_1849/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the laundry machine."}, {"from": "gpt", "value": "<point>[[78, 425], [213, 181], [349, 122]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach inside the laundry machine."}, {"from": "gpt", "value": "<point>[[341, 103], [183, 340], [127, 643]]</point>"}]}
{"id": 1450, "image": ["rtx_frames_success_8/49_bridge#episode_1858/frame_0.png", "rtx_frames_success_8/49_bridge#episode_1859/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[476, 41], [460, 451], [557, 528]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[562, 565], [482, 244], [407, 207]]</point>"}]}
{"id": 1452, "image": ["rtx_frames_success_8/49_bridge#episode_1859/frame_15.png", "rtx_frames_success_8/49_bridge#episode_1860/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bowl next to the canned good."}, {"from": "gpt", "value": "<point>[[341, 199], [194, 380], [144, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red capsicum."}, {"from": "gpt", "value": "<point>[[410, 336], [191, 392], [150, 499]]</point>"}]}
{"id": 1454, "image": ["rtx_frames_success_8/49_bridge#episode_1864/frame_0.png", "rtx_frames_success_8/49_bridge#episode_1913/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[565, 85], [706, 702], [737, 801]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave door."}, {"from": "gpt", "value": "<point>[[474, 698], [338, 417], [208, 332]]</point>"}]}
{"id": 1456, "image": ["rtx_frames_success_8/49_bridge#episode_1913/frame_15.png", "rtx_frames_success_8/49_bridge#episode_17955/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the door to close it."}, {"from": "gpt", "value": "<point>[[258, 347], [510, 491], [587, 292]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[327, 377], [524, 395], [612, 440]]</point>"}]}
{"id": 1458, "image": ["rtx_frames_success_8/49_bridge#episode_17963/frame_0.png", "rtx_frames_success_8/49_bridge#episode_17995/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[338, 196], [230, 288], [205, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[645, 151], [510, 595], [554, 735]]</point>"}]}
{"id": 1460, "image": ["rtx_frames_success_8/49_bridge#episode_18032/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18036/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the rag."}, {"from": "gpt", "value": "<point>[[388, 229], [349, 417], [421, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[529, 240], [476, 277], [435, 314]]</point>"}]}
{"id": 1462, "image": ["rtx_frames_success_8/49_bridge#episode_18039/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18048/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the brush."}, {"from": "gpt", "value": "<point>[[515, 240], [576, 325], [659, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[499, 199], [302, 332], [202, 447]]</point>"}]}
{"id": 1464, "image": ["rtx_frames_success_8/49_bridge#episode_18050/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18050/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the yellow pepper."}, {"from": "gpt", "value": "<point>[[402, 163], [288, 222], [238, 303]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the yellow pepper towards the bowl."}, {"from": "gpt", "value": "<point>[[263, 299], [402, 270], [515, 351]]</point>"}]}
{"id": 1466, "image": ["rtx_frames_success_8/49_bridge#episode_18051/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18057/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[565, 270], [252, 358], [152, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the lid."}, {"from": "gpt", "value": "<point>[[668, 277], [496, 495], [488, 591]]</point>"}]}
{"id": 1468, "image": ["rtx_frames_success_8/49_bridge#episode_18076/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18078/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the mushroom."}, {"from": "gpt", "value": "<point>[[371, 465], [454, 351], [537, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[651, 288], [532, 318], [499, 410]]</point>"}]}
{"id": 1470, "image": ["rtx_frames_success_8/49_bridge#episode_18080/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18081/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spatula in the sink."}, {"from": "gpt", "value": "<point>[[532, 355], [679, 513], [695, 620]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[285, 214], [601, 550], [684, 687]]</point>"}]}
{"id": 1472, "image": ["rtx_frames_success_8/49_bridge#episode_18083/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18103/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[512, 521], [335, 469], [310, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup on the plate."}, {"from": "gpt", "value": "<point>[[341, 170], [568, 572], [684, 539]]</point>"}]}
{"id": 1474, "image": ["rtx_frames_success_8/49_bridge#episode_18103/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18110/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cup on the plate."}, {"from": "gpt", "value": "<point>[[742, 462], [468, 355], [368, 380]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[260, 15], [607, 617], [698, 805]]</point>"}]}
{"id": 1476, "image": ["rtx_frames_success_8/49_bridge#episode_18120/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18124/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot on the plate."}, {"from": "gpt", "value": "<point>[[360, 299], [291, 399], [294, 524]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice bottle."}, {"from": "gpt", "value": "<point>[[512, 273], [382, 347], [296, 421]]</point>"}]}
{"id": 1478, "image": ["rtx_frames_success_8/49_bridge#episode_18125/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18127/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green towel."}, {"from": "gpt", "value": "<point>[[352, 129], [643, 480], [720, 705]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[499, 288], [670, 358], [745, 510]]</point>"}]}
{"id": 1480, "image": ["rtx_frames_success_8/49_bridge#episode_18129/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18155/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[587, 355], [460, 432], [416, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[596, 52], [355, 447], [346, 517]]</point>"}]}
{"id": 1482, "image": ["rtx_frames_success_8/49_bridge#episode_18165/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18165/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the potato."}, {"from": "gpt", "value": "<point>[[518, 192], [230, 443], [158, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the potato."}, {"from": "gpt", "value": "<point>[[197, 561], [310, 521], [404, 613]]</point>"}]}
{"id": 1484, "image": ["rtx_frames_success_8/49_bridge#episode_18166/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18168/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the cheese wedge."}, {"from": "gpt", "value": "<point>[[493, 266], [618, 288], [687, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice container."}, {"from": "gpt", "value": "<point>[[313, 476], [543, 395], [643, 473]]</point>"}]}
{"id": 1486, "image": ["rtx_frames_success_8/49_bridge#episode_18169/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18169/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fish."}, {"from": "gpt", "value": "<point>[[283, 107], [116, 366], [105, 495]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the fish towards the blue cloth."}, {"from": "gpt", "value": "<point>[[288, 177], [515, 351], [560, 458]]</point>"}]}
{"id": 1488, "image": ["rtx_frames_success_8/49_bridge#episode_18176/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18189/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place a carrot to the right of the pot."}, {"from": "gpt", "value": "<point>[[457, 225], [573, 325], [579, 403]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[596, 262], [543, 524], [507, 654]]</point>"}]}
{"id": 1490, "image": ["rtx_frames_success_8/49_bridge#episode_18195/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18200/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[432, 251], [307, 432], [327, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[499, 196], [593, 395], [615, 491]]</point>"}]}
{"id": 1492, "image": ["rtx_frames_success_8/49_bridge#episode_18212/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18241/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[717, 421], [623, 406], [573, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple rag."}, {"from": "gpt", "value": "<point>[[476, 222], [579, 591], [526, 691]]</point>"}]}
{"id": 1494, "image": ["rtx_frames_success_8/49_bridge#episode_18250/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18253/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the vegetable."}, {"from": "gpt", "value": "<point>[[321, 207], [238, 229], [205, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[776, 410], [382, 432], [271, 558]]</point>"}]}
{"id": 1496, "image": ["rtx_frames_success_8/49_bridge#episode_18254/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18254/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[452, 114], [601, 369], [662, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the corn towards the purple towel."}, {"from": "gpt", "value": "<point>[[460, 262], [307, 329], [247, 465]]</point>"}]}
{"id": 1498, "image": ["rtx_frames_success_8/49_bridge#episode_18292/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18296/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the sushi piece in the pan."}, {"from": "gpt", "value": "<point>[[507, 554], [424, 536], [416, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[468, 299], [377, 251], [285, 318]]</point>"}]}
{"id": 1500, "image": ["rtx_frames_success_8/49_bridge#episode_18296/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18309/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[307, 321], [230, 524], [283, 606]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[479, 126], [643, 536], [618, 669]]</point>"}]}
{"id": 1502, "image": ["rtx_frames_success_8/49_bridge#episode_18309/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18314/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to cut the sushi roll."}, {"from": "gpt", "value": "<point>[[565, 469], [382, 421], [335, 325]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[341, 199], [446, 251], [496, 388]]</point>"}]}
{"id": 1504, "image": ["rtx_frames_success_8/49_bridge#episode_18318/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18320/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pepper."}, {"from": "gpt", "value": "<point>[[546, 561], [717, 395], [704, 270]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the egg."}, {"from": "gpt", "value": "<point>[[479, 163], [429, 447], [313, 561], [338, 362]]</point>"}]}
{"id": 1506, "image": ["rtx_frames_success_8/49_bridge#episode_18320/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18332/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the egg towards the bowl."}, {"from": "gpt", "value": "<point>[[393, 314], [598, 410], [657, 550]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[399, 181], [526, 591], [629, 639]]</point>"}]}
{"id": 1508, "image": ["rtx_frames_success_8/49_bridge#episode_18337/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18337/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green pepper."}, {"from": "gpt", "value": "<point>[[380, 63], [432, 129], [404, 203]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the green pepper in the pot."}, {"from": "gpt", "value": "<point>[[352, 78], [216, 229], [241, 410]]</point>"}]}
{"id": 1510, "image": ["rtx_frames_success_8/49_bridge#episode_18341/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18348/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the colander."}, {"from": "gpt", "value": "<point>[[565, 288], [402, 281], [341, 299]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[526, 48], [499, 480], [438, 635]]</point>"}]}
{"id": 1512, "image": ["rtx_frames_success_8/49_bridge#episode_18350/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18353/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange towel."}, {"from": "gpt", "value": "<point>[[524, 188], [584, 366], [562, 513]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[601, 410], [463, 421], [438, 303]]</point>"}]}
{"id": 1514, "image": ["rtx_frames_success_8/49_bridge#episode_18356/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18363/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[482, 292], [554, 502], [684, 539]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sauce pan."}, {"from": "gpt", "value": "<point>[[546, 421], [432, 458], [388, 502]]</point>"}]}
{"id": 1516, "image": ["rtx_frames_success_8/49_bridge#episode_18373/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18373/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[654, 192], [454, 266], [330, 373]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pan towards the towel."}, {"from": "gpt", "value": "<point>[[391, 303], [693, 654], [765, 827]]</point>"}]}
{"id": 1518, "image": ["rtx_frames_success_8/49_bridge#episode_18387/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18392/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[130, 103], [504, 355], [612, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the onion."}, {"from": "gpt", "value": "<point>[[100, 543], [324, 465], [355, 362]]</point>"}]}
{"id": 1520, "image": ["rtx_frames_success_8/49_bridge#episode_18392/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18394/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the onion towards the cardboard fence."}, {"from": "gpt", "value": "<point>[[341, 392], [247, 421], [155, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[501, 188], [485, 414], [380, 554]]</point>"}]}
{"id": 1522, "image": ["rtx_frames_success_8/49_bridge#episode_18400/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18409/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the chicken leg."}, {"from": "gpt", "value": "<point>[[191, 48], [482, 554], [645, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[321, 266], [288, 465], [399, 558]]</point>"}]}
{"id": 1524, "image": ["rtx_frames_success_8/49_bridge#episode_18430/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18438/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the brush."}, {"from": "gpt", "value": "<point>[[551, 244], [238, 369], [127, 473]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[541, 180], [458, 203], [391, 220], [334, 293]]</point>"}]}
{"id": 1526, "image": ["rtx_frames_success_8/49_bridge#episode_18446/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18446/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the toy."}, {"from": "gpt", "value": "<point>[[167, 259], [404, 403], [587, 642], [678, 868]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the toy towards the towel."}, {"from": "gpt", "value": "<point>[[551, 592], [389, 473], [277, 434], [116, 468]]</point>"}]}
{"id": 1528, "image": ["rtx_frames_success_8/49_bridge#episode_18449/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18449/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[518, 33], [657, 240], [709, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[626, 388], [416, 321], [299, 406]]</point>"}]}
{"id": 1530, "image": ["rtx_frames_success_8/49_bridge#episode_18455/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18461/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[482, 211], [416, 233], [449, 325]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange towel."}, {"from": "gpt", "value": "<point>[[274, 318], [518, 399], [670, 528]]</point>"}]}
{"id": 1532, "image": ["rtx_frames_success_8/49_bridge#episode_18462/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18469/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush towards the pot and can."}, {"from": "gpt", "value": "<point>[[302, 270], [435, 332], [546, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[380, 140], [568, 476], [560, 650]]</point>"}]}
{"id": 1534, "image": ["rtx_frames_success_8/49_bridge#episode_18473/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18473/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn in the pot."}, {"from": "gpt", "value": "<point>[[413, 299], [294, 325], [211, 443]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the corn on the purple towel."}, {"from": "gpt", "value": "<point>[[319, 295], [535, 295], [740, 392]]</point>"}]}
{"id": 1536, "image": ["rtx_frames_success_8/49_bridge#episode_18478/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18479/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange object in the bowl."}, {"from": "gpt", "value": "<point>[[391, 259], [321, 392], [219, 388]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[562, 137], [452, 284], [463, 395]]</point>"}]}
{"id": 1538, "image": ["rtx_frames_success_8/49_bridge#episode_18480/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18507/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sauce pan."}, {"from": "gpt", "value": "<point>[[512, 351], [596, 528], [681, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[404, 181], [396, 325], [507, 432]]</point>"}]}
{"id": 1540, "image": ["rtx_frames_success_8/49_bridge#episode_18508/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18518/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue utensil."}, {"from": "gpt", "value": "<point>[[438, 358], [582, 602], [604, 739]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container with the cheese."}, {"from": "gpt", "value": "<point>[[310, 211], [582, 440], [690, 691]]</point>"}]}
{"id": 1542, "image": ["rtx_frames_success_8/49_bridge#episode_18543/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18551/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[224, 89], [427, 606], [537, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the laundry machine."}, {"from": "gpt", "value": "<point>[[191, 417], [418, 270], [576, 229]]</point>"}]}
{"id": 1544, "image": ["rtx_frames_success_8/49_bridge#episode_18551/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18557/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the clothing out of the machine."}, {"from": "gpt", "value": "<point>[[535, 244], [363, 565], [402, 731]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pan."}, {"from": "gpt", "value": "<point>[[316, 410], [391, 343], [432, 292]]</point>"}]}
{"id": 1546, "image": ["rtx_frames_success_8/49_bridge#episode_18557/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18563/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan towards the cloth."}, {"from": "gpt", "value": "<point>[[424, 292], [590, 524], [579, 669]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[801, 454], [643, 499], [612, 425]]</point>"}]}
{"id": 1548, "image": ["rtx_frames_success_8/49_bridge#episode_18566/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18567/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[402, 236], [269, 281], [205, 380]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[643, 85], [319, 303], [266, 403]]</point>"}]}
{"id": 1550, "image": ["rtx_frames_success_8/49_bridge#episode_18570/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18578/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[382, 347], [529, 443], [568, 547]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the capsicum."}, {"from": "gpt", "value": "<point>[[601, 550], [438, 480], [327, 517]]</point>"}]}
{"id": 1552, "image": ["rtx_frames_success_8/49_bridge#episode_18581/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18589/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the spoon."}, {"from": "gpt", "value": "<point>[[676, 473], [540, 543], [501, 495]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[324, 380], [260, 480], [219, 451]]</point>"}]}
{"id": 1554, "image": ["rtx_frames_success_8/49_bridge#episode_18589/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18594/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon in front of the microwave."}, {"from": "gpt", "value": "<point>[[244, 440], [363, 299], [357, 207], [316, 203], [330, 247]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[529, 465], [679, 414], [731, 506]]</point>"}]}
{"id": 1556, "image": ["rtx_frames_success_8/49_bridge#episode_18598/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18603/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot on the plate."}, {"from": "gpt", "value": "<point>[[463, 233], [687, 484], [773, 606]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the food item."}, {"from": "gpt", "value": "<point>[[468, 170], [363, 244], [313, 325]]</point>"}]}
{"id": 1558, "image": ["rtx_frames_success_8/49_bridge#episode_18603/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18609/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the food item towards the cloth."}, {"from": "gpt", "value": "<point>[[313, 321], [355, 510], [432, 628], [465, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[371, 270], [543, 340], [618, 458]]</point>"}]}
{"id": 1560, "image": ["rtx_frames_success_8/49_bridge#episode_18617/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18622/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[294, 329], [363, 318], [429, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drop the pear into the pot."}, {"from": "gpt", "value": "<point>[[471, 277], [501, 635], [626, 709]]</point>"}]}
{"id": 1562, "image": ["rtx_frames_success_8/49_bridge#episode_18625/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18635/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the vessel."}, {"from": "gpt", "value": "<point>[[499, 262], [496, 362], [568, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[504, 111], [529, 314], [476, 458]]</point>"}]}
{"id": 1564, "image": ["rtx_frames_success_8/49_bridge#episode_18652/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18662/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange towel."}, {"from": "gpt", "value": "<point>[[587, 199], [681, 513], [643, 728]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow spoon."}, {"from": "gpt", "value": "<point>[[252, 222], [449, 295], [540, 417]]</point>"}]}
{"id": 1566, "image": ["rtx_frames_success_8/49_bridge#episode_18667/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18668/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[399, 288], [551, 325], [670, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana in the colander."}, {"from": "gpt", "value": "<point>[[701, 177], [748, 687], [601, 750]]</point>"}]}
{"id": 1568, "image": ["rtx_frames_success_8/49_bridge#episode_18682/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18685/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the corn."}, {"from": "gpt", "value": "<point>[[435, 225], [615, 255], [687, 414]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the cloth."}, {"from": "gpt", "value": "<point>[[274, 295], [463, 465], [518, 687]]</point>"}]}
{"id": 1570, "image": ["rtx_frames_success_8/49_bridge#episode_18695/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18697/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[496, 358], [587, 499], [557, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[393, 170], [537, 572], [554, 702]]</point>"}]}
{"id": 1572, "image": ["rtx_frames_success_8/49_bridge#episode_18716/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18716/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the object."}, {"from": "gpt", "value": "<point>[[576, 196], [446, 314], [368, 288]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[341, 199], [618, 576], [609, 709]]</point>"}]}
{"id": 1574, "image": ["rtx_frames_success_8/49_bridge#episode_18723/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18736/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[399, 310], [620, 395], [615, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[449, 92], [310, 114], [247, 273], [316, 262]]</point>"}]}
{"id": 1576, "image": ["rtx_frames_success_8/49_bridge#episode_18749/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18749/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[269, 384], [493, 410], [684, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the fork towards the dish towel."}, {"from": "gpt", "value": "<point>[[632, 343], [277, 451], [213, 547]]</point>"}]}
{"id": 1578, "image": ["rtx_frames_success_8/49_bridge#episode_18762/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18764/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[438, 78], [504, 233], [435, 310]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower the mushroom into the bowl."}, {"from": "gpt", "value": "<point>[[316, 244], [632, 473], [681, 624]]</point>"}]}
{"id": 1580, "image": ["rtx_frames_success_8/49_bridge#episode_18769/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18777/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the spice container."}, {"from": "gpt", "value": "<point>[[296, 347], [576, 318], [668, 388]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the cheese."}, {"from": "gpt", "value": "<point>[[573, 443], [432, 480], [388, 602]]</point>"}]}
{"id": 1582, "image": ["rtx_frames_success_8/49_bridge#episode_18777/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18790/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cheese to the upper right corner of the table."}, {"from": "gpt", "value": "<point>[[557, 539], [609, 661], [695, 639], [767, 513], [665, 595]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the avocado."}, {"from": "gpt", "value": "<point>[[319, 321], [457, 476], [518, 432]]</point>"}]}
{"id": 1584, "image": ["rtx_frames_success_8/49_bridge#episode_18790/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18793/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the avocado towards the pot."}, {"from": "gpt", "value": "<point>[[454, 310], [332, 425], [269, 369]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the oven door handle."}, {"from": "gpt", "value": "<point>[[416, 476], [598, 425], [748, 513]]</point>"}]}
{"id": 1586, "image": ["rtx_frames_success_8/49_bridge#episode_18800/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18819/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green pepper."}, {"from": "gpt", "value": "<point>[[410, 458], [607, 325], [742, 380]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up a piece of clothing from the basket."}, {"from": "gpt", "value": "<point>[[537, 569], [435, 713], [335, 728]]</point>"}]}
{"id": 1588, "image": ["rtx_frames_success_8/49_bridge#episode_18819/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18821/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the clothing inside the washing machine."}, {"from": "gpt", "value": "<point>[[324, 731], [449, 739], [540, 502], [548, 351], [482, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[693, 307], [562, 436], [479, 410]]</point>"}]}
{"id": 1590, "image": ["rtx_frames_success_8/49_bridge#episode_18828/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18841/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the towel towards the vessel."}, {"from": "gpt", "value": "<point>[[355, 336], [460, 469], [440, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the item in the container."}, {"from": "gpt", "value": "<point>[[446, 247], [493, 436], [393, 521]]</point>"}]}
{"id": 1592, "image": ["rtx_frames_success_8/49_bridge#episode_18841/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18843/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the item towards the edge of the surface."}, {"from": "gpt", "value": "<point>[[255, 233], [161, 314], [202, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the orange pot."}, {"from": "gpt", "value": "<point>[[452, 151], [598, 539], [687, 536]]</point>"}]}
{"id": 1594, "image": ["rtx_frames_success_8/49_bridge#episode_18843/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18850/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot to the left."}, {"from": "gpt", "value": "<point>[[440, 358], [271, 395], [199, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[659, 214], [501, 244], [377, 332]]</point>"}]}
{"id": 1596, "image": ["rtx_frames_success_8/49_bridge#episode_18850/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18856/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the towel."}, {"from": "gpt", "value": "<point>[[432, 273], [626, 521], [609, 724]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[416, 284], [551, 451], [579, 598]]</point>"}]}
{"id": 1598, "image": ["rtx_frames_success_8/49_bridge#episode_18875/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18878/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[551, 613], [452, 624], [440, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[413, 620], [535, 598], [551, 473]]</point>"}]}
{"id": 1600, "image": ["rtx_frames_success_8/49_bridge#episode_18878/frame_15.png", "rtx_frames_success_8/49_bridge#episode_18883/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[529, 495], [402, 517], [366, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the pan containing a mushroom."}, {"from": "gpt", "value": "<point>[[432, 609], [573, 484], [596, 355]]</point>"}]}
{"id": 1602, "image": ["rtx_frames_success_8/49_bridge#episode_18890/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18893/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[418, 355], [360, 591], [263, 628]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[640, 81], [571, 484], [690, 635]]</point>"}]}
{"id": 1604, "image": ["rtx_frames_success_8/49_bridge#episode_18904/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18907/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[410, 635], [535, 484], [618, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[515, 281], [449, 395], [368, 406]]</point>"}]}
{"id": 1606, "image": ["rtx_frames_success_8/49_bridge#episode_18921/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18921/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[404, 414], [524, 392], [596, 458]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[576, 465], [440, 392], [360, 403]]</point>"}]}
{"id": 1608, "image": ["rtx_frames_success_8/49_bridge#episode_18923/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18937/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[693, 236], [712, 380], [604, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the basil bottle."}, {"from": "gpt", "value": "<point>[[609, 421], [427, 336], [349, 417]]</point>"}]}
{"id": 1610, "image": ["rtx_frames_success_8/49_bridge#episode_18944/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18952/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[432, 303], [515, 436], [537, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow cloth."}, {"from": "gpt", "value": "<point>[[280, 177], [449, 440], [479, 602]]</point>"}]}
{"id": 1612, "image": ["rtx_frames_success_8/49_bridge#episode_18961/frame_0.png", "rtx_frames_success_8/49_bridge#episode_18977/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[330, 277], [449, 462], [488, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the carrot away from the container."}, {"from": "gpt", "value": "<point>[[504, 454], [482, 676], [560, 720]]</point>"}]}
{"id": 1614, "image": ["rtx_frames_success_8/49_bridge#episode_19004/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19011/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bread loaf in the pan."}, {"from": "gpt", "value": "<point>[[432, 295], [562, 447], [590, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[227, 318], [532, 325], [687, 428]]</point>"}]}
{"id": 1616, "image": ["rtx_frames_success_8/49_bridge#episode_19071/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19071/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue spoon."}, {"from": "gpt", "value": "<point>[[302, 332], [418, 255], [499, 251]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green pepper."}, {"from": "gpt", "value": "<point>[[288, 181], [188, 380], [133, 414]]</point>"}]}
{"id": 1618, "image": ["rtx_frames_success_8/49_bridge#episode_19078/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19083/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green cloth."}, {"from": "gpt", "value": "<point>[[590, 177], [366, 325], [271, 406]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[706, 137], [579, 565], [662, 672]]</point>"}]}
{"id": 1620, "image": ["rtx_frames_success_8/49_bridge#episode_19092/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19111/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the yellow cloth."}, {"from": "gpt", "value": "<point>[[341, 247], [657, 617], [693, 768]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven handle."}, {"from": "gpt", "value": "<point>[[734, 436], [604, 329], [612, 229]]</point>"}]}
{"id": 1622, "image": ["rtx_frames_success_8/49_bridge#episode_19117/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19134/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[382, 233], [263, 355], [241, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula on the plate."}, {"from": "gpt", "value": "<point>[[839, 521], [717, 661], [659, 643]]</point>"}]}
{"id": 1624, "image": ["rtx_frames_success_8/49_bridge#episode_19140/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19140/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the oven door."}, {"from": "gpt", "value": "<point>[[474, 436], [319, 495], [183, 428]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to close the oven door."}, {"from": "gpt", "value": "<point>[[562, 307], [765, 284], [798, 166]]</point>"}]}
{"id": 1626, "image": ["rtx_frames_success_8/49_bridge#episode_19145/frame_15.png", "rtx_frames_success_8/49_bridge#episode_19163/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the metal pot over the purple towel."}, {"from": "gpt", "value": "<point>[[673, 325], [562, 532], [598, 694]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[410, 181], [374, 414], [399, 502]]</point>"}]}
{"id": 1628, "image": ["rtx_frames_success_8/49_bridge#episode_19163/frame_15.png", "rtx_frames_success_8/49_bridge#episode_19164/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot."}, {"from": "gpt", "value": "<point>[[380, 384], [562, 458], [623, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pickle."}, {"from": "gpt", "value": "<point>[[607, 233], [607, 332], [673, 392]]</point>"}]}
{"id": 1630, "image": ["rtx_frames_success_8/49_bridge#episode_19165/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19169/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[283, 185], [343, 532], [438, 606]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[355, 122], [449, 657], [579, 698]]</point>"}]}
{"id": 1632, "image": ["rtx_frames_success_8/49_bridge#episode_19179/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19180/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[385, 362], [335, 454], [269, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[632, 358], [488, 458], [521, 628]]</point>"}]}
{"id": 1634, "image": ["rtx_frames_success_8/49_bridge#episode_19185/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19190/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[573, 211], [482, 266], [427, 406]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[377, 488], [499, 373], [573, 417]]</point>"}]}
{"id": 1636, "image": ["rtx_frames_success_8/49_bridge#episode_19190/frame_15.png", "rtx_frames_success_8/49_bridge#episode_19194/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the bowl on the blue towel."}, {"from": "gpt", "value": "<point>[[296, 325], [352, 454], [296, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[413, 244], [399, 436], [449, 528]]</point>"}]}
{"id": 1638, "image": ["rtx_frames_success_8/49_bridge#episode_19199/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19199/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[745, 310], [690, 395], [598, 388]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[568, 369], [382, 384], [310, 295]]</point>"}]}
{"id": 1640, "image": ["rtx_frames_success_8/49_bridge#episode_19200/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19208/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spatula on the plate."}, {"from": "gpt", "value": "<point>[[371, 307], [435, 343], [518, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[499, 118], [524, 443], [435, 628]]</point>"}]}
{"id": 1642, "image": ["rtx_frames_success_8/49_bridge#episode_19211/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19211/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up a piece of clothing from the basket."}, {"from": "gpt", "value": "<point>[[427, 281], [393, 532], [316, 650]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the clothing towards the washing machine."}, {"from": "gpt", "value": "<point>[[343, 705], [584, 447], [618, 277]]</point>"}]}
{"id": 1644, "image": ["rtx_frames_success_8/49_bridge#episode_19213/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19213/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the bottle."}, {"from": "gpt", "value": "<point>[[294, 159], [294, 244], [363, 266]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the carrot in the pot."}, {"from": "gpt", "value": "<point>[[357, 129], [299, 403], [172, 476]]</point>"}]}
{"id": 1646, "image": ["rtx_frames_success_8/49_bridge#episode_19232/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19237/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[227, 207], [579, 524], [645, 709]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the glue tube."}, {"from": "gpt", "value": "<point>[[914, 174], [565, 185], [501, 277]]</point>"}]}
{"id": 1648, "image": ["rtx_frames_success_8/49_bridge#episode_19238/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19249/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the fork."}, {"from": "gpt", "value": "<point>[[357, 303], [537, 513], [607, 665]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the cheese."}, {"from": "gpt", "value": "<point>[[817, 166], [740, 332], [839, 469]]</point>"}]}
{"id": 1650, "image": ["rtx_frames_success_8/49_bridge#episode_19250/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19274/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[562, 137], [266, 321], [163, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[476, 211], [227, 332], [144, 462]]</point>"}]}
{"id": 1652, "image": ["rtx_frames_success_8/49_bridge#episode_19274/frame_15.png", "rtx_frames_success_8/49_bridge#episode_19276/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[150, 432], [565, 624], [662, 794]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl containing the spoon."}, {"from": "gpt", "value": "<point>[[435, 211], [512, 273], [524, 380]]</point>"}]}
{"id": 1654, "image": ["rtx_frames_success_8/49_bridge#episode_19290/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19305/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[488, 587], [593, 314], [745, 251]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[413, 207], [615, 680], [742, 702]]</point>"}]}
{"id": 1656, "image": ["rtx_frames_success_8/49_bridge#episode_19313/frame_0.png", "rtx_frames_success_8/49_bridge#episode_19358/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[438, 114], [587, 366], [670, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue fork."}, {"from": "gpt", "value": "<point>[[524, 369], [399, 373], [391, 318]]</point>"}]}
{"id": 1658, "image": ["rtx_frames_success_8/49_bridge#episode_19362/frame_0.png", "rtx_frames_success_9/49_bridge#episode_2/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[479, 307], [410, 262], [352, 288]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[261, 70], [284, 248], [323, 436], [378, 624]]</point>"}]}
{"id": 1660, "image": ["rtx_frames_success_9/49_bridge#episode_2/frame_15.png", "rtx_frames_success_9/49_bridge#episode_1964/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to slide the cloth diagonally towards the spoon."}, {"from": "gpt", "value": "<point>[[367, 591], [485, 602], [598, 596], [705, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[145, 331], [301, 401], [432, 489], [562, 591]]</point>"}]}
{"id": 1662, "image": ["rtx_frames_success_9/49_bridge#episode_1973/frame_0.png", "rtx_frames_success_9/49_bridge#episode_1974/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[647, 393], [662, 454], [682, 519], [709, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the scrub brush towards the pan."}, {"from": "gpt", "value": "<point>[[430, 185], [523, 288], [615, 401], [695, 521]]</point>"}]}
{"id": 1664, "image": ["rtx_frames_success_9/49_bridge#episode_1980/frame_0.png", "rtx_frames_success_9/49_bridge#episode_1980/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[515, 308], [442, 311], [363, 331], [297, 373]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the carrot on the plate."}, {"from": "gpt", "value": "<point>[[273, 358], [383, 404], [479, 459], [588, 541]]</point>"}]}
{"id": 1666, "image": ["rtx_frames_success_9/49_bridge#episode_2006/frame_0.png", "rtx_frames_success_9/49_bridge#episode_2018/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[239, 368], [383, 445], [503, 555], [602, 670]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift a plate."}, {"from": "gpt", "value": "<point>[[486, 319], [379, 291], [247, 341], [165, 456]]</point>"}]}
{"id": 1668, "image": ["rtx_frames_success_9/49_bridge#episode_2018/frame_15.png", "rtx_frames_success_9/49_bridge#episode_2033/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spice container."}, {"from": "gpt", "value": "<point>[[198, 418], [284, 423], [338, 489], [375, 621]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the refrigerator handle."}, {"from": "gpt", "value": "<point>[[222, 402], [343, 432], [489, 437], [636, 417]]</point>"}]}
{"id": 1670, "image": ["rtx_frames_success_9/49_bridge#episode_2035/frame_0.png", "rtx_frames_success_9/49_bridge#episode_2041/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[203, 540], [369, 606], [517, 646], [676, 667]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[210, 366], [366, 379], [523, 409], [688, 462]]</point>"}]}
{"id": 1672, "image": ["rtx_frames_success_9/49_bridge#episode_2076/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19381/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[463, 178], [371, 228], [261, 317], [200, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[53, 216], [256, 343], [438, 516], [553, 704]]</point>"}]}
{"id": 1674, "image": ["rtx_frames_success_9/49_bridge#episode_19386/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19407/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[633, 263], [622, 373], [605, 466], [568, 596]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom in the bowl."}, {"from": "gpt", "value": "<point>[[523, 316], [611, 361], [679, 419], [724, 474]]</point>"}]}
{"id": 1676, "image": ["rtx_frames_success_9/49_bridge#episode_19416/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19469/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[660, 336], [549, 341], [438, 368], [306, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[419, 158], [491, 221], [555, 301], [607, 383]]</point>"}]}
{"id": 1678, "image": ["rtx_frames_success_9/49_bridge#episode_19469/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19487/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[626, 346], [524, 376], [455, 416], [374, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the plate."}, {"from": "gpt", "value": "<point>[[414, 596], [320, 571], [233, 561], [143, 561]]</point>"}]}
{"id": 1680, "image": ["rtx_frames_success_9/49_bridge#episode_19489/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19490/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the croissant."}, {"from": "gpt", "value": "<point>[[643, 411], [560, 461], [485, 519], [427, 594]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[376, 93], [327, 221], [273, 378], [231, 531]]</point>"}]}
{"id": 1682, "image": ["rtx_frames_success_9/49_bridge#episode_19492/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19494/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[370, 30], [305, 125], [212, 301], [156, 461]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[410, 5], [477, 195], [528, 386], [568, 586]]</point>"}]}
{"id": 1684, "image": ["rtx_frames_success_9/49_bridge#episode_19496/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19496/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[502, 85], [370, 216], [259, 353], [152, 486]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[111, 486], [254, 529], [408, 594], [556, 677]]</point>"}]}
{"id": 1686, "image": ["rtx_frames_success_9/49_bridge#episode_19507/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19510/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[216, 153], [382, 203], [568, 278], [763, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[461, 291], [483, 459], [496, 669], [500, 857]]</point>"}]}
{"id": 1688, "image": ["rtx_frames_success_9/49_bridge#episode_19518/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19521/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the sushi on the plate."}, {"from": "gpt", "value": "<point>[[656, 108], [489, 201], [368, 291], [211, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spice bottle."}, {"from": "gpt", "value": "<point>[[267, 135], [331, 228], [387, 326], [445, 439]]</point>"}]}
{"id": 1690, "image": ["rtx_frames_success_9/49_bridge#episode_19522/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19530/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the small metal pan."}, {"from": "gpt", "value": "<point>[[466, 175], [414, 281], [342, 381], [276, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[466, 228], [370, 263], [286, 313], [192, 396]]</point>"}]}
{"id": 1692, "image": ["rtx_frames_success_9/49_bridge#episode_19530/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19537/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[199, 378], [329, 386], [481, 419], [620, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the clothes."}, {"from": "gpt", "value": "<point>[[378, 193], [444, 343], [519, 556], [562, 747]]</point>"}]}
{"id": 1694, "image": ["rtx_frames_success_9/49_bridge#episode_19546/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19547/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[353, 263], [508, 336], [652, 426], [788, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the utensil."}, {"from": "gpt", "value": "<point>[[222, 120], [271, 246], [303, 376], [331, 499]]</point>"}]}
{"id": 1696, "image": ["rtx_frames_success_9/49_bridge#episode_19566/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19571/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon in the sink."}, {"from": "gpt", "value": "<point>[[673, 208], [600, 266], [543, 333], [485, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[423, 70], [368, 165], [301, 253], [212, 336]]</point>"}]}
{"id": 1698, "image": ["rtx_frames_success_9/49_bridge#episode_19587/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19667/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[410, 5], [506, 130], [641, 256], [782, 333]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the colander towards the towel."}, {"from": "gpt", "value": "<point>[[301, 286], [318, 414], [321, 494], [314, 591]]</point>"}]}
{"id": 1700, "image": ["rtx_frames_success_9/49_bridge#episode_19669/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19707/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cheese slice."}, {"from": "gpt", "value": "<point>[[321, 288], [457, 456], [553, 602], [650, 777]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[628, 33], [682, 243], [727, 464], [769, 737]]</point>"}]}
{"id": 1702, "image": ["rtx_frames_success_9/49_bridge#episode_19707/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19738/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the sushi towards the bowl."}, {"from": "gpt", "value": "<point>[[684, 426], [592, 489], [519, 554], [451, 642]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[720, 138], [703, 213], [673, 306], [632, 381]]</point>"}]}
{"id": 1704, "image": ["rtx_frames_success_9/49_bridge#episode_19750/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19762/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan containing broccoli."}, {"from": "gpt", "value": "<point>[[560, 667], [577, 602], [605, 544], [647, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[447, 201], [393, 276], [353, 361], [323, 466]]</point>"}]}
{"id": 1706, "image": ["rtx_frames_success_9/49_bridge#episode_19762/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19772/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the carrot towards the pot."}, {"from": "gpt", "value": "<point>[[306, 446], [398, 449], [504, 494], [594, 566]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[425, 298], [549, 343], [639, 391], [739, 469]]</point>"}]}
{"id": 1708, "image": ["rtx_frames_success_9/49_bridge#episode_19772/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19782/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[746, 416], [643, 391], [558, 378], [461, 378]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the pan."}, {"from": "gpt", "value": "<point>[[479, 5], [427, 85], [393, 170], [374, 268]]</point>"}]}
{"id": 1710, "image": ["rtx_frames_success_9/49_bridge#episode_19801/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19814/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[357, 175], [361, 293], [376, 421], [402, 531]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position above the hotdog bun."}, {"from": "gpt", "value": "<point>[[372, 213], [494, 288], [615, 398], [703, 511]]</point>"}]}
{"id": 1712, "image": ["rtx_frames_success_9/49_bridge#episode_19819/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19821/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[442, 45], [429, 203], [429, 429], [440, 662]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[639, 211], [583, 356], [539, 541], [517, 709]]</point>"}]}
{"id": 1714, "image": ["rtx_frames_success_9/49_bridge#episode_19823/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19829/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[372, 501], [410, 429], [451, 383], [515, 356]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[462, 343], [534, 341], [598, 343], [658, 371]]</point>"}]}
{"id": 1716, "image": ["rtx_frames_success_9/49_bridge#episode_19836/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19837/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[395, 40], [335, 155], [286, 271], [250, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the rag."}, {"from": "gpt", "value": "<point>[[489, 98], [521, 311], [573, 514], [647, 687]]</point>"}]}
{"id": 1718, "image": ["rtx_frames_success_9/49_bridge#episode_19846/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19846/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the bowl from the countertop."}, {"from": "gpt", "value": "<point>[[342, 288], [481, 436], [607, 604], [711, 787]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the knife on the middle of the table."}, {"from": "gpt", "value": "<point>[[690, 629], [626, 551], [556, 499], [485, 459]]</point>"}]}
{"id": 1720, "image": ["rtx_frames_success_9/49_bridge#episode_19875/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19884/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up a piece of red cloth."}, {"from": "gpt", "value": "<point>[[624, 436], [586, 564], [568, 679], [553, 817]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the spoon."}, {"from": "gpt", "value": "<point>[[568, 236], [543, 368], [509, 514], [466, 659]]</point>"}]}
{"id": 1722, "image": ["rtx_frames_success_9/49_bridge#episode_19888/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19890/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the blue block in the pot."}, {"from": "gpt", "value": "<point>[[697, 246], [660, 353], [633, 484], [615, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[278, 173], [224, 273], [188, 396], [162, 521]]</point>"}]}
{"id": 1724, "image": ["rtx_frames_success_9/49_bridge#episode_19890/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19892/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the towel."}, {"from": "gpt", "value": "<point>[[73, 434], [137, 396], [216, 376], [291, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the croissant in the bowl."}, {"from": "gpt", "value": "<point>[[406, 303], [387, 401], [383, 471], [387, 579]]</point>"}]}
{"id": 1726, "image": ["rtx_frames_success_9/49_bridge#episode_19920/frame_15.png", "rtx_frames_success_9/49_bridge#episode_19923/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the stove."}, {"from": "gpt", "value": "<point>[[288, 108], [398, 236], [524, 419], [618, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the egg."}, {"from": "gpt", "value": "<point>[[383, 0], [408, 168], [419, 366], [421, 546]]</point>"}]}
{"id": 1728, "image": ["rtx_frames_success_9/49_bridge#episode_19931/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19931/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[365, 80], [320, 155], [282, 246], [252, 346]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[246, 306], [346, 363], [429, 426], [532, 506]]</point>"}]}
{"id": 1730, "image": ["rtx_frames_success_9/49_bridge#episode_19948/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19965/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[250, 80], [372, 271], [502, 439], [650, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[363, 246], [502, 404], [637, 599], [752, 787]]</point>"}]}
{"id": 1732, "image": ["rtx_frames_success_9/49_bridge#episode_19966/frame_0.png", "rtx_frames_success_9/49_bridge#episode_19991/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[609, 391], [528, 376], [436, 386], [350, 404]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the microwave door closed."}, {"from": "gpt", "value": "<point>[[338, 195], [408, 198], [487, 180], [543, 130]]</point>"}]}
{"id": 1734, "image": ["rtx_frames_success_9/49_bridge#episode_20009/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20043/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[491, 98], [389, 180], [284, 311], [211, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[430, 193], [506, 363], [577, 561], [635, 752]]</point>"}]}
{"id": 1736, "image": ["rtx_frames_success_9/49_bridge#episode_20048/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20050/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[408, 33], [352, 138], [306, 258], [265, 396]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[464, 170], [352, 251], [239, 371], [111, 539]]</point>"}]}
{"id": 1738, "image": ["rtx_frames_success_9/49_bridge#episode_20070/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20071/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[313, 60], [235, 154], [152, 269], [95, 423]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl on the plate."}, {"from": "gpt", "value": "<point>[[462, 231], [495, 390], [495, 511], [470, 687]]</point>"}]}
{"id": 1740, "image": ["rtx_frames_success_9/49_bridge#episode_20074/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20095/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[317, 88], [445, 170], [577, 280], [717, 423]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[598, 269], [672, 352], [713, 429], [746, 533]]</point>"}]}
{"id": 1742, "image": ["rtx_frames_success_9/49_bridge#episode_20102/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20107/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[152, 203], [152, 286], [161, 374], [185, 434]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[210, 363], [404, 401], [560, 423], [771, 429]]</point>"}]}
{"id": 1744, "image": ["rtx_frames_success_9/49_bridge#episode_20113/frame_15.png", "rtx_frames_success_9/49_bridge#episode_20118/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the green vegetable on the blue towel."}, {"from": "gpt", "value": "<point>[[387, 137], [499, 324], [577, 544], [639, 791]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[445, 77], [466, 247], [470, 418], [457, 632]]</point>"}]}
{"id": 1746, "image": ["rtx_frames_success_9/49_bridge#episode_20141/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20144/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[25, 302], [115, 291], [235, 291], [338, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[474, 577], [396, 555], [313, 500], [214, 401]]</point>"}]}
{"id": 1748, "image": ["rtx_frames_success_9/49_bridge#episode_20154/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20154/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[206, 242], [321, 231], [420, 264], [499, 357]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon to the right of the pot."}, {"from": "gpt", "value": "<point>[[536, 247], [581, 352], [663, 489], [738, 610]]</point>"}]}
{"id": 1750, "image": ["rtx_frames_success_9/49_bridge#episode_20183/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20191/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[779, 451], [659, 396], [560, 363], [420, 368]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[387, 44], [387, 203], [404, 379], [453, 566]]</point>"}]}
{"id": 1752, "image": ["rtx_frames_success_9/49_bridge#episode_20206/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20208/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[239, 363], [359, 423], [495, 445], [602, 434]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon in the sink."}, {"from": "gpt", "value": "<point>[[371, 49], [420, 187], [453, 291], [466, 445]]</point>"}]}
{"id": 1754, "image": ["rtx_frames_success_9/49_bridge#episode_20212/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20212/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[288, 368], [354, 363], [404, 379], [470, 423]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the sushi towards the pot."}, {"from": "gpt", "value": "<point>[[466, 379], [359, 352], [247, 390], [169, 478]]</point>"}]}
{"id": 1756, "image": ["rtx_frames_success_9/49_bridge#episode_20217/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20217/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[255, 154], [449, 225], [585, 308], [738, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[746, 445], [630, 516], [499, 544], [354, 566]]</point>"}]}
{"id": 1758, "image": ["rtx_frames_success_9/49_bridge#episode_20220/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20220/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[398, 129], [434, 265], [456, 412], [479, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the sushi in a pot."}, {"from": "gpt", "value": "<point>[[468, 540], [438, 455], [407, 409], [360, 379], [294, 386]]</point>"}]}
{"id": 1760, "image": ["rtx_frames_success_9/49_bridge#episode_20223/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20225/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[202, 379], [313, 401], [412, 401], [536, 379]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[373, 61], [348, 167], [316, 260], [267, 381]]</point>"}]}
{"id": 1762, "image": ["rtx_frames_success_9/49_bridge#episode_20225/frame_15.png", "rtx_frames_success_9/49_bridge#episode_20226/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[252, 402], [354, 402], [466, 434], [561, 492]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green towel."}, {"from": "gpt", "value": "<point>[[647, 401], [655, 511], [643, 593], [602, 659]]</point>"}]}
{"id": 1764, "image": ["rtx_frames_success_9/49_bridge#episode_20242/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20242/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spatula."}, {"from": "gpt", "value": "<point>[[288, 434], [301, 335], [338, 247], [400, 143]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the green vegetable on the middle of the table next to a green papper."}, {"from": "gpt", "value": "<point>[[400, 93], [433, 236], [457, 401], [445, 571]]</point>"}]}
{"id": 1766, "image": ["rtx_frames_success_9/49_bridge#episode_20254/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20254/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the pan."}, {"from": "gpt", "value": "<point>[[502, 394], [413, 414], [305, 427], [203, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spatula to the left of the pot."}, {"from": "gpt", "value": "<point>[[169, 399], [248, 419], [345, 412], [430, 391]]</point>"}]}
{"id": 1768, "image": ["rtx_frames_success_9/49_bridge#episode_20273/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20294/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[439, 136], [441, 207], [422, 258], [381, 301]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the small metal vessel."}, {"from": "gpt", "value": "<point>[[375, 215], [439, 311], [502, 444], [536, 591]]</point>"}]}
{"id": 1770, "image": ["rtx_frames_success_9/49_bridge#episode_20302/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20312/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[93, 237], [210, 220], [324, 220], [439, 250]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[138, 247], [225, 308], [301, 376], [366, 465]]</point>"}]}
{"id": 1772, "image": ["rtx_frames_success_9/49_bridge#episode_20339/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20340/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the towel."}, {"from": "gpt", "value": "<point>[[509, 157], [460, 235], [428, 341], [417, 447]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the broccoli downwards."}, {"from": "gpt", "value": "<point>[[479, 172], [483, 364], [506, 591], [545, 753]]</point>"}]}
{"id": 1774, "image": ["rtx_frames_success_9/49_bridge#episode_20360/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20366/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[333, 91], [286, 169], [246, 260], [214, 374]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[386, 157], [299, 240], [203, 366], [131, 485]]</point>"}]}
{"id": 1776, "image": ["rtx_frames_success_9/49_bridge#episode_20370/frame_15.png", "rtx_frames_success_9/49_bridge#episode_20379/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to upright the spice jar."}, {"from": "gpt", "value": "<point>[[354, 354], [331, 301], [282, 245], [227, 230]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[74, 197], [222, 391], [367, 533], [513, 652]]</point>"}]}
{"id": 1778, "image": ["rtx_frames_success_9/49_bridge#episode_20386/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20397/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the plate."}, {"from": "gpt", "value": "<point>[[451, 187], [356, 336], [294, 505], [267, 672]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[540, 177], [555, 298], [559, 492], [553, 601]]</point>"}]}
{"id": 1780, "image": ["rtx_frames_success_9/49_bridge#episode_20404/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20410/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[542, 187], [460, 217], [383, 270], [324, 333]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[127, 359], [214, 414], [280, 472], [347, 561]]</point>"}]}
{"id": 1782, "image": ["rtx_frames_success_9/49_bridge#episode_20410/frame_15.png", "rtx_frames_success_9/49_bridge#episode_20427/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[328, 497], [392, 472], [468, 465], [553, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the door closed."}, {"from": "gpt", "value": "<point>[[191, 341], [282, 318], [339, 293], [388, 253]]</point>"}]}
{"id": 1784, "image": ["rtx_frames_success_9/49_bridge#episode_20498/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20519/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[405, 311], [392, 399], [396, 503], [420, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana in the pan."}, {"from": "gpt", "value": "<point>[[593, 424], [447, 374], [320, 351], [201, 354]]</point>"}]}
{"id": 1786, "image": ["rtx_frames_success_9/49_bridge#episode_20519/frame_15.png", "rtx_frames_success_9/49_bridge#episode_20534/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana in the pan."}, {"from": "gpt", "value": "<point>[[195, 298], [383, 359], [576, 442], [778, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[271, 187], [312, 290], [347, 409], [364, 505]]</point>"}]}
{"id": 1788, "image": ["rtx_frames_success_9/49_bridge#episode_20549/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20549/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the chicken drumstick."}, {"from": "gpt", "value": "<point>[[619, 437], [544, 384], [477, 364], [392, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the chicken drumstick over the table."}, {"from": "gpt", "value": "<point>[[388, 258], [286, 288], [195, 341], [112, 399]]</point>"}]}
{"id": 1790, "image": ["rtx_frames_success_9/49_bridge#episode_20558/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20565/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[475, 40], [449, 144], [413, 268], [371, 361]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the green and yellow item."}, {"from": "gpt", "value": "<point>[[589, 245], [513, 235], [447, 245], [384, 273]]</point>"}]}
{"id": 1792, "image": ["rtx_frames_success_9/49_bridge#episode_20567/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20574/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[614, 328], [551, 399], [509, 472], [468, 566]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[750, 205], [627, 318], [504, 472], [388, 657]]</point>"}]}
{"id": 1794, "image": ["rtx_frames_success_9/49_bridge#episode_20581/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20601/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[706, 404], [608, 389], [487, 381], [373, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[278, 146], [305, 225], [318, 298], [320, 374]]</point>"}]}
{"id": 1796, "image": ["rtx_frames_success_9/49_bridge#episode_20604/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20604/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bread."}, {"from": "gpt", "value": "<point>[[288, 5], [345, 129], [411, 301], [453, 424]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bread on the green towel."}, {"from": "gpt", "value": "<point>[[504, 293], [388, 326], [280, 374], [184, 442]]</point>"}]}
{"id": 1798, "image": ["rtx_frames_success_9/49_bridge#episode_20616/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20629/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[587, 146], [610, 273], [631, 442], [636, 581]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the toy cat in the bowl."}, {"from": "gpt", "value": "<point>[[449, 5], [422, 177], [407, 386], [403, 588]]</point>"}]}
{"id": 1800, "image": ["rtx_frames_success_9/49_bridge#episode_20629/frame_15.png", "rtx_frames_success_9/49_bridge#episode_20647/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the toy cat towards the purple cloth."}, {"from": "gpt", "value": "<point>[[445, 487], [538, 490], [627, 480], [714, 447], [733, 298]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[579, 270], [597, 336], [602, 409], [608, 483]]</point>"}]}
{"id": 1802, "image": ["rtx_frames_success_9/49_bridge#episode_20649/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20664/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[232, 313], [313, 386], [400, 514], [460, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[819, 320], [753, 351], [695, 459], [634, 568], [579, 660]]</point>"}]}
{"id": 1804, "image": ["rtx_frames_success_9/49_bridge#episode_20666/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20672/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[472, 266], [391, 251], [313, 282], [269, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl containing the corn."}, {"from": "gpt", "value": "<point>[[278, 97], [382, 124], [466, 243], [510, 317], [573, 429]]</point>"}]}
{"id": 1806, "image": ["rtx_frames_success_9/49_bridge#episode_20675/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20675/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the cube."}, {"from": "gpt", "value": "<point>[[582, 429], [478, 336], [379, 336], [287, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the cube on the plate."}, {"from": "gpt", "value": "<point>[[324, 31], [316, 154], [298, 270], [287, 394]]</point>"}]}
{"id": 1808, "image": ["rtx_frames_success_9/49_bridge#episode_20684/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20694/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[452, 69], [388, 181], [359, 266], [330, 332], [287, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[151, 544], [284, 525], [420, 514], [588, 548]]</point>"}]}
{"id": 1810, "image": ["rtx_frames_success_9/49_bridge#episode_20706/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20720/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the tomato in the bowl."}, {"from": "gpt", "value": "<point>[[171, 718], [310, 664], [449, 633], [582, 614], [733, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave door handle."}, {"from": "gpt", "value": "<point>[[944, 506], [857, 467], [724, 440], [573, 452], [486, 533]]</point>"}]}
{"id": 1812, "image": ["rtx_frames_success_9/49_bridge#episode_20729/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20750/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[365, 170], [400, 278], [440, 398], [486, 517], [521, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[591, 371], [489, 332], [333, 340], [217, 375]]</point>"}]}
{"id": 1814, "image": ["rtx_frames_success_9/49_bridge#episode_20750/frame_15.png", "rtx_frames_success_9/49_bridge#episode_20762/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the fork towards the right edge of the stove."}, {"from": "gpt", "value": "<point>[[200, 390], [379, 371], [458, 386], [536, 409]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container with the carrot."}, {"from": "gpt", "value": "<point>[[657, 571], [568, 579], [495, 595], [423, 614]]</point>"}]}
{"id": 1816, "image": ["rtx_frames_success_9/49_bridge#episode_20765/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20765/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[226, 162], [319, 239], [385, 297], [443, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl to the left of a spoon."}, {"from": "gpt", "value": "<point>[[405, 263], [527, 328], [625, 425]]</point>"}]}
{"id": 1818, "image": ["rtx_frames_success_9/49_bridge#episode_20778/frame_0.png", "rtx_frames_success_9/49_bridge#episode_20784/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the refrigerator handle."}, {"from": "gpt", "value": "<point>[[400, 205], [544, 309], [611, 359], [686, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet handle."}, {"from": "gpt", "value": "<point>[[232, 768], [356, 761], [481, 757], [654, 707]]</point>"}]}
{"id": 1820, "image": ["rtx_frames_success_10/49_bridge#episode_214/frame_0.png", "rtx_frames_success_10/49_bridge#episode_217/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the mushroom."}, {"from": "gpt", "value": "<point>[[539, 573], [430, 446], [271, 376], [123, 352]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[335, 80], [472, 183], [592, 390], [669, 676]]</point>"}]}
{"id": 1822, "image": ["rtx_frames_success_10/49_bridge#episode_221/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2084/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[236, 474], [278, 394], [349, 352], [496, 333]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the red spoon."}, {"from": "gpt", "value": "<point>[[461, 479], [398, 404], [356, 366], [285, 338]]</point>"}]}
{"id": 1824, "image": ["rtx_frames_success_10/49_bridge#episode_2084/frame_15.png", "rtx_frames_success_10/49_bridge#episode_2087/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the spoon next to the pan and in front of the banana."}, {"from": "gpt", "value": "<point>[[261, 254], [349, 310], [437, 394], [451, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[391, 122], [496, 207], [525, 296], [539, 469]]</point>"}]}
{"id": 1826, "image": ["rtx_frames_success_10/49_bridge#episode_2088/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2088/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the vessel."}, {"from": "gpt", "value": "<point>[[426, 169], [391, 207], [394, 254], [419, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place a sushi roll on the middle of the talbe."}, {"from": "gpt", "value": "<point>[[394, 225], [444, 362], [458, 451], [437, 620]]</point>"}]}
{"id": 1828, "image": ["rtx_frames_success_10/49_bridge#episode_2096/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2104/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[250, 460], [313, 390], [391, 319], [563, 310]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[56, 296], [261, 329], [430, 399], [507, 615]]</point>"}]}
{"id": 1830, "image": ["rtx_frames_success_10/49_bridge#episode_2113/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2113/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[500, 390], [430, 366], [342, 366], [222, 423]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the knife to the right of the pot."}, {"from": "gpt", "value": "<point>[[155, 286], [342, 282], [528, 329], [697, 479]]</point>"}]}
{"id": 1832, "image": ["rtx_frames_success_10/49_bridge#episode_2125/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2126/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[67, 413], [151, 460], [264, 479], [398, 498]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow utensil."}, {"from": "gpt", "value": "<point>[[384, 390], [440, 441], [567, 469], [648, 465]]</point>"}]}
{"id": 1834, "image": ["rtx_frames_success_10/49_bridge#episode_2131/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2134/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[327, 404], [430, 460], [542, 469], [694, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the tin can."}, {"from": "gpt", "value": "<point>[[320, 178], [278, 272], [285, 366], [335, 479]]</point>"}]}
{"id": 1836, "image": ["rtx_frames_success_10/49_bridge#episode_2142/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2148/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[394, 357], [451, 390], [475, 441], [475, 516]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[437, 258], [352, 282], [306, 329], [271, 446]]</point>"}]}
{"id": 1838, "image": ["rtx_frames_success_10/49_bridge#episode_2160/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2177/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green rag."}, {"from": "gpt", "value": "<point>[[447, 244], [335, 286], [275, 352], [243, 418]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[451, 296], [539, 366], [616, 469], [637, 577]]</point>"}]}
{"id": 1840, "image": ["rtx_frames_success_10/49_bridge#episode_2177/frame_15.png", "rtx_frames_success_10/49_bridge#episode_2193/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife to the left edge of the table."}, {"from": "gpt", "value": "<point>[[437, 239], [359, 263], [303, 324], [257, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cheese."}, {"from": "gpt", "value": "<point>[[271, 249], [394, 399], [518, 441], [697, 460]]</point>"}]}
{"id": 1842, "image": ["rtx_frames_success_10/49_bridge#episode_2193/frame_15.png", "rtx_frames_success_10/49_bridge#episode_2194/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cheese."}, {"from": "gpt", "value": "<point>[[715, 469], [609, 488], [542, 535], [486, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[743, 169], [609, 169], [532, 216], [440, 310]]</point>"}]}
{"id": 1844, "image": ["rtx_frames_success_10/49_bridge#episode_2205/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2209/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[535, 160], [496, 244], [479, 352], [518, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the beet."}, {"from": "gpt", "value": "<point>[[419, 315], [342, 286], [299, 282], [229, 300]]</point>"}]}
{"id": 1846, "image": ["rtx_frames_success_10/49_bridge#episode_2212/frame_0.png", "rtx_frames_success_10/49_bridge#episode_2214/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[440, 225], [387, 254], [349, 291], [327, 357]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[585, 258], [521, 230], [454, 235], [391, 277]]</point>"}]}
{"id": 1848, "image": ["rtx_frames_success_10/49_bridge#episode_20787/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20788/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[609, 315], [546, 324], [479, 432], [472, 563]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange towel."}, {"from": "gpt", "value": "<point>[[651, 357], [680, 432], [694, 498], [669, 582]]</point>"}]}
{"id": 1850, "image": ["rtx_frames_success_10/49_bridge#episode_20789/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20789/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the utensil."}, {"from": "gpt", "value": "<point>[[651, 394], [574, 329], [472, 296], [373, 380]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the utensil towards the cloth."}, {"from": "gpt", "value": "<point>[[320, 380], [387, 394], [454, 451], [475, 601]]</point>"}]}
{"id": 1852, "image": ["rtx_frames_success_10/49_bridge#episode_20801/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20801/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[447, 324], [482, 296], [528, 296], [570, 310]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the red object in the pot."}, {"from": "gpt", "value": "<point>[[373, 324], [342, 300], [299, 305], [246, 347]]</point>"}]}
{"id": 1854, "image": ["rtx_frames_success_10/49_bridge#episode_20820/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20820/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[222, 146], [292, 174], [320, 216], [363, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot towards the spatula."}, {"from": "gpt", "value": "<point>[[599, 207], [651, 324], [676, 437], [676, 531]]</point>"}]}
{"id": 1856, "image": ["rtx_frames_success_10/49_bridge#episode_20821/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20821/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[349, 192], [335, 319], [349, 432], [440, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the bowl."}, {"from": "gpt", "value": "<point>[[359, 352], [306, 310], [261, 310], [204, 338]]</point>"}]}
{"id": 1858, "image": ["rtx_frames_success_10/49_bridge#episode_20828/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20828/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green item."}, {"from": "gpt", "value": "<point>[[144, 385], [296, 338], [479, 376], [665, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the green item at the left rear of the table."}, {"from": "gpt", "value": "<point>[[616, 413], [553, 338], [486, 300], [401, 282]]</point>"}]}
{"id": 1860, "image": ["rtx_frames_success_10/49_bridge#episode_20834/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20834/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[222, 263], [433, 385], [521, 469], [595, 667]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the can at the left rear of the table."}, {"from": "gpt", "value": "<point>[[585, 554], [570, 437], [525, 357], [465, 315]]</point>"}]}
{"id": 1862, "image": ["rtx_frames_success_10/49_bridge#episode_20901/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20902/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the bowl."}, {"from": "gpt", "value": "<point>[[349, 315], [468, 338], [574, 432], [665, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[264, 347], [324, 291], [373, 300], [433, 371]]</point>"}]}
{"id": 1864, "image": ["rtx_frames_success_10/49_bridge#episode_20916/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20916/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[511, 282], [475, 254], [423, 244], [335, 296]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pan in front of the cans."}, {"from": "gpt", "value": "<point>[[335, 192], [454, 244], [514, 286], [599, 357]]</point>"}]}
{"id": 1866, "image": ["rtx_frames_success_10/49_bridge#episode_20947/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20971/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave door."}, {"from": "gpt", "value": "<point>[[461, 291], [342, 254], [225, 263], [116, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the sushi roll."}, {"from": "gpt", "value": "<point>[[736, 188], [585, 249], [496, 366], [454, 507]]</point>"}]}
{"id": 1868, "image": ["rtx_frames_success_10/49_bridge#episode_20971/frame_15.png", "rtx_frames_success_10/49_bridge#episode_20975/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the sushi roll towards the violet cloth."}, {"from": "gpt", "value": "<point>[[507, 385], [599, 394], [658, 432], [704, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[461, 300], [546, 338], [599, 376], [630, 441]]</point>"}]}
{"id": 1870, "image": ["rtx_frames_success_10/49_bridge#episode_20975/frame_15.png", "rtx_frames_success_10/49_bridge#episode_20980/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the pot."}, {"from": "gpt", "value": "<point>[[528, 263], [458, 272], [387, 319], [349, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[680, 249], [588, 404], [546, 526], [546, 662]]</point>"}]}
{"id": 1872, "image": ["rtx_frames_success_10/49_bridge#episode_20981/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20988/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[215, 376], [366, 333], [500, 347], [637, 427]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[236, 254], [356, 427], [384, 545], [426, 770]]</point>"}]}
{"id": 1874, "image": ["rtx_frames_success_10/49_bridge#episode_20988/frame_15.png", "rtx_frames_success_10/49_bridge#episode_20991/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the knife in the pot."}, {"from": "gpt", "value": "<point>[[359, 704], [370, 592], [426, 507], [500, 469]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the food in the bowl."}, {"from": "gpt", "value": "<point>[[609, 221], [521, 291], [486, 362], [465, 441]]</point>"}]}
{"id": 1876, "image": ["rtx_frames_success_10/49_bridge#episode_20996/frame_0.png", "rtx_frames_success_10/49_bridge#episode_20996/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[197, 207], [345, 310], [430, 427], [454, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place a pot next to a sushi roll."}, {"from": "gpt", "value": "<point>[[451, 451], [496, 394], [570, 362], [630, 394]]</point>"}]}
{"id": 1878, "image": ["rtx_frames_success_10/49_bridge#episode_21010/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21010/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[683, 366], [525, 376], [384, 437], [246, 531]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the carrot towards the top right side of the table."}, {"from": "gpt", "value": "<point>[[296, 394], [412, 394], [532, 394], [687, 479]]</point>"}]}
{"id": 1880, "image": ["rtx_frames_success_10/49_bridge#episode_21014/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21018/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange towel."}, {"from": "gpt", "value": "<point>[[415, 131], [373, 197], [356, 310], [391, 516]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position above the green vegetable."}, {"from": "gpt", "value": "<point>[[454, 146], [556, 291], [581, 404], [595, 582]]</point>"}]}
{"id": 1882, "image": ["rtx_frames_success_10/49_bridge#episode_21018/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21021/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the green vegetable to pot."}, {"from": "gpt", "value": "<point>[[511, 362], [458, 338], [387, 333], [327, 352]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red pepper."}, {"from": "gpt", "value": "<point>[[454, 310], [366, 310], [320, 338], [278, 404]]</point>"}]}
{"id": 1884, "image": ["rtx_frames_success_10/49_bridge#episode_21035/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21035/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cube."}, {"from": "gpt", "value": "<point>[[630, 385], [680, 385], [711, 404], [732, 455]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the cube in front of the spoon."}, {"from": "gpt", "value": "<point>[[472, 282], [419, 291], [370, 324], [331, 394]]</point>"}]}
{"id": 1886, "image": ["rtx_frames_success_10/49_bridge#episode_21042/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21042/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[440, 437], [363, 418], [268, 423], [162, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spoon in the pot."}, {"from": "gpt", "value": "<point>[[113, 371], [225, 333], [303, 329], [423, 347]]</point>"}]}
{"id": 1888, "image": ["rtx_frames_success_10/49_bridge#episode_21048/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21048/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the fork."}, {"from": "gpt", "value": "<point>[[637, 263], [799, 376], [838, 507], [835, 756]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to scoop the grapes with the fork."}, {"from": "gpt", "value": "<point>[[694, 531], [472, 465], [349, 446], [222, 455]]</point>"}]}
{"id": 1890, "image": ["rtx_frames_success_10/49_bridge#episode_21060/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21062/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[197, 300], [254, 404], [342, 531], [482, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[673, 493], [634, 559], [630, 620], [651, 718]]</point>"}]}
{"id": 1892, "image": ["rtx_frames_success_10/49_bridge#episode_21063/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21063/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[412, 136], [504, 254], [525, 376], [539, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the object on the green towel."}, {"from": "gpt", "value": "<point>[[549, 554], [401, 441], [268, 404], [158, 427]]</point>"}]}
{"id": 1894, "image": ["rtx_frames_success_10/49_bridge#episode_21091/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21093/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[577, 371], [641, 385], [697, 408], [768, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the hot sauce bottle."}, {"from": "gpt", "value": "<point>[[613, 437], [708, 437], [792, 437], [856, 502]]</point>"}]}
{"id": 1896, "image": ["rtx_frames_success_10/49_bridge#episode_21093/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21111/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to knock over the hot sauce bottle."}, {"from": "gpt", "value": "<point>[[820, 563], [785, 563], [739, 582], [697, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[342, 310], [447, 394], [500, 502], [560, 714]]</point>"}]}
{"id": 1898, "image": ["rtx_frames_success_10/49_bridge#episode_21111/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21117/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the knife at the left rear of the table."}, {"from": "gpt", "value": "<point>[[493, 488], [475, 413], [423, 338], [345, 300]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the carrot."}, {"from": "gpt", "value": "<point>[[412, 66], [472, 127], [475, 150], [482, 211]]</point>"}]}
{"id": 1900, "image": ["rtx_frames_success_10/49_bridge#episode_21117/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21128/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the carrot over the spatula."}, {"from": "gpt", "value": "<point>[[486, 239], [430, 347], [384, 451], [370, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the pan."}, {"from": "gpt", "value": "<point>[[563, 117], [489, 272], [444, 404], [475, 582]]</point>"}]}
{"id": 1902, "image": ["rtx_frames_success_10/49_bridge#episode_21130/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21130/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[370, 263], [433, 352], [468, 451], [539, 657]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the fork to the left."}, {"from": "gpt", "value": "<point>[[542, 601], [408, 455], [317, 423], [197, 441]]</point>"}]}
{"id": 1904, "image": ["rtx_frames_success_10/49_bridge#episode_21135/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21135/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[504, 188], [468, 329], [482, 479], [602, 718]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spoon on front of the pot."}, {"from": "gpt", "value": "<point>[[542, 582], [394, 516], [275, 469], [162, 479]]</point>"}]}
{"id": 1906, "image": ["rtx_frames_success_10/49_bridge#episode_21148/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21148/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[454, 310], [320, 282], [246, 310], [190, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[232, 437], [299, 404], [363, 408], [437, 460]]</point>"}]}
{"id": 1908, "image": ["rtx_frames_success_10/49_bridge#episode_21155/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21173/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[609, 380], [521, 371], [468, 385], [398, 441]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[426, 263], [454, 357], [482, 385], [556, 413]]</point>"}]}
{"id": 1910, "image": ["rtx_frames_success_10/49_bridge#episode_21178/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21178/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana in the colander."}, {"from": "gpt", "value": "<point>[[553, 160], [475, 150], [433, 160], [391, 216]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move away from the colander with the banana."}, {"from": "gpt", "value": "<point>[[430, 225], [518, 310], [546, 413], [528, 526]]</point>"}]}
{"id": 1912, "image": ["rtx_frames_success_10/49_bridge#episode_21183/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21184/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the orange container."}, {"from": "gpt", "value": "<point>[[165, 399], [264, 319], [349, 296], [451, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[296, 310], [394, 394], [504, 446], [630, 441]]</point>"}]}
{"id": 1914, "image": ["rtx_frames_success_10/49_bridge#episode_21184/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21189/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[585, 300], [451, 291], [356, 305], [285, 366]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple towel."}, {"from": "gpt", "value": "<point>[[673, 460], [630, 512], [606, 568], [602, 648]]</point>"}]}
{"id": 1916, "image": ["rtx_frames_success_10/49_bridge#episode_21209/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21227/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[434, 308], [526, 338], [599, 376], [675, 491]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the potato."}, {"from": "gpt", "value": "<point>[[377, 333], [475, 413], [549, 451], [694, 479]]</point>"}]}
{"id": 1918, "image": ["rtx_frames_success_10/49_bridge#episode_21228/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21231/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[496, 338], [560, 404], [588, 488], [585, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[472, 122], [405, 188], [356, 282], [363, 423]]</point>"}]}
{"id": 1920, "image": ["rtx_frames_success_10/49_bridge#episode_21235/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21239/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[130, 235], [137, 324], [165, 394], [232, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[271, 286], [423, 371], [454, 432], [479, 610]]</point>"}]}
{"id": 1922, "image": ["rtx_frames_success_10/49_bridge#episode_21249/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21260/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[151, 343], [254, 300], [373, 300], [525, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the potato."}, {"from": "gpt", "value": "<point>[[637, 249], [546, 366], [514, 516], [542, 746]]</point>"}]}
{"id": 1924, "image": ["rtx_frames_success_10/49_bridge#episode_21271/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21272/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[250, 254], [201, 300], [187, 338], [180, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[137, 282], [183, 310], [229, 329], [282, 300]]</point>"}]}
{"id": 1926, "image": ["rtx_frames_success_10/49_bridge#episode_21272/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21279/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[408, 249], [500, 380], [556, 493], [592, 671]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[88, 178], [130, 244], [180, 272], [271, 268]]</point>"}]}
{"id": 1928, "image": ["rtx_frames_success_10/49_bridge#episode_21288/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21288/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[496, 427], [370, 404], [306, 441], [215, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the carrot on the plate."}, {"from": "gpt", "value": "<point>[[204, 498], [282, 455], [359, 451], [444, 484]]</point>"}]}
{"id": 1930, "image": ["rtx_frames_success_10/49_bridge#episode_21296/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21299/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[271, 282], [335, 338], [465, 394], [560, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[349, 235], [320, 272], [303, 300], [296, 343]]</point>"}]}
{"id": 1932, "image": ["rtx_frames_success_10/49_bridge#episode_21303/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21303/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[602, 352], [570, 329], [539, 329], [482, 352]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the front right corner of the table."}, {"from": "gpt", "value": "<point>[[451, 319], [574, 432], [644, 531], [690, 704]]</point>"}]}
{"id": 1934, "image": ["rtx_frames_success_10/49_bridge#episode_21306/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21323/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[366, 235], [289, 258], [243, 300], [180, 418]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[282, 113], [433, 272], [486, 423], [532, 620]]</point>"}]}
{"id": 1936, "image": ["rtx_frames_success_10/49_bridge#episode_21337/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21352/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place pot on the stove."}, {"from": "gpt", "value": "<point>[[687, 192], [609, 188], [560, 207], [511, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[701, 545], [546, 418], [384, 380], [239, 394]]</point>"}]}
{"id": 1938, "image": ["rtx_frames_success_10/49_bridge#episode_21352/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21373/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon."}, {"from": "gpt", "value": "<point>[[215, 258], [335, 315], [423, 394], [486, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[278, 380], [243, 380], [194, 408], [173, 465]]</point>"}]}
{"id": 1940, "image": ["rtx_frames_success_10/49_bridge#episode_21373/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21386/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush across the table."}, {"from": "gpt", "value": "<point>[[415, 272], [504, 272], [577, 300], [662, 418]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lemon."}, {"from": "gpt", "value": "<point>[[440, 385], [493, 423], [528, 437], [585, 432]]</point>"}]}
{"id": 1942, "image": ["rtx_frames_success_10/49_bridge#episode_21400/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21416/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[518, 338], [412, 310], [324, 319], [232, 357]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[320, 113], [398, 150], [444, 211], [447, 310]]</point>"}]}
{"id": 1944, "image": ["rtx_frames_success_10/49_bridge#episode_21416/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21423/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the potato towards the lower left corner."}, {"from": "gpt", "value": "<point>[[187, 300], [162, 362], [134, 413], [144, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[553, 465], [451, 441], [356, 446], [285, 484]]</point>"}]}
{"id": 1946, "image": ["rtx_frames_success_10/49_bridge#episode_21439/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21443/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spice container."}, {"from": "gpt", "value": "<point>[[187, 254], [250, 343], [310, 385], [426, 404]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[845, 310], [877, 376], [891, 432], [863, 568]]</point>"}]}
{"id": 1948, "image": ["rtx_frames_success_10/49_bridge#episode_21443/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21454/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the potato over the pot."}, {"from": "gpt", "value": "<point>[[648, 291], [595, 319], [539, 376], [518, 423]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red spoon."}, {"from": "gpt", "value": "<point>[[296, 188], [356, 376], [511, 493], [708, 624]]</point>"}]}
{"id": 1950, "image": ["rtx_frames_success_10/49_bridge#episode_21458/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21458/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bell pepper."}, {"from": "gpt", "value": "<point>[[546, 272], [518, 272], [489, 291], [472, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bell pepper over the pot."}, {"from": "gpt", "value": "<point>[[514, 319], [577, 404], [592, 469], [592, 554]]</point>"}]}
{"id": 1952, "image": ["rtx_frames_success_10/49_bridge#episode_21476/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21491/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the meat."}, {"from": "gpt", "value": "<point>[[257, 300], [349, 385], [408, 394], [532, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[359, 141], [468, 268], [532, 423], [609, 685]]</point>"}]}
{"id": 1954, "image": ["rtx_frames_success_10/49_bridge#episode_21507/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21509/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[398, 329], [419, 498], [489, 629], [637, 770]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the blue spoon."}, {"from": "gpt", "value": "<point>[[331, 291], [486, 376], [574, 493], [658, 667]]</point>"}]}
{"id": 1956, "image": ["rtx_frames_success_10/49_bridge#episode_21517/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21517/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[546, 286], [486, 244], [426, 244], [370, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[377, 235], [472, 300], [514, 385], [535, 479]]</point>"}]}
{"id": 1958, "image": ["rtx_frames_success_10/49_bridge#episode_21518/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21518/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[570, 455], [500, 376], [412, 347], [324, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spatula to the right side of the stove."}, {"from": "gpt", "value": "<point>[[303, 366], [394, 366], [486, 394], [592, 460]]</point>"}]}
{"id": 1960, "image": ["rtx_frames_success_10/49_bridge#episode_21547/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21552/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot containing the corn."}, {"from": "gpt", "value": "<point>[[398, 516], [496, 592], [602, 592], [708, 559]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the grapes."}, {"from": "gpt", "value": "<point>[[602, 277], [440, 268], [285, 310], [197, 394]]</point>"}]}
{"id": 1962, "image": ["rtx_frames_success_10/49_bridge#episode_21564/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21581/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spice jar."}, {"from": "gpt", "value": "<point>[[127, 272], [215, 211], [292, 216], [387, 235]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the jelly."}, {"from": "gpt", "value": "<point>[[426, 225], [370, 347], [366, 446], [391, 549]]</point>"}]}
{"id": 1964, "image": ["rtx_frames_success_10/49_bridge#episode_21587/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21589/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[447, 305], [317, 268], [229, 286], [109, 357]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bread."}, {"from": "gpt", "value": "<point>[[306, 305], [391, 362], [412, 413], [426, 568]]</point>"}]}
{"id": 1966, "image": ["rtx_frames_success_10/49_bridge#episode_21590/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21605/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the container."}, {"from": "gpt", "value": "<point>[[581, 178], [475, 235], [433, 362], [419, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[408, 512], [479, 469], [539, 455], [609, 460]]</point>"}]}
{"id": 1968, "image": ["rtx_frames_success_10/49_bridge#episode_21609/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21609/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the mushroom."}, {"from": "gpt", "value": "<point>[[261, 526], [356, 516], [433, 554], [535, 667]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[630, 638], [546, 582], [440, 549], [317, 545]]</point>"}]}
{"id": 1970, "image": ["rtx_frames_success_10/49_bridge#episode_21623/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21623/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[426, 211], [377, 207], [342, 225], [271, 291]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the hot dog and towel."}, {"from": "gpt", "value": "<point>[[373, 380], [430, 469], [444, 545], [458, 634]]</point>"}]}
{"id": 1972, "image": ["rtx_frames_success_10/49_bridge#episode_21643/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21648/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow cloth."}, {"from": "gpt", "value": "<point>[[567, 103], [507, 160], [475, 216], [451, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the stuffed toy."}, {"from": "gpt", "value": "<point>[[518, 197], [408, 207], [342, 239], [285, 338]]</point>"}]}
{"id": 1974, "image": ["rtx_frames_success_10/49_bridge#episode_21649/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21649/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue spoon."}, {"from": "gpt", "value": "<point>[[342, 225], [246, 272], [180, 319], [144, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the blue spoon towards the yellow cloth."}, {"from": "gpt", "value": "<point>[[165, 418], [225, 352], [285, 333], [373, 333]]</point>"}]}
{"id": 1976, "image": ["rtx_frames_success_10/49_bridge#episode_21661/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21662/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[278, 268], [299, 347], [384, 385], [479, 385]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice container."}, {"from": "gpt", "value": "<point>[[500, 272], [528, 366], [549, 488], [525, 601]]</point>"}]}
{"id": 1978, "image": ["rtx_frames_success_10/49_bridge#episode_21666/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21666/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[514, 460], [408, 390], [327, 376], [239, 385]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon."}, {"from": "gpt", "value": "<point>[[366, 376], [415, 418], [451, 484], [444, 545]]</point>"}]}
{"id": 1980, "image": ["rtx_frames_success_10/49_bridge#episode_21710/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21718/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[299, 192], [285, 268], [306, 338], [370, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[349, 333], [335, 394], [342, 451], [380, 512]]</point>"}]}
{"id": 1982, "image": ["rtx_frames_success_10/49_bridge#episode_21730/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21745/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[335, 347], [380, 366], [412, 399], [440, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[482, 94], [415, 150], [398, 225], [391, 352]]</point>"}]}
{"id": 1984, "image": ["rtx_frames_success_10/49_bridge#episode_21745/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21753/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife towards the cloth."}, {"from": "gpt", "value": "<point>[[761, 319], [778, 451], [768, 563], [729, 714]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the corn from the bowl."}, {"from": "gpt", "value": "<point>[[243, 423], [282, 347], [335, 291], [401, 310]]</point>"}]}
{"id": 1986, "image": ["rtx_frames_success_10/49_bridge#episode_21753/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21776/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the corn above the stove."}, {"from": "gpt", "value": "<point>[[535, 207], [585, 268], [599, 357], [599, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave door."}, {"from": "gpt", "value": "<point>[[387, 418], [461, 460], [489, 516], [514, 592]]</point>"}]}
{"id": 1988, "image": ["rtx_frames_success_10/49_bridge#episode_21779/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21779/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the green vegetable from the bowl."}, {"from": "gpt", "value": "<point>[[581, 113], [465, 131], [408, 188], [370, 305]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the green vegetable towards the right bottom of the table."}, {"from": "gpt", "value": "<point>[[387, 235], [504, 282], [592, 376], [665, 531]]</point>"}]}
{"id": 1990, "image": ["rtx_frames_success_10/49_bridge#episode_21781/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21781/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue fork."}, {"from": "gpt", "value": "<point>[[570, 197], [468, 277], [391, 432], [335, 559]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the blue fork towards the top right corner, above the yellow cloth."}, {"from": "gpt", "value": "<point>[[563, 329], [644, 390], [683, 451], [732, 554]]</point>"}]}
{"id": 1992, "image": ["rtx_frames_success_10/49_bridge#episode_21808/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21842/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[232, 207], [331, 300], [447, 333], [560, 300]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[218, 192], [250, 366], [352, 549], [570, 685]]</point>"}]}
{"id": 1994, "image": ["rtx_frames_success_10/49_bridge#episode_21842/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21846/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[570, 648], [535, 559], [482, 479], [415, 437]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the plate."}, {"from": "gpt", "value": "<point>[[743, 540], [694, 502], [651, 498], [556, 488]]</point>"}]}
{"id": 1996, "image": ["rtx_frames_success_10/49_bridge#episode_21846/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21862/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pear on the plate."}, {"from": "gpt", "value": "<point>[[637, 357], [592, 286], [542, 272], [461, 244]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the orange block."}, {"from": "gpt", "value": "<point>[[423, 352], [384, 329], [331, 329], [285, 366]]</point>"}]}
{"id": 1998, "image": ["rtx_frames_success_10/49_bridge#episode_21880/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21896/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[303, 131], [331, 249], [468, 319], [616, 357]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lemon."}, {"from": "gpt", "value": "<point>[[222, 225], [437, 319], [637, 455], [782, 690]]</point>"}]}
{"id": 2000, "image": ["rtx_frames_success_10/49_bridge#episode_21899/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21932/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red cup."}, {"from": "gpt", "value": "<point>[[352, 282], [292, 300], [254, 347], [225, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[518, 239], [377, 291], [278, 347], [208, 498]]</point>"}]}
{"id": 2002, "image": ["rtx_frames_success_10/49_bridge#episode_21933/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21938/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the cloth."}, {"from": "gpt", "value": "<point>[[567, 296], [475, 413], [451, 512], [472, 681]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[542, 305], [482, 362], [451, 432], [426, 507]]</point>"}]}
{"id": 2004, "image": ["rtx_frames_success_10/49_bridge#episode_21944/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21945/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[722, 333], [560, 329], [405, 376], [225, 469]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[306, 103], [387, 254], [465, 329], [588, 343]]</point>"}]}
{"id": 2006, "image": ["rtx_frames_success_10/49_bridge#episode_21946/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21952/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cleaning brush."}, {"from": "gpt", "value": "<point>[[602, 413], [627, 469], [634, 526], [623, 620]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the yellow knife."}, {"from": "gpt", "value": "<point>[[408, 141], [356, 174], [313, 221], [278, 291]]</point>"}]}
{"id": 2008, "image": ["rtx_frames_success_10/49_bridge#episode_21952/frame_15.png", "rtx_frames_success_10/49_bridge#episode_21953/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the knife down."}, {"from": "gpt", "value": "<point>[[239, 404], [366, 413], [532, 465], [637, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the container."}, {"from": "gpt", "value": "<point>[[370, 211], [384, 300], [426, 366], [511, 385]]</point>"}]}
{"id": 2010, "image": ["rtx_frames_success_10/49_bridge#episode_21971/frame_0.png", "rtx_frames_success_10/49_bridge#episode_21972/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[405, 451], [482, 343], [574, 291], [725, 268]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pan containing the broccoli."}, {"from": "gpt", "value": "<point>[[493, 394], [451, 437], [437, 507], [440, 629]]</point>"}]}
{"id": 2012, "image": ["rtx_frames_success_10/49_bridge#episode_21989/frame_0.png", "rtx_frames_success_10/49_bridge#episode_22003/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[701, 197], [651, 338], [655, 507], [708, 648]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana in the colander."}, {"from": "gpt", "value": "<point>[[634, 282], [532, 404], [472, 554], [475, 770]]</point>"}]}
{"id": 2014, "image": ["rtx_frames_success_10/49_bridge#episode_22011/frame_0.png", "rtx_frames_success_10/49_bridge#episode_22017/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[197, 498], [285, 413], [440, 371], [658, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[419, 338], [447, 451], [521, 507], [641, 540]]</point>"}]}
{"id": 2016, "image": ["rtx_frames_success_10/49_bridge#episode_22025/frame_0.png", "rtx_frames_success_10/49_bridge#episode_22053/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the pot on the stove."}, {"from": "gpt", "value": "<point>[[644, 413], [592, 376], [556, 376], [496, 404]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the blue towel."}, {"from": "gpt", "value": "<point>[[606, 75], [644, 150], [662, 239], [637, 347]]</point>"}]}
{"id": 2018, "image": ["rtx_frames_success_10/49_bridge#episode_22057/frame_0.png", "rtx_frames_success_10/49_bridge#episode_22089/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the mushroom."}, {"from": "gpt", "value": "<point>[[518, 94], [468, 230], [454, 376], [486, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[331, 197], [370, 300], [447, 376], [577, 404]]</point>"}]}
{"id": 2020, "image": ["rtx_frames_success_10/49_bridge#episode_22100/frame_0.png", "rtx_frames_success_10/49_bridge#episode_22104/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[437, 164], [581, 291], [673, 441], [694, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[539, 300], [465, 286], [391, 319], [345, 366]]</point>"}]}
{"id": 2022, "image": ["rtx_frames_success_10/49_bridge#episode_22107/frame_0.png", "rtx_frames_success_10/49_bridge#episode_22107/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[539, 469], [440, 432], [320, 432], [243, 469]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot to the back."}, {"from": "gpt", "value": "<point>[[278, 460], [415, 366], [542, 333], [676, 343]]</point>"}]}
{"id": 2024, "image": ["rtx_frames_success_10/49_bridge#episode_22125/frame_0.png", "rtx_frames_success_10/49_bridge#episode_22147/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal bowl containing the ear of corn."}, {"from": "gpt", "value": "<point>[[415, 239], [317, 272], [278, 329], [236, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[197, 315], [370, 488], [567, 568], [736, 601]]</point>"}]}
{"id": 2026, "image": ["rtx_frames_success_10/49_bridge#episode_22147/frame_15.png", "rtx_frames_success_10/49_bridge#episode_22167/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[521, 418], [430, 390], [310, 390], [176, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[444, 277], [525, 333], [623, 366], [754, 371]]</point>"}]}
{"id": 2028, "image": ["rtx_frames_success_10/49_bridge#episode_22173/frame_0.png", "rtx_frames_success_10/49_bridge#episode_22211/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[549, 338], [658, 390], [704, 432], [739, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[401, 178], [352, 258], [320, 347], [320, 516]]</point>"}]}
{"id": 2030, "image": ["rtx_frames_success_10/49_bridge#episode_22211/frame_15.png", "rtx_frames_success_11/49_bridge#episode_230/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife to the right."}, {"from": "gpt", "value": "<point>[[232, 404], [331, 479], [458, 554], [651, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[356, 135], [411, 212], [475, 324], [544, 440], [643, 618]]</point>"}]}
{"id": 2032, "image": ["rtx_frames_success_11/49_bridge#episode_233/frame_0.png", "rtx_frames_success_11/49_bridge#episode_234/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[353, 77], [376, 197], [394, 290], [408, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red object."}, {"from": "gpt", "value": "<point>[[350, 259], [521, 293], [582, 344], [649, 421]]</point>"}]}
{"id": 2034, "image": ["rtx_frames_success_11/49_bridge#episode_2226/frame_0.png", "rtx_frames_success_11/49_bridge#episode_2243/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[188, 351], [298, 328], [376, 351], [458, 363]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the knife."}, {"from": "gpt", "value": "<point>[[264, 185], [319, 278], [391, 386], [466, 514]]</point>"}]}
{"id": 2036, "image": ["rtx_frames_success_11/49_bridge#episode_2250/frame_0.png", "rtx_frames_success_11/49_bridge#episode_2256/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green towel."}, {"from": "gpt", "value": "<point>[[417, 120], [374, 181], [307, 278], [220, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[426, 290], [359, 297], [313, 332], [243, 398]]</point>"}]}
{"id": 2038, "image": ["rtx_frames_success_11/49_bridge#episode_2258/frame_0.png", "rtx_frames_success_11/49_bridge#episode_2284/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the towel."}, {"from": "gpt", "value": "<point>[[449, 116], [423, 212], [403, 320], [391, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom towards the pot."}, {"from": "gpt", "value": "<point>[[570, 193], [547, 274], [524, 371], [492, 533]]</point>"}]}
{"id": 2040, "image": ["rtx_frames_success_11/49_bridge#episode_2296/frame_0.png", "rtx_frames_success_11/49_bridge#episode_2301/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pepper."}, {"from": "gpt", "value": "<point>[[347, 120], [321, 205], [327, 301]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[530, 703], [597, 575], [666, 471], [796, 317]]</point>"}]}
{"id": 2042, "image": ["rtx_frames_success_11/49_bridge#episode_2315/frame_0.png", "rtx_frames_success_11/49_bridge#episode_2334/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[194, 251], [321, 274], [414, 336], [472, 398]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[290, 351], [403, 417], [504, 486], [588, 579]]</point>"}]}
{"id": 2044, "image": ["rtx_frames_success_11/49_bridge#episode_2337/frame_0.png", "rtx_frames_success_11/49_bridge#episode_2346/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[182, 154], [272, 201], [353, 297], [452, 402], [539, 490], [614, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[197, 290], [319, 266], [446, 259], [539, 274], [602, 290]]</point>"}]}
{"id": 2046, "image": ["rtx_frames_success_11/49_bridge#episode_2346/frame_15.png", "rtx_frames_success_11/49_bridge#episode_2363/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cucumber."}, {"from": "gpt", "value": "<point>[[640, 274], [588, 232], [533, 193], [446, 162], [353, 189]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red cloth."}, {"from": "gpt", "value": "<point>[[796, 386], [707, 425], [631, 479], [582, 541], [562, 595]]</point>"}]}
{"id": 2048, "image": ["rtx_frames_success_11/49_bridge#episode_22250/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22269/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow knife."}, {"from": "gpt", "value": "<point>[[356, 205], [394, 278], [420, 359], [434, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[449, 278], [426, 398], [414, 568]]</point>"}]}
{"id": 2050, "image": ["rtx_frames_success_11/49_bridge#episode_22269/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22277/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife towards the cutting board."}, {"from": "gpt", "value": "<point>[[405, 552], [498, 517], [579, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pear."}, {"from": "gpt", "value": "<point>[[188, 386], [281, 378], [397, 402], [507, 432], [634, 475], [782, 568]]</point>"}]}
{"id": 2052, "image": ["rtx_frames_success_11/49_bridge#episode_22306/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22315/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[324, 73], [388, 174], [443, 266], [495, 347], [570, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[437, 216], [437, 320], [434, 444], [440, 571], [440, 707]]</point>"}]}
{"id": 2054, "image": ["rtx_frames_success_11/49_bridge#episode_22316/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22321/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[489, 139], [434, 259], [365, 363], [313, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the scoop."}, {"from": "gpt", "value": "<point>[[295, 170], [359, 239], [408, 320]]</point>"}]}
{"id": 2056, "image": ["rtx_frames_success_11/49_bridge#episode_22322/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22334/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[385, 266], [486, 355], [585, 448], [660, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow pepper."}, {"from": "gpt", "value": "<point>[[417, 46], [420, 108], [426, 174], [443, 278]]</point>"}]}
{"id": 2058, "image": ["rtx_frames_success_11/49_bridge#episode_22335/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22339/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[365, 278], [403, 394], [452, 595]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[759, 324], [712, 347], [625, 432]]</point>"}]}
{"id": 2060, "image": ["rtx_frames_success_11/49_bridge#episode_22347/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22357/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cleaning tool on the table."}, {"from": "gpt", "value": "<point>[[405, 89], [475, 151], [536, 251], [588, 398]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[188, 359], [261, 382], [365, 390], [486, 378]]</point>"}]}
{"id": 2062, "image": ["rtx_frames_success_11/49_bridge#episode_22376/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22376/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave door handle."}, {"from": "gpt", "value": "<point>[[599, 598], [550, 533], [452, 429], [376, 359], [292, 305], [188, 251]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the microwave door closed."}, {"from": "gpt", "value": "<point>[[429, 382], [553, 328], [649, 297], [721, 263]]</point>"}]}
{"id": 2064, "image": ["rtx_frames_success_11/49_bridge#episode_22384/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22402/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[313, 112], [411, 290], [495, 490], [559, 749]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[203, 429], [290, 375], [347, 317]]</point>"}]}
{"id": 2066, "image": ["rtx_frames_success_11/49_bridge#episode_22403/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22408/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to open the microwave door."}, {"from": "gpt", "value": "<point>[[449, 328], [376, 382], [292, 425], [240, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[715, 116], [588, 124], [486, 166], [408, 205], [266, 301]]</point>"}]}
{"id": 2068, "image": ["rtx_frames_success_11/49_bridge#episode_22408/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22413/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth between the pot and the spoon."}, {"from": "gpt", "value": "<point>[[301, 274], [391, 367], [463, 483], [513, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[417, 286], [319, 324], [214, 378], [110, 456], [58, 502]]</point>"}]}
{"id": 2070, "image": ["rtx_frames_success_11/49_bridge#episode_22413/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22427/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[58, 521], [133, 529], [252, 548], [339, 575], [446, 614]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[582, 429], [547, 405], [466, 367], [345, 336]]</point>"}]}
{"id": 2072, "image": ["rtx_frames_success_11/49_bridge#episode_22438/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22442/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green spatula."}, {"from": "gpt", "value": "<point>[[252, 324], [394, 340], [539, 375], [695, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[264, 143], [261, 224], [278, 351]]</point>"}]}
{"id": 2074, "image": ["rtx_frames_success_11/49_bridge#episode_22451/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22456/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[507, 278], [568, 371], [620, 444], [654, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[692, 216], [623, 336], [585, 471], [553, 622]]</point>"}]}
{"id": 2076, "image": ["rtx_frames_success_11/49_bridge#episode_22465/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22476/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[501, 286], [443, 320], [333, 386], [269, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cube."}, {"from": "gpt", "value": "<point>[[417, 112], [362, 170], [275, 293], [177, 405]]</point>"}]}
{"id": 2078, "image": ["rtx_frames_success_11/49_bridge#episode_22480/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22482/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[101, 471], [252, 429], [356, 386], [394, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[617, 247], [576, 255], [429, 301], [310, 347]]</point>"}]}
{"id": 2080, "image": ["rtx_frames_success_11/49_bridge#episode_22486/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22489/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[180, 243], [194, 375], [200, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[698, 386], [568, 386], [452, 425], [298, 486]]</point>"}]}
{"id": 2082, "image": ["rtx_frames_success_11/49_bridge#episode_22489/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22493/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[284, 405], [400, 402], [510, 402], [573, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[292, 417], [426, 398], [599, 382], [762, 421]]</point>"}]}
{"id": 2084, "image": ["rtx_frames_success_11/49_bridge#episode_22494/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22500/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cheese wedge."}, {"from": "gpt", "value": "<point>[[319, 205], [423, 239], [507, 293], [608, 363]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[206, 201], [298, 293], [374, 378], [466, 502], [556, 695]]</point>"}]}
{"id": 2086, "image": ["rtx_frames_success_11/49_bridge#episode_22500/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22502/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the microwave."}, {"from": "gpt", "value": "<point>[[559, 548], [434, 413], [362, 317], [321, 255]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the sweetpotato."}, {"from": "gpt", "value": "<point>[[669, 332], [637, 417], [608, 537]]</point>"}]}
{"id": 2088, "image": ["rtx_frames_success_11/49_bridge#episode_22502/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22504/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up a stuffed elephant."}, {"from": "gpt", "value": "<point>[[819, 193], [727, 255], [602, 371], [458, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[507, 320], [431, 313], [382, 328], [307, 378]]</point>"}]}
{"id": 2090, "image": ["rtx_frames_success_11/49_bridge#episode_22511/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22523/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[17, 290], [133, 290], [220, 290], [368, 320], [478, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[359, 201], [449, 317], [568, 452], [701, 610]]</point>"}]}
{"id": 2092, "image": ["rtx_frames_success_11/49_bridge#episode_22524/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22541/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the drumstick."}, {"from": "gpt", "value": "<point>[[533, 232], [414, 344], [353, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[550, 205], [527, 278], [481, 417], [452, 564]]</point>"}]}
{"id": 2094, "image": ["rtx_frames_success_11/49_bridge#episode_22542/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22549/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[440, 181], [397, 224], [330, 324], [304, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[319, 143], [374, 270], [452, 429], [565, 641]]</point>"}]}
{"id": 2096, "image": ["rtx_frames_success_11/49_bridge#episode_22551/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22567/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom in the bowl."}, {"from": "gpt", "value": "<point>[[675, 54], [559, 154], [484, 232], [394, 324], [235, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the steel pan."}, {"from": "gpt", "value": "<point>[[284, 212], [379, 282], [495, 355], [573, 390], [649, 444]]</point>"}]}
{"id": 2098, "image": ["rtx_frames_success_11/49_bridge#episode_22569/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22575/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[206, 93], [310, 174], [446, 297], [562, 432], [582, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[484, 355], [475, 448], [475, 583], [481, 699]]</point>"}]}
{"id": 2100, "image": ["rtx_frames_success_11/49_bridge#episode_22579/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22580/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the bowl."}, {"from": "gpt", "value": "<point>[[510, 382], [594, 402], [672, 475], [721, 544]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the vessel."}, {"from": "gpt", "value": "<point>[[417, 147], [458, 297], [481, 483]]</point>"}]}
{"id": 2102, "image": ["rtx_frames_success_11/49_bridge#episode_22589/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22605/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[484, 290], [507, 440], [547, 548], [597, 703]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[243, 263], [385, 270], [507, 293], [623, 324]]</point>"}]}
{"id": 2104, "image": ["rtx_frames_success_11/49_bridge#episode_22633/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22649/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[177, 301], [316, 309], [388, 340], [510, 375]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue spoon."}, {"from": "gpt", "value": "<point>[[394, 131], [319, 181], [264, 274], [214, 344], [188, 425]]</point>"}]}
{"id": 2106, "image": ["rtx_frames_success_11/49_bridge#episode_22652/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22687/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[634, 112], [643, 228], [660, 390], [675, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[142, 266], [185, 332], [226, 402], [261, 448]]</point>"}]}
{"id": 2108, "image": ["rtx_frames_success_11/49_bridge#episode_22691/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22706/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[405, 181], [460, 239], [515, 324], [591, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[368, 270], [463, 332], [495, 378], [513, 440]]</point>"}]}
{"id": 2110, "image": ["rtx_frames_success_11/49_bridge#episode_22709/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22717/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the towel."}, {"from": "gpt", "value": "<point>[[533, 178], [414, 274], [362, 363], [330, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the object from the cloth."}, {"from": "gpt", "value": "<point>[[466, 162], [527, 224], [597, 309], [669, 413], [738, 483]]</point>"}]}
{"id": 2112, "image": ["rtx_frames_success_11/49_bridge#episode_22722/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22727/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cheese."}, {"from": "gpt", "value": "<point>[[359, 154], [478, 239], [553, 359], [620, 502], [724, 695]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[275, 185], [431, 255], [530, 293], [637, 367], [712, 432]]</point>"}]}
{"id": 2114, "image": ["rtx_frames_success_11/49_bridge#episode_22732/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22737/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[481, 174], [507, 324], [533, 467], [556, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the yellow cloth to the right corner."}, {"from": "gpt", "value": "<point>[[319, 502], [469, 552], [559, 598], [654, 656]]</point>"}]}
{"id": 2116, "image": ["rtx_frames_success_11/49_bridge#episode_22745/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22750/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the blue fork."}, {"from": "gpt", "value": "<point>[[582, 425], [472, 367], [356, 336], [310, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the clothes."}, {"from": "gpt", "value": "<point>[[617, 699], [524, 637], [429, 633], [347, 622]]</point>"}]}
{"id": 2118, "image": ["rtx_frames_success_11/49_bridge#episode_22750/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22759/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move a piece of clothing to the washing machine."}, {"from": "gpt", "value": "<point>[[356, 637], [434, 544], [498, 463], [597, 278]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice bottle."}, {"from": "gpt", "value": "<point>[[281, 143], [356, 178], [426, 247], [530, 355]]</point>"}]}
{"id": 2120, "image": ["rtx_frames_success_11/49_bridge#episode_22766/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22768/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spice container."}, {"from": "gpt", "value": "<point>[[542, 158], [484, 239], [443, 317], [417, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull open the refrigerator door."}, {"from": "gpt", "value": "<point>[[605, 629], [559, 660], [513, 699], [397, 761]]</point>"}]}
{"id": 2122, "image": ["rtx_frames_success_11/49_bridge#episode_22769/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22797/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the block."}, {"from": "gpt", "value": "<point>[[156, 162], [197, 239], [217, 305], [232, 355]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl on the green towel."}, {"from": "gpt", "value": "<point>[[481, 301], [414, 278], [356, 297], [313, 328]]</point>"}]}
{"id": 2124, "image": ["rtx_frames_success_11/49_bridge#episode_22814/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22843/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[788, 363], [649, 336], [576, 340], [420, 355]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[206, 85], [266, 185], [333, 263], [385, 324]]</point>"}]}
{"id": 2126, "image": ["rtx_frames_success_11/49_bridge#episode_22847/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22854/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cucumber in the orange bowl."}, {"from": "gpt", "value": "<point>[[159, 371], [290, 432], [362, 475], [426, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the knife on the cloth."}, {"from": "gpt", "value": "<point>[[469, 328], [362, 344], [269, 394], [223, 452]]</point>"}]}
{"id": 2128, "image": ["rtx_frames_success_11/49_bridge#episode_22861/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22869/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[507, 502], [539, 568], [582, 691], [591, 842]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[484, 185], [559, 251], [608, 305], [707, 436]]</point>"}]}
{"id": 2130, "image": ["rtx_frames_success_11/49_bridge#episode_22920/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22927/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[365, 143], [420, 216], [481, 355], [530, 456], [611, 653]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[504, 39], [475, 178], [443, 293], [417, 429]]</point>"}]}
{"id": 2132, "image": ["rtx_frames_success_11/49_bridge#episode_22930/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22945/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the meat."}, {"from": "gpt", "value": "<point>[[513, 208], [458, 232], [368, 324], [310, 375]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green pepper."}, {"from": "gpt", "value": "<point>[[220, 425], [336, 402], [478, 359], [675, 317]]</point>"}]}
{"id": 2134, "image": ["rtx_frames_success_11/49_bridge#episode_22949/frame_0.png", "rtx_frames_success_11/49_bridge#episode_22961/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[489, 208], [446, 251], [359, 297], [272, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drumstick."}, {"from": "gpt", "value": "<point>[[411, 324], [394, 394], [374, 483]]</point>"}]}
{"id": 2136, "image": ["rtx_frames_success_11/49_bridge#episode_22993/frame_15.png", "rtx_frames_success_11/49_bridge#episode_22998/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the bowl."}, {"from": "gpt", "value": "<point>[[217, 336], [359, 371], [492, 398], [631, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[718, 266], [637, 301], [547, 371], [472, 448], [397, 517]]</point>"}]}
{"id": 2138, "image": ["rtx_frames_success_11/49_bridge#episode_23017/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23017/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[469, 181], [417, 274], [382, 332], [313, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spoon down."}, {"from": "gpt", "value": "<point>[[414, 452], [542, 560], [591, 625], [637, 680]]</point>"}]}
{"id": 2140, "image": ["rtx_frames_success_11/49_bridge#episode_23026/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23026/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[310, 429], [469, 417], [585, 444], [770, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl on the blue towel."}, {"from": "gpt", "value": "<point>[[301, 328], [356, 444], [371, 483]]</point>"}]}
{"id": 2142, "image": ["rtx_frames_success_11/49_bridge#episode_23039/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23045/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the blue arch."}, {"from": "gpt", "value": "<point>[[429, 39], [365, 124], [301, 259], [237, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the knife on the yellow towel."}, {"from": "gpt", "value": "<point>[[148, 174], [261, 205], [359, 278], [446, 355]]</point>"}]}
{"id": 2144, "image": ["rtx_frames_success_11/49_bridge#episode_23059/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23066/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the knief on the orange towel."}, {"from": "gpt", "value": "<point>[[130, 259], [272, 355], [374, 440], [530, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato in the pan."}, {"from": "gpt", "value": "<point>[[391, 355], [542, 355], [654, 375], [730, 413]]</point>"}]}
{"id": 2146, "image": ["rtx_frames_success_11/49_bridge#episode_23071/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23071/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[507, 143], [599, 197], [666, 309], [727, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to close the microwave door."}, {"from": "gpt", "value": "<point>[[628, 568], [582, 440], [530, 332], [472, 201]]</point>"}]}
{"id": 2148, "image": ["rtx_frames_success_11/49_bridge#episode_23079/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23084/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the strawberry."}, {"from": "gpt", "value": "<point>[[455, 328], [376, 336], [272, 355], [240, 378]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the mushroom in the pot."}, {"from": "gpt", "value": "<point>[[434, 189], [472, 293], [527, 429], [597, 660]]</point>"}]}
{"id": 2150, "image": ["rtx_frames_success_11/49_bridge#episode_23099/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23101/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[550, 313], [440, 355], [301, 386], [194, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[125, 456], [266, 510], [397, 571], [573, 641]]</point>"}]}
{"id": 2152, "image": ["rtx_frames_success_11/49_bridge#episode_23141/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23142/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[191, 232], [284, 317], [382, 394], [466, 479], [559, 598], [625, 683]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[640, 197], [672, 293], [698, 425], [718, 525]]</point>"}]}
{"id": 2154, "image": ["rtx_frames_success_11/49_bridge#episode_23151/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23155/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[498, 224], [458, 266], [405, 359], [368, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[385, 193], [362, 290], [342, 409], [345, 587]]</point>"}]}
{"id": 2156, "image": ["rtx_frames_success_11/49_bridge#episode_23161/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23165/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to open the microwave door."}, {"from": "gpt", "value": "<point>[[573, 448], [472, 525], [382, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue spoon."}, {"from": "gpt", "value": "<point>[[559, 205], [608, 270], [652, 336], [698, 432]]</point>"}]}
{"id": 2158, "image": ["rtx_frames_success_11/49_bridge#episode_23170/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23171/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the banana next to the bottle."}, {"from": "gpt", "value": "<point>[[278, 208], [333, 363], [374, 517], [426, 649]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the cube."}, {"from": "gpt", "value": "<point>[[559, 255], [553, 340], [576, 432]]</point>"}]}
{"id": 2160, "image": ["rtx_frames_success_11/49_bridge#episode_23171/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23179/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the tomato."}, {"from": "gpt", "value": "<point>[[264, 174], [278, 243], [290, 305], [298, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[324, 239], [408, 317], [458, 413], [504, 544]]</point>"}]}
{"id": 2162, "image": ["rtx_frames_success_11/49_bridge#episode_23183/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23203/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cashew container."}, {"from": "gpt", "value": "<point>[[764, 324], [712, 402], [643, 471], [576, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[261, 371], [379, 390], [513, 413], [669, 452]]</point>"}]}
{"id": 2164, "image": ["rtx_frames_success_11/49_bridge#episode_23223/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23231/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[333, 224], [379, 359], [437, 510], [501, 668]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[611, 93], [646, 197], [672, 336]]</point>"}]}
{"id": 2166, "image": ["rtx_frames_success_11/49_bridge#episode_23232/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23233/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[313, 81], [388, 232], [466, 402], [495, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[324, 147], [295, 278], [290, 390], [264, 475]]</point>"}]}
{"id": 2168, "image": ["rtx_frames_success_11/49_bridge#episode_23271/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23272/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[620, 185], [495, 193], [362, 208], [266, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple object."}, {"from": "gpt", "value": "<point>[[180, 270], [327, 290], [524, 328], [657, 382], [770, 456]]</point>"}]}
{"id": 2170, "image": ["rtx_frames_success_11/49_bridge#episode_23285/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23293/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the apple."}, {"from": "gpt", "value": "<point>[[168, 359], [281, 363], [368, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[405, 336], [403, 417], [394, 537]]</point>"}]}
{"id": 2172, "image": ["rtx_frames_success_11/49_bridge#episode_23309/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23327/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[220, 201], [301, 290], [374, 402], [434, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red spoon on the board."}, {"from": "gpt", "value": "<point>[[582, 62], [628, 232], [663, 371], [683, 502], [707, 672]]</point>"}]}
{"id": 2174, "image": ["rtx_frames_success_11/49_bridge#episode_23338/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23345/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the cloth and knife."}, {"from": "gpt", "value": "<point>[[330, 263], [460, 340], [536, 432], [568, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[466, 344], [368, 413], [295, 486], [237, 571]]</point>"}]}
{"id": 2176, "image": ["rtx_frames_success_11/49_bridge#episode_23350/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23427/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the bowl."}, {"from": "gpt", "value": "<point>[[371, 178], [452, 239], [565, 320], [625, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up a plate."}, {"from": "gpt", "value": "<point>[[562, 483], [501, 486], [403, 506], [255, 587]]</point>"}]}
{"id": 2178, "image": ["rtx_frames_success_11/49_bridge#episode_23427/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23445/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the carrot on the plate."}, {"from": "gpt", "value": "<point>[[258, 583], [362, 502], [431, 459], [533, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[420, 50], [469, 154], [521, 266], [559, 375], [579, 506]]</point>"}]}
{"id": 2180, "image": ["rtx_frames_success_11/49_bridge#episode_23446/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23461/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the knife."}, {"from": "gpt", "value": "<point>[[527, 622], [475, 517], [431, 467]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[411, 336], [501, 351], [559, 367], [620, 398]]</point>"}]}
{"id": 2182, "image": ["rtx_frames_success_11/49_bridge#episode_23461/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23463/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the pan."}, {"from": "gpt", "value": "<point>[[408, 236], [316, 286], [255, 332], [171, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[310, 517], [382, 421], [429, 398], [498, 371]]</point>"}]}
{"id": 2184, "image": ["rtx_frames_success_11/49_bridge#episode_23478/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23483/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[495, 320], [599, 471], [672, 560], [727, 695]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[119, 224], [208, 236], [330, 317], [420, 378], [498, 432]]</point>"}]}
{"id": 2186, "image": ["rtx_frames_success_11/49_bridge#episode_23484/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23487/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the item."}, {"from": "gpt", "value": "<point>[[281, 259], [394, 297], [513, 347], [620, 386], [704, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bowl."}, {"from": "gpt", "value": "<point>[[582, 340], [495, 371], [382, 405], [298, 425], [240, 440]]</point>"}]}
{"id": 2188, "image": ["rtx_frames_success_11/49_bridge#episode_23487/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23524/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bowl towards the spatula."}, {"from": "gpt", "value": "<point>[[246, 471], [356, 463], [504, 494], [594, 575], [663, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[791, 432], [620, 425], [498, 417], [379, 413], [307, 413], [235, 417]]</point>"}]}
{"id": 2190, "image": ["rtx_frames_success_11/49_bridge#episode_23537/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23541/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[660, 208], [614, 208], [556, 239], [518, 270], [458, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[379, 23], [368, 139], [382, 332], [394, 429]]</point>"}]}
{"id": 2192, "image": ["rtx_frames_success_11/49_bridge#episode_23586/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23595/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spice container."}, {"from": "gpt", "value": "<point>[[180, 251], [243, 255], [327, 259], [403, 278]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[489, 208], [504, 320], [542, 402], [576, 494]]</point>"}]}
{"id": 2194, "image": ["rtx_frames_success_11/49_bridge#episode_23604/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23622/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow knife."}, {"from": "gpt", "value": "<point>[[197, 143], [226, 251], [246, 336], [272, 444], [295, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[631, 309], [565, 359], [486, 421], [446, 475]]</point>"}]}
{"id": 2196, "image": ["rtx_frames_success_11/49_bridge#episode_23625/frame_15.png", "rtx_frames_success_11/49_bridge#episode_23626/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the rag."}, {"from": "gpt", "value": "<point>[[307, 324], [385, 371], [475, 417], [553, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[180, 286], [232, 340], [316, 429], [356, 486]]</point>"}]}
{"id": 2198, "image": ["rtx_frames_success_11/49_bridge#episode_23629/frame_0.png", "rtx_frames_success_11/49_bridge#episode_23633/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[463, 100], [434, 181], [411, 243], [394, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[701, 293], [599, 336], [501, 371], [423, 421], [382, 475]]</point>"}]}
{"id": 2200, "image": ["rtx_frames_success_11/49_bridge#episode_23645/frame_0.png", "rtx_frames_success_12/49_bridge#episode_239/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[229, 131], [324, 208], [455, 313], [542, 375], [701, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[562, 62], [498, 97], [426, 170], [371, 239], [330, 363]]</point>"}]}
{"id": 2202, "image": ["rtx_frames_success_12/49_bridge#episode_239/frame_15.png", "rtx_frames_success_12/49_bridge#episode_243/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot to the target location."}, {"from": "gpt", "value": "<point>[[287, 355], [345, 367], [420, 340], [495, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[229, 274], [316, 305], [405, 351], [455, 425]]</point>"}]}
{"id": 2204, "image": ["rtx_frames_success_12/49_bridge#episode_2372/frame_0.png", "rtx_frames_success_12/49_bridge#episode_2375/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the corn."}, {"from": "gpt", "value": "<point>[[290, 232], [420, 220], [527, 259], [663, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[591, 475], [492, 525], [458, 579], [434, 637]]</point>"}]}
{"id": 2206, "image": ["rtx_frames_success_12/49_bridge#episode_2384/frame_0.png", "rtx_frames_success_12/49_bridge#episode_2384/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[550, 320], [446, 320], [278, 371], [145, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cloth towards the spoon."}, {"from": "gpt", "value": "<point>[[130, 490], [240, 463], [333, 444], [458, 390]]</point>"}]}
{"id": 2208, "image": ["rtx_frames_success_12/49_bridge#episode_2426/frame_0.png", "rtx_frames_success_12/49_bridge#episode_2440/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the pan."}, {"from": "gpt", "value": "<point>[[258, 162], [388, 189], [492, 243], [599, 359]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow object."}, {"from": "gpt", "value": "<point>[[678, 139], [585, 185], [495, 224], [417, 332]]</point>"}]}
{"id": 2210, "image": ["rtx_frames_success_12/49_bridge#episode_2448/frame_0.png", "rtx_frames_success_12/49_bridge#episode_2449/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the scissors."}, {"from": "gpt", "value": "<point>[[290, 371], [495, 371], [599, 378], [701, 413], [791, 498]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[657, 139], [625, 208], [605, 324], [602, 402], [620, 598]]</point>"}]}
{"id": 2212, "image": ["rtx_frames_success_12/49_bridge#episode_2462/frame_0.png", "rtx_frames_success_12/49_bridge#episode_2465/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the bowl."}, {"from": "gpt", "value": "<point>[[614, 212], [565, 259], [515, 332], [472, 409], [405, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[692, 475], [608, 463], [521, 436], [455, 320]]</point>"}]}
{"id": 2214, "image": ["rtx_frames_success_12/49_bridge#episode_2468/frame_0.png", "rtx_frames_success_12/49_bridge#episode_2486/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the towel."}, {"from": "gpt", "value": "<point>[[542, 475], [452, 425], [356, 436], [249, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[96, 154], [211, 116], [290, 120], [388, 139], [458, 181], [498, 239], [550, 359]]</point>"}]}
{"id": 2216, "image": ["rtx_frames_success_12/49_bridge#episode_23655/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23663/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the stuffed animal."}, {"from": "gpt", "value": "<point>[[628, 139], [542, 162], [423, 232], [319, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[365, 154], [319, 313], [307, 413], [310, 656]]</point>"}]}
{"id": 2218, "image": ["rtx_frames_success_12/49_bridge#episode_23664/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23666/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[486, 340], [405, 324], [307, 344], [208, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the jar over the pot."}, {"from": "gpt", "value": "<point>[[258, 355], [304, 301], [400, 274], [481, 313]]</point>"}]}
{"id": 2220, "image": ["rtx_frames_success_12/49_bridge#episode_23683/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23703/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[478, 8], [400, 131], [365, 255], [368, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the red fruit."}, {"from": "gpt", "value": "<point>[[397, 189], [513, 216], [585, 309], [657, 475]]</point>"}]}
{"id": 2222, "image": ["rtx_frames_success_12/49_bridge#episode_23706/frame_15.png", "rtx_frames_success_12/49_bridge#episode_23720/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hold the carrot above the container."}, {"from": "gpt", "value": "<point>[[388, 448], [515, 479], [617, 541], [709, 645]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[510, 220], [565, 317], [614, 429], [643, 649]]</point>"}]}
{"id": 2224, "image": ["rtx_frames_success_12/49_bridge#episode_23722/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23724/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position the gripper above the carrot."}, {"from": "gpt", "value": "<point>[[472, 297], [492, 409], [481, 498], [446, 625]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[266, 243], [417, 266], [585, 320], [759, 432]]</point>"}]}
{"id": 2226, "image": ["rtx_frames_success_12/49_bridge#episode_23724/frame_15.png", "rtx_frames_success_12/49_bridge#episode_23732/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the strawberry towards the cloth."}, {"from": "gpt", "value": "<point>[[559, 212], [504, 216], [449, 259], [408, 317]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the plate."}, {"from": "gpt", "value": "<point>[[788, 290], [773, 405], [776, 521], [796, 710]]</point>"}]}
{"id": 2228, "image": ["rtx_frames_success_12/49_bridge#episode_23732/frame_15.png", "rtx_frames_success_12/49_bridge#episode_23736/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the mushroom on the plate."}, {"from": "gpt", "value": "<point>[[602, 201], [568, 274], [550, 340], [573, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green cloth."}, {"from": "gpt", "value": "<point>[[324, 0], [353, 85], [411, 181], [446, 332], [469, 502]]</point>"}]}
{"id": 2230, "image": ["rtx_frames_success_12/49_bridge#episode_23745/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23761/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[269, 81], [371, 143], [455, 297], [495, 417], [515, 722]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[330, 131], [304, 290], [327, 421], [374, 618]]</point>"}]}
{"id": 2232, "image": ["rtx_frames_success_12/49_bridge#episode_23776/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23801/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[608, 166], [672, 259], [704, 363], [715, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[327, 270], [316, 344], [301, 448], [350, 533]]</point>"}]}
{"id": 2234, "image": ["rtx_frames_success_12/49_bridge#episode_23806/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23806/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up clothes from a basket."}, {"from": "gpt", "value": "<point>[[284, 50], [310, 151], [336, 270], [339, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the clothes into the washing machine."}, {"from": "gpt", "value": "<point>[[278, 560], [426, 475], [527, 355], [617, 151]]</point>"}]}
{"id": 2236, "image": ["rtx_frames_success_12/49_bridge#episode_23812/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23824/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the yellow fruit."}, {"from": "gpt", "value": "<point>[[654, 193], [643, 255], [611, 328], [542, 413], [478, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[562, 332], [463, 436], [400, 537], [391, 680]]</point>"}]}
{"id": 2238, "image": ["rtx_frames_success_12/49_bridge#episode_23832/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23833/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[426, 290], [533, 359], [579, 452], [625, 691]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[620, 15], [573, 127], [521, 282], [518, 517]]</point>"}]}
{"id": 2240, "image": ["rtx_frames_success_12/49_bridge#episode_23840/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23841/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the pot."}, {"from": "gpt", "value": "<point>[[597, 212], [530, 297], [463, 398], [417, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[316, 58], [359, 143], [385, 293], [391, 595]]</point>"}]}
{"id": 2242, "image": ["rtx_frames_success_12/49_bridge#episode_23841/frame_15.png", "rtx_frames_success_12/49_bridge#episode_23870/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the sushi towards the pan."}, {"from": "gpt", "value": "<point>[[362, 425], [510, 506], [611, 506], [741, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple towel."}, {"from": "gpt", "value": "<point>[[374, 170], [298, 208], [214, 301], [148, 429]]</point>"}]}
{"id": 2244, "image": ["rtx_frames_success_12/49_bridge#episode_23879/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23879/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object in the pot."}, {"from": "gpt", "value": "<point>[[492, 139], [397, 151], [327, 228], [266, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the object towards the cloth."}, {"from": "gpt", "value": "<point>[[513, 166], [582, 236], [620, 347], [631, 490]]</point>"}]}
{"id": 2246, "image": ["rtx_frames_success_12/49_bridge#episode_23892/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23894/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the vegetable."}, {"from": "gpt", "value": "<point>[[469, 266], [530, 309], [562, 367], [585, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[423, 247], [507, 278], [556, 344], [597, 448]]</point>"}]}
{"id": 2248, "image": ["rtx_frames_success_12/49_bridge#episode_23895/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23896/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[643, 293], [582, 278], [463, 278], [368, 290], [237, 344]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[588, 297], [536, 266], [489, 251], [400, 224]]</point>"}]}
{"id": 2250, "image": ["rtx_frames_success_12/49_bridge#episode_23914/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23928/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[614, 328], [634, 413], [628, 525], [582, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red object."}, {"from": "gpt", "value": "<point>[[403, 259], [504, 320], [565, 425], [597, 541], [666, 734]]</point>"}]}
{"id": 2252, "image": ["rtx_frames_success_12/49_bridge#episode_23930/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23933/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[475, 54], [542, 120], [608, 193], [666, 297], [715, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[643, 452], [504, 405], [353, 409], [208, 452]]</point>"}]}
{"id": 2254, "image": ["rtx_frames_success_12/49_bridge#episode_23933/frame_15.png", "rtx_frames_success_12/49_bridge#episode_23941/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan closer to the corn."}, {"from": "gpt", "value": "<point>[[258, 351], [405, 371], [559, 429], [709, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[287, 108], [350, 151], [385, 208], [397, 328]]</point>"}]}
{"id": 2256, "image": ["rtx_frames_success_12/49_bridge#episode_23941/frame_15.png", "rtx_frames_success_12/49_bridge#episode_23948/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the banan on the green towel."}, {"from": "gpt", "value": "<point>[[362, 124], [460, 158], [556, 224], [611, 309], [681, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spatula."}, {"from": "gpt", "value": "<point>[[423, 313], [353, 324], [261, 359], [194, 463]]</point>"}]}
{"id": 2258, "image": ["rtx_frames_success_12/49_bridge#episode_23962/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23968/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[469, 147], [431, 239], [411, 317], [394, 405], [379, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[220, 135], [275, 205], [319, 324], [330, 552]]</point>"}]}
{"id": 2260, "image": ["rtx_frames_success_12/49_bridge#episode_23968/frame_15.png", "rtx_frames_success_12/49_bridge#episode_23975/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[226, 459], [330, 452], [501, 425], [692, 340]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[298, 270], [371, 290], [414, 340], [469, 432]]</point>"}]}
{"id": 2262, "image": ["rtx_frames_success_12/49_bridge#episode_23977/frame_0.png", "rtx_frames_success_12/49_bridge#episode_23984/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the scoop."}, {"from": "gpt", "value": "<point>[[608, 247], [530, 243], [431, 247], [295, 286], [214, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle in the pot."}, {"from": "gpt", "value": "<point>[[437, 224], [345, 270], [298, 386]]</point>"}]}
{"id": 2264, "image": ["rtx_frames_success_12/49_bridge#episode_23984/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24006/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bottle towards the table."}, {"from": "gpt", "value": "<point>[[559, 479], [652, 571], [707, 764]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[307, 270], [446, 290], [588, 290], [701, 378]]</point>"}]}
{"id": 2266, "image": ["rtx_frames_success_12/49_bridge#episode_24006/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24025/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull open the refrigerator door."}, {"from": "gpt", "value": "<point>[[498, 483], [431, 571], [336, 606], [223, 637]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the towel."}, {"from": "gpt", "value": "<point>[[345, 297], [400, 413], [426, 598]]</point>"}]}
{"id": 2268, "image": ["rtx_frames_success_12/49_bridge#episode_24030/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24030/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spatula."}, {"from": "gpt", "value": "<point>[[235, 143], [414, 224], [533, 347], [715, 625]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spatula over the pot."}, {"from": "gpt", "value": "<point>[[411, 197], [342, 243], [295, 332]]</point>"}]}
{"id": 2270, "image": ["rtx_frames_success_12/49_bridge#episode_24037/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24038/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[429, 181], [356, 201], [313, 263], [264, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower the gripper towards the pot."}, {"from": "gpt", "value": "<point>[[153, 444], [272, 463], [388, 479], [515, 521], [623, 633]]</point>"}]}
{"id": 2272, "image": ["rtx_frames_success_12/49_bridge#episode_24041/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24049/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan with the red cube."}, {"from": "gpt", "value": "<point>[[678, 347], [594, 301], [533, 278], [455, 282]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[246, 174], [350, 193], [397, 278], [443, 413], [466, 606]]</point>"}]}
{"id": 2274, "image": ["rtx_frames_success_12/49_bridge#episode_24059/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24063/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the towel."}, {"from": "gpt", "value": "<point>[[177, 197], [287, 224], [431, 274], [559, 340], [689, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spatula."}, {"from": "gpt", "value": "<point>[[539, 498], [458, 471], [403, 421], [371, 317]]</point>"}]}
{"id": 2276, "image": ["rtx_frames_success_12/49_bridge#episode_24082/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24085/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon."}, {"from": "gpt", "value": "<point>[[298, 178], [423, 243], [527, 328], [588, 463], [620, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[423, 228], [350, 205], [295, 236], [258, 328]]</point>"}]}
{"id": 2278, "image": ["rtx_frames_success_12/49_bridge#episode_24085/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24093/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the broccoli in the pot."}, {"from": "gpt", "value": "<point>[[246, 135], [226, 232], [235, 375], [243, 479], [281, 649]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[284, 205], [388, 205], [472, 243], [542, 290], [605, 371]]</point>"}]}
{"id": 2280, "image": ["rtx_frames_success_12/49_bridge#episode_24099/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24100/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[226, 363], [339, 347], [484, 347], [625, 382], [753, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[301, 112], [397, 135], [429, 189], [458, 259], [458, 305]]</point>"}]}
{"id": 2282, "image": ["rtx_frames_success_12/49_bridge#episode_24102/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24104/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[637, 224], [625, 313], [631, 456], [646, 683]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[313, 247], [426, 247], [542, 274], [707, 432]]</point>"}]}
{"id": 2284, "image": ["rtx_frames_success_12/49_bridge#episode_24113/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24115/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[292, 0], [368, 69], [420, 151], [463, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[298, 166], [240, 201], [226, 243], [217, 347]]</point>"}]}
{"id": 2286, "image": ["rtx_frames_success_12/49_bridge#episode_24115/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24122/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the sushi roll in the pot."}, {"from": "gpt", "value": "<point>[[232, 320], [333, 278], [455, 278], [527, 320], [614, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the capsicum."}, {"from": "gpt", "value": "<point>[[643, 305], [565, 290], [484, 278], [379, 305]]</point>"}]}
{"id": 2288, "image": ["rtx_frames_success_12/49_bridge#episode_24132/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24141/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green towel."}, {"from": "gpt", "value": "<point>[[628, 479], [588, 537], [550, 602], [556, 691]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pot on the table."}, {"from": "gpt", "value": "<point>[[513, 448], [547, 525], [568, 606], [542, 687]]</point>"}]}
{"id": 2290, "image": ["rtx_frames_success_12/49_bridge#episode_24148/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24154/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[266, 197], [417, 216], [495, 247], [565, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[347, 112], [460, 135], [573, 205], [657, 263], [750, 386]]</point>"}]}
{"id": 2292, "image": ["rtx_frames_success_12/49_bridge#episode_24186/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24189/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[310, 224], [397, 212], [513, 247], [649, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[391, 143], [310, 143], [226, 239], [162, 367]]</point>"}]}
{"id": 2294, "image": ["rtx_frames_success_12/49_bridge#episode_24192/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24192/frame_27.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[718, 46], [631, 85], [498, 162], [345, 259], [220, 355], [119, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to release the broccoli."}, {"from": "gpt", "value": "<point>[[272, 228], [327, 290], [371, 347], [385, 467]]</point>"}]}
{"id": 2296, "image": ["rtx_frames_success_12/49_bridge#episode_24207/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24213/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[669, 274], [573, 317], [481, 413], [434, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the knife."}, {"from": "gpt", "value": "<point>[[559, 8], [475, 46], [403, 112], [353, 313]]</point>"}]}
{"id": 2298, "image": ["rtx_frames_success_12/49_bridge#episode_24214/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24235/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the tiffin box."}, {"from": "gpt", "value": "<point>[[550, 463], [507, 413], [426, 347], [281, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[223, 278], [275, 332], [327, 448], [371, 602]]</point>"}]}
{"id": 2300, "image": ["rtx_frames_success_12/49_bridge#episode_24263/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24280/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[681, 228], [556, 228], [455, 290], [391, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[715, 205], [663, 228], [576, 266], [469, 313], [388, 386], [284, 502]]</point>"}]}
{"id": 2302, "image": ["rtx_frames_success_12/49_bridge#episode_24284/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24300/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the eggplant."}, {"from": "gpt", "value": "<point>[[437, 263], [475, 336], [504, 429], [521, 653]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[194, 178], [304, 236], [376, 347], [443, 556]]</point>"}]}
{"id": 2304, "image": ["rtx_frames_success_12/49_bridge#episode_24307/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24317/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[220, 104], [295, 139], [342, 197], [431, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[295, 69], [391, 89], [455, 162], [501, 243], [547, 448]]</point>"}]}
{"id": 2306, "image": ["rtx_frames_success_12/49_bridge#episode_24351/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24353/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the can."}, {"from": "gpt", "value": "<point>[[345, 224], [513, 259], [634, 317], [738, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the can."}, {"from": "gpt", "value": "<point>[[605, 239], [513, 224], [350, 205], [237, 232]]</point>"}]}
{"id": 2308, "image": ["rtx_frames_success_12/49_bridge#episode_24358/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24364/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[533, 274], [556, 378], [605, 502], [666, 680]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[272, 263], [446, 293], [530, 355], [625, 479]]</point>"}]}
{"id": 2310, "image": ["rtx_frames_success_12/49_bridge#episode_24364/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24377/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull open the refrigerator door."}, {"from": "gpt", "value": "<point>[[469, 664], [405, 710], [310, 726], [194, 707]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[374, 216], [345, 297], [342, 340], [345, 390]]</point>"}]}
{"id": 2312, "image": ["rtx_frames_success_12/49_bridge#episode_24388/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24429/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[544, 216], [489, 255], [452, 317], [417, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[237, 344], [379, 320], [521, 324], [695, 402]]</point>"}]}
{"id": 2314, "image": ["rtx_frames_success_12/49_bridge#episode_24446/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24449/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[275, 181], [319, 216], [368, 251], [417, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pan containing the broccoli."}, {"from": "gpt", "value": "<point>[[704, 251], [637, 247], [530, 270], [417, 328]]</point>"}]}
{"id": 2316, "image": ["rtx_frames_success_12/49_bridge#episode_24451/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24453/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[440, 139], [481, 228], [527, 347], [588, 514], [591, 672]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[217, 205], [301, 247], [379, 324], [481, 425], [565, 649]]</point>"}]}
{"id": 2318, "image": ["rtx_frames_success_12/49_bridge#episode_24458/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24463/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cucumber."}, {"from": "gpt", "value": "<point>[[843, 351], [767, 363], [669, 371], [605, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[588, 367], [559, 440], [504, 494], [434, 560]]</point>"}]}
{"id": 2320, "image": ["rtx_frames_success_12/49_bridge#episode_24463/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24465/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the target location."}, {"from": "gpt", "value": "<point>[[385, 614], [304, 583], [249, 506], [220, 432], [211, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the fridge handle."}, {"from": "gpt", "value": "<point>[[171, 425], [264, 425], [417, 475], [489, 514], [605, 598]]</point>"}]}
{"id": 2322, "image": ["rtx_frames_success_12/49_bridge#episode_24465/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24467/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to open the fridge door."}, {"from": "gpt", "value": "<point>[[434, 741], [374, 772], [295, 815], [165, 834]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[359, 112], [304, 178], [258, 286], [237, 444]]</point>"}]}
{"id": 2324, "image": ["rtx_frames_success_12/49_bridge#episode_24467/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24476/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the fork towards the cloth."}, {"from": "gpt", "value": "<point>[[145, 232], [266, 263], [342, 394], [417, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the orange towel."}, {"from": "gpt", "value": "<point>[[397, 228], [353, 247], [307, 301], [281, 402]]</point>"}]}
{"id": 2326, "image": ["rtx_frames_success_12/49_bridge#episode_24492/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24494/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[379, 124], [307, 166], [298, 243], [292, 347], [290, 444]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the towel."}, {"from": "gpt", "value": "<point>[[776, 205], [712, 228], [602, 251], [507, 305], [414, 394]]</point>"}]}
{"id": 2328, "image": ["rtx_frames_success_12/49_bridge#episode_24512/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24512/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the brush."}, {"from": "gpt", "value": "<point>[[295, 143], [397, 201], [469, 297], [521, 456], [559, 649]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the brush next to the bread."}, {"from": "gpt", "value": "<point>[[544, 456], [460, 459], [431, 402], [417, 309]]</point>"}]}
{"id": 2330, "image": ["rtx_frames_success_12/49_bridge#episode_24517/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24517/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[637, 514], [568, 483], [501, 417], [489, 340]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the broccoli away from the pan."}, {"from": "gpt", "value": "<point>[[492, 347], [518, 421], [536, 537], [550, 691]]</point>"}]}
{"id": 2332, "image": ["rtx_frames_success_12/49_bridge#episode_24520/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24520/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the cheese in the small metal bowl."}, {"from": "gpt", "value": "<point>[[316, 301], [414, 274], [489, 274], [608, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bowl with the cheese over the stovetop."}, {"from": "gpt", "value": "<point>[[614, 243], [527, 293], [455, 344], [388, 452]]</point>"}]}
{"id": 2334, "image": ["rtx_frames_success_12/49_bridge#episode_24524/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24529/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lime in the pot."}, {"from": "gpt", "value": "<point>[[414, 301], [339, 351], [272, 386], [269, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[429, 135], [411, 205], [374, 305], [359, 417]]</point>"}]}
{"id": 2336, "image": ["rtx_frames_success_12/49_bridge#episode_24529/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24530/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot over the stove."}, {"from": "gpt", "value": "<point>[[411, 347], [481, 409], [559, 502], [594, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[258, 139], [353, 170], [434, 236], [469, 309], [513, 405], [527, 541]]</point>"}]}
{"id": 2338, "image": ["rtx_frames_success_12/49_bridge#episode_24535/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24538/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the grapes."}, {"from": "gpt", "value": "<point>[[371, 181], [495, 212], [594, 274], [669, 359], [733, 429], [776, 521], [822, 683]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[350, 69], [304, 116], [275, 181], [255, 286], [235, 452]]</point>"}]}
{"id": 2340, "image": ["rtx_frames_success_12/49_bridge#episode_24544/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24566/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue measuring spoon."}, {"from": "gpt", "value": "<point>[[301, 185], [356, 205], [376, 239], [403, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[258, 367], [350, 328], [423, 328], [478, 355], [573, 421]]</point>"}]}
{"id": 2342, "image": ["rtx_frames_success_12/49_bridge#episode_24566/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24568/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[573, 394], [515, 386], [452, 351], [365, 324], [269, 301]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[327, 236], [313, 305], [316, 402], [310, 544]]</point>"}]}
{"id": 2344, "image": ["rtx_frames_success_12/49_bridge#episode_24572/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24572/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the chicken leg."}, {"from": "gpt", "value": "<point>[[388, 224], [446, 224], [495, 274], [530, 324], [547, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the chicken leg towards the bowl."}, {"from": "gpt", "value": "<point>[[515, 266], [443, 293], [339, 328], [290, 402], [278, 459]]</point>"}]}
{"id": 2346, "image": ["rtx_frames_success_12/49_bridge#episode_24573/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24574/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[762, 154], [744, 224], [718, 320], [721, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[646, 359], [643, 425], [625, 459], [576, 479]]</point>"}]}
{"id": 2348, "image": ["rtx_frames_success_12/49_bridge#episode_24591/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24641/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[513, 97], [515, 170], [524, 239], [547, 398]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[310, 193], [313, 293], [301, 382], [246, 479]]</point>"}]}
{"id": 2350, "image": ["rtx_frames_success_12/49_bridge#episode_24641/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24658/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the metal pot on the blue towel."}, {"from": "gpt", "value": "<point>[[458, 77], [518, 97], [611, 135], [686, 178], [759, 266], [796, 355], [822, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[287, 270], [229, 243], [188, 259], [127, 301]]</point>"}]}
{"id": 2352, "image": ["rtx_frames_success_12/49_bridge#episode_24658/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24666/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the spoon on the table."}, {"from": "gpt", "value": "<point>[[232, 208], [365, 239], [484, 317], [605, 606]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the towel."}, {"from": "gpt", "value": "<point>[[255, 278], [321, 263], [388, 259], [437, 305]]</point>"}]}
{"id": 2354, "image": ["rtx_frames_success_12/49_bridge#episode_24674/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24681/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[284, 54], [324, 97], [356, 197], [374, 297], [371, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spatula in the sink."}, {"from": "gpt", "value": "<point>[[678, 286], [646, 309], [597, 332], [539, 382], [515, 475]]</point>"}]}
{"id": 2356, "image": ["rtx_frames_success_12/49_bridge#episode_24682/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24686/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the almond container."}, {"from": "gpt", "value": "<point>[[481, 181], [463, 247], [446, 305], [440, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi in the container."}, {"from": "gpt", "value": "<point>[[330, 170], [278, 270], [266, 359], [284, 525]]</point>"}]}
{"id": 2358, "image": ["rtx_frames_success_12/49_bridge#episode_24697/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24697/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[747, 166], [660, 162], [565, 197], [486, 266], [446, 317]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the stuffed toy on the towel."}, {"from": "gpt", "value": "<point>[[475, 181], [562, 205], [634, 255], [681, 332], [730, 448]]</point>"}]}
{"id": 2360, "image": ["rtx_frames_success_12/49_bridge#episode_24708/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24708/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[492, 185], [582, 216], [623, 266], [675, 347], [704, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the pepper."}, {"from": "gpt", "value": "<point>[[704, 429], [640, 417], [553, 425], [452, 490]]</point>"}]}
{"id": 2362, "image": ["rtx_frames_success_12/49_bridge#episode_24712/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24713/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cube."}, {"from": "gpt", "value": "<point>[[321, 290], [405, 309], [463, 371], [515, 440], [544, 494], [614, 691]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana in the colander."}, {"from": "gpt", "value": "<point>[[458, 216], [515, 286], [550, 394], [565, 514], [585, 714]]</point>"}]}
{"id": 2364, "image": ["rtx_frames_success_12/49_bridge#episode_24724/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24736/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[333, 162], [391, 228], [437, 340], [463, 440], [475, 645]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the scissors."}, {"from": "gpt", "value": "<point>[[281, 467], [443, 459], [585, 483], [782, 587]]</point>"}]}
{"id": 2366, "image": ["rtx_frames_success_12/49_bridge#episode_24740/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24754/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[568, 220], [544, 259], [504, 297], [446, 363], [353, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the yellow bell pepper."}, {"from": "gpt", "value": "<point>[[550, 147], [602, 197], [666, 259], [704, 390]]</point>"}]}
{"id": 2368, "image": ["rtx_frames_success_12/49_bridge#episode_24757/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24760/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the yellow pepper."}, {"from": "gpt", "value": "<point>[[808, 386], [791, 452], [785, 541], [805, 683]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[136, 367], [229, 363], [342, 394], [429, 436], [550, 556]]</point>"}]}
{"id": 2370, "image": ["rtx_frames_success_12/49_bridge#episode_24784/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24790/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[440, 151], [507, 212], [533, 286], [562, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the corn."}, {"from": "gpt", "value": "<point>[[472, 15], [498, 62], [539, 139], [565, 224], [582, 328], [591, 440], [602, 541], [599, 726]]</point>"}]}
{"id": 2372, "image": ["rtx_frames_success_12/49_bridge#episode_24797/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24802/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the clothes."}, {"from": "gpt", "value": "<point>[[321, 328], [371, 359], [405, 429], [437, 498], [481, 660]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the container."}, {"from": "gpt", "value": "<point>[[597, 278], [611, 336], [611, 405], [579, 537]]</point>"}]}
{"id": 2374, "image": ["rtx_frames_success_12/49_bridge#episode_24805/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24808/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the yellow towel."}, {"from": "gpt", "value": "<point>[[611, 344], [492, 270], [417, 259], [356, 259]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the egg."}, {"from": "gpt", "value": "<point>[[764, 378], [605, 320], [460, 324], [330, 347], [174, 402]]</point>"}]}
{"id": 2376, "image": ["rtx_frames_success_12/49_bridge#episode_24813/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24816/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[652, 197], [707, 286], [736, 371], [744, 471], [753, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the fork."}, {"from": "gpt", "value": "<point>[[484, 402], [368, 390], [281, 378], [145, 429]]</point>"}]}
{"id": 2378, "image": ["rtx_frames_success_12/49_bridge#episode_24816/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24832/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the fork towards the spoon."}, {"from": "gpt", "value": "<point>[[275, 529], [362, 533], [472, 568], [559, 625], [631, 714]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[345, 486], [405, 498], [507, 494], [591, 475], [649, 444], [744, 351]]</point>"}]}
{"id": 2380, "image": ["rtx_frames_success_12/49_bridge#episode_24832/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24836/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to open the microwave door."}, {"from": "gpt", "value": "<point>[[721, 332], [704, 378], [649, 456], [562, 517], [492, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the can."}, {"from": "gpt", "value": "<point>[[640, 236], [634, 297], [623, 398], [562, 463]]</point>"}]}
{"id": 2382, "image": ["rtx_frames_success_12/49_bridge#episode_24846/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24849/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[156, 378], [208, 332], [284, 309], [403, 320]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[486, 745], [568, 722], [646, 649], [683, 537], [701, 463], [730, 317]]</point>"}]}
{"id": 2384, "image": ["rtx_frames_success_12/49_bridge#episode_24857/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24861/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[411, 178], [365, 197], [307, 228], [255, 297], [220, 355], [214, 409]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the rag away from the bowl."}, {"from": "gpt", "value": "<point>[[469, 309], [611, 363], [715, 390], [799, 452]]</point>"}]}
{"id": 2386, "image": ["rtx_frames_success_12/49_bridge#episode_24867/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24867/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[374, 127], [423, 178], [466, 243], [495, 355], [515, 448], [530, 548], [530, 614]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to slide the cloth to the left."}, {"from": "gpt", "value": "<point>[[486, 645], [408, 602], [304, 556], [185, 471]]</point>"}]}
{"id": 2388, "image": ["rtx_frames_success_12/49_bridge#episode_24880/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24880/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[133, 197], [197, 201], [301, 236], [391, 363]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the wok."}, {"from": "gpt", "value": "<point>[[426, 201], [440, 305], [460, 398], [469, 502], [455, 656]]</point>"}]}
{"id": 2390, "image": ["rtx_frames_success_12/49_bridge#episode_24888/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24917/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[391, 108], [481, 170], [542, 263], [582, 359], [623, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[530, 197], [469, 208], [408, 243], [368, 305], [324, 394]]</point>"}]}
{"id": 2392, "image": ["rtx_frames_success_12/49_bridge#episode_24930/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24933/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[324, 131], [336, 216], [327, 309], [264, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[272, 286], [336, 259], [437, 243], [515, 247], [597, 274], [654, 305], [747, 425]]</point>"}]}
{"id": 2394, "image": ["rtx_frames_success_12/49_bridge#episode_24934/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24940/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[597, 174], [597, 293], [599, 413], [631, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[405, 297], [350, 317], [287, 328], [171, 440]]</point>"}]}
{"id": 2396, "image": ["rtx_frames_success_12/49_bridge#episode_24940/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24945/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the napkin."}, {"from": "gpt", "value": "<point>[[156, 390], [232, 398], [278, 336], [327, 282]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the eggplant."}, {"from": "gpt", "value": "<point>[[333, 127], [376, 185], [417, 282], [443, 355], [463, 471], [466, 606]]</point>"}]}
{"id": 2398, "image": ["rtx_frames_success_12/49_bridge#episode_24947/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24950/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush to the left of the towel."}, {"from": "gpt", "value": "<point>[[510, 390], [513, 479], [533, 552], [556, 625]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blueberries."}, {"from": "gpt", "value": "<point>[[637, 444], [756, 463], [848, 498], [930, 568], [982, 722]]</point>"}]}
{"id": 2400, "image": ["rtx_frames_success_12/49_bridge#episode_24950/frame_15.png", "rtx_frames_success_12/49_bridge#episode_24960/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the blueberries."}, {"from": "gpt", "value": "<point>[[958, 753], [912, 699], [834, 676], [767, 672], [721, 683]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[208, 131], [281, 170], [365, 224], [429, 293], [533, 432]]</point>"}]}
{"id": 2402, "image": ["rtx_frames_success_12/49_bridge#episode_24964/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24978/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position above the fruit."}, {"from": "gpt", "value": "<point>[[704, 197], [681, 270], [649, 328], [614, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[724, 324], [707, 386], [695, 429], [657, 444]]</point>"}]}
{"id": 2404, "image": ["rtx_frames_success_12/49_bridge#episode_24979/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24979/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[576, 351], [570, 429], [573, 506], [602, 595], [652, 734]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spatula."}, {"from": "gpt", "value": "<point>[[579, 228], [501, 224], [374, 243], [272, 309]]</point>"}]}
{"id": 2406, "image": ["rtx_frames_success_12/49_bridge#episode_24991/frame_0.png", "rtx_frames_success_12/49_bridge#episode_24998/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[559, 317], [559, 413], [573, 529], [608, 695]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[533, 903], [452, 822], [356, 753], [206, 749], [139, 792]]</point>"}]}
{"id": 2408, "image": ["rtx_frames_success_12/49_bridge#episode_24998/frame_15.png", "rtx_frames_success_12/49_bridge#episode_25008/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the cabinet door closed."}, {"from": "gpt", "value": "<point>[[608, 710], [652, 714], [730, 699], [840, 653]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[611, 328], [565, 402], [518, 502], [489, 710]]</point>"}]}
{"id": 2410, "image": ["rtx_frames_success_12/49_bridge#episode_25011/frame_0.png", "rtx_frames_success_12/49_bridge#episode_25012/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[846, 425], [727, 425], [652, 432], [559, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[327, 147], [272, 193], [243, 263], [229, 344], [208, 471]]</point>"}]}
{"id": 2412, "image": ["rtx_frames_success_12/49_bridge#episode_25017/frame_0.png", "rtx_frames_success_12/49_bridge#episode_25017/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spice container."}, {"from": "gpt", "value": "<point>[[510, 556], [585, 514], [654, 456], [692, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the spice container."}, {"from": "gpt", "value": "<point>[[637, 440], [568, 498], [513, 560], [455, 668]]</point>"}]}
{"id": 2414, "image": ["rtx_frames_success_12/49_bridge#episode_25020/frame_0.png", "rtx_frames_success_12/49_bridge#episode_25041/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red spoon."}, {"from": "gpt", "value": "<point>[[295, 27], [284, 104], [275, 181], [287, 278], [301, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[226, 239], [298, 259], [388, 274], [486, 301], [542, 351], [605, 471]]</point>"}]}
{"id": 2416, "image": ["rtx_frames_success_13/49_bridge#episode_27/frame_0.png", "rtx_frames_success_13/49_bridge#episode_30/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the yellow towel."}, {"from": "gpt", "value": "<point>[[692, 501], [586, 429], [402, 401], [286, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[919, 143], [820, 271], [705, 481], [652, 602]]</point>"}]}
{"id": 2418, "image": ["rtx_frames_success_13/49_bridge#episode_30/frame_15.png", "rtx_frames_success_13/49_bridge#episode_252/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the fork over the rag."}, {"from": "gpt", "value": "<point>[[530, 366], [430, 371], [323, 388], [226, 419]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the banana."}, {"from": "gpt", "value": "<point>[[429, 211], [421, 258], [410, 291], [383, 326]]</point>"}]}
{"id": 2420, "image": ["rtx_frames_success_13/49_bridge#episode_252/frame_15.png", "rtx_frames_success_13/49_bridge#episode_257/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana over the table."}, {"from": "gpt", "value": "<point>[[201, 175], [154, 236], [133, 356], [141, 461]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pear."}, {"from": "gpt", "value": "<point>[[547, 313], [455, 318], [314, 388], [218, 466]]</point>"}]}
{"id": 2422, "image": ["rtx_frames_success_13/49_bridge#episode_270/frame_0.png", "rtx_frames_success_13/49_bridge#episode_273/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the cheese."}, {"from": "gpt", "value": "<point>[[462, 198], [558, 236], [635, 283], [699, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the eggplant."}, {"from": "gpt", "value": "<point>[[579, 311], [489, 313], [380, 348], [293, 406]]</point>"}]}
{"id": 2424, "image": ["rtx_frames_success_13/49_bridge#episode_280/frame_0.png", "rtx_frames_success_13/49_bridge#episode_313/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[665, 491], [579, 424], [455, 381], [325, 381]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green bar."}, {"from": "gpt", "value": "<point>[[335, 155], [312, 203], [305, 293], [320, 351]]</point>"}]}
{"id": 2426, "image": ["rtx_frames_success_13/49_bridge#episode_321/frame_0.png", "rtx_frames_success_13/49_bridge#episode_337/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[256, 130], [258, 241], [297, 416], [350, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[534, 8], [498, 140], [489, 331], [500, 549]]</point>"}]}
{"id": 2428, "image": ["rtx_frames_success_13/49_bridge#episode_349/frame_0.png", "rtx_frames_success_13/49_bridge#episode_349/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[207, 55], [286, 103], [370, 183], [444, 303]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the pot."}, {"from": "gpt", "value": "<point>[[451, 301], [378, 281], [271, 341], [233, 419]]</point>"}]}
{"id": 2430, "image": ["rtx_frames_success_13/49_bridge#episode_355/frame_0.png", "rtx_frames_success_13/49_bridge#episode_355/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[463, 69], [417, 278], [405, 440], [385, 618]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the pot."}, {"from": "gpt", "value": "<point>[[388, 637], [478, 537], [559, 463], [654, 467]]</point>"}]}
{"id": 2432, "image": ["rtx_frames_success_13/49_bridge#episode_2506/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2543/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green grater."}, {"from": "gpt", "value": "<point>[[259, 216], [241, 278], [244, 331], [263, 398]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the tomato."}, {"from": "gpt", "value": "<point>[[641, 451], [517, 439], [344, 459], [235, 494]]</point>"}]}
{"id": 2434, "image": ["rtx_frames_success_13/49_bridge#episode_2543/frame_15.png", "rtx_frames_success_13/49_bridge#episode_2545/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the tomato towards the top right corner."}, {"from": "gpt", "value": "<point>[[220, 451], [367, 411], [555, 419], [731, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[430, 313], [342, 353], [280, 393], [194, 466]]</point>"}]}
{"id": 2436, "image": ["rtx_frames_success_13/49_bridge#episode_2553/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2562/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the bowl."}, {"from": "gpt", "value": "<point>[[656, 105], [586, 223], [509, 323], [423, 396]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[327, 95], [333, 211], [321, 351], [297, 446]]</point>"}]}
{"id": 2438, "image": ["rtx_frames_success_13/49_bridge#episode_2576/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2576/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pan."}, {"from": "gpt", "value": "<point>[[880, 449], [865, 549], [836, 659], [805, 749]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to put the pan on the stove."}, {"from": "gpt", "value": "<point>[[671, 100], [769, 188], [868, 321], [938, 436]]</point>"}]}
{"id": 2440, "image": ["rtx_frames_success_13/49_bridge#episode_2593/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2593/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[184, 148], [331, 218], [474, 321], [588, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[588, 406], [496, 446], [391, 466], [299, 466]]</point>"}]}
{"id": 2442, "image": ["rtx_frames_success_13/49_bridge#episode_2598/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2600/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[442, 100], [496, 190], [543, 306], [564, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[474, 55], [583, 180], [679, 321], [748, 464]]</point>"}]}
{"id": 2444, "image": ["rtx_frames_success_13/49_bridge#episode_2604/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2604/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the can."}, {"from": "gpt", "value": "<point>[[393, 446], [316, 371], [265, 336], [162, 306]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the can."}, {"from": "gpt", "value": "<point>[[246, 221], [397, 251], [538, 311], [665, 401]]</point>"}]}
{"id": 2446, "image": ["rtx_frames_success_13/49_bridge#episode_2612/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2615/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the vessel."}, {"from": "gpt", "value": "<point>[[684, 466], [633, 411], [530, 348], [459, 331]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[374, 301], [412, 406], [425, 504], [427, 612]]</point>"}]}
{"id": 2448, "image": ["rtx_frames_success_13/49_bridge#episode_2627/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2627/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[459, 103], [566, 180], [675, 288], [769, 401]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl on the blue towel."}, {"from": "gpt", "value": "<point>[[498, 195], [417, 228], [312, 321], [259, 401]]</point>"}]}
{"id": 2450, "image": ["rtx_frames_success_13/49_bridge#episode_2630/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2634/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fabric."}, {"from": "gpt", "value": "<point>[[211, 206], [244, 276], [282, 341], [327, 378]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[372, 140], [417, 183], [466, 256], [500, 321]]</point>"}]}
{"id": 2452, "image": ["rtx_frames_success_13/49_bridge#episode_2644/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2644/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[494, 135], [556, 195], [613, 276], [656, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bowl towards the lower right side of the table."}, {"from": "gpt", "value": "<point>[[291, 150], [288, 248], [256, 396], [207, 461]]</point>"}]}
{"id": 2454, "image": ["rtx_frames_success_13/49_bridge#episode_2650/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2657/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[558, 105], [545, 246], [494, 434], [434, 544]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the clothes."}, {"from": "gpt", "value": "<point>[[346, 419], [425, 486], [536, 662], [575, 772]]</point>"}]}
{"id": 2456, "image": ["rtx_frames_success_13/49_bridge#episode_2658/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2658/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the stuffed animal."}, {"from": "gpt", "value": "<point>[[293, 70], [429, 158], [575, 276], [684, 386]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the stuffed animal to the left."}, {"from": "gpt", "value": "<point>[[581, 180], [444, 251], [299, 348], [195, 441]]</point>"}]}
{"id": 2458, "image": ["rtx_frames_success_13/49_bridge#episode_2661/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2664/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sausage."}, {"from": "gpt", "value": "<point>[[361, 40], [286, 138], [207, 273], [158, 406]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[430, 195], [408, 243], [363, 291], [312, 323]]</point>"}]}
{"id": 2460, "image": ["rtx_frames_success_13/49_bridge#episode_2666/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2680/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the orange rectangle."}, {"from": "gpt", "value": "<point>[[397, 90], [485, 178], [553, 278], [598, 383]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[716, 145], [682, 221], [641, 276], [583, 336]]</point>"}]}
{"id": 2462, "image": ["rtx_frames_success_13/49_bridge#episode_2683/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2727/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the can."}, {"from": "gpt", "value": "<point>[[282, 150], [430, 291], [605, 494], [771, 707]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[382, 323], [451, 306], [508, 321], [568, 363]]</point>"}]}
{"id": 2464, "image": ["rtx_frames_success_13/49_bridge#episode_2731/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2738/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pepper shaker."}, {"from": "gpt", "value": "<point>[[457, 221], [402, 276], [336, 388], [306, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[449, 251], [434, 331], [440, 404], [464, 469]]</point>"}]}
{"id": 2466, "image": ["rtx_frames_success_13/49_bridge#episode_2743/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2743/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pear."}, {"from": "gpt", "value": "<point>[[461, 10], [402, 183], [363, 406], [346, 574]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spoon on the right corner of the table."}, {"from": "gpt", "value": "<point>[[306, 168], [466, 258], [568, 351], [724, 521]]</point>"}]}
{"id": 2468, "image": ["rtx_frames_success_13/49_bridge#episode_2744/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2744/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the sweet potato."}, {"from": "gpt", "value": "<point>[[917, 406], [863, 491], [816, 591], [789, 682]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the red object."}, {"from": "gpt", "value": "<point>[[724, 206], [669, 246], [615, 313], [575, 409]]</point>"}]}
{"id": 2470, "image": ["rtx_frames_success_13/49_bridge#episode_2786/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2786/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato in the bowl."}, {"from": "gpt", "value": "<point>[[325, 53], [368, 140], [397, 231], [417, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the potato towards the knife."}, {"from": "gpt", "value": "<point>[[318, 140], [284, 238], [265, 341], [258, 459]]</point>"}]}
{"id": 2472, "image": ["rtx_frames_success_13/49_bridge#episode_2799/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2803/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[308, 80], [346, 148], [365, 203], [365, 268]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container with the carrot."}, {"from": "gpt", "value": "<point>[[295, 128], [391, 258], [494, 449], [556, 607]]</point>"}]}
{"id": 2474, "image": ["rtx_frames_success_13/49_bridge#episode_2806/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2827/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green pepper."}, {"from": "gpt", "value": "<point>[[684, 566], [594, 516], [440, 494], [335, 509]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[539, 108], [470, 233], [410, 401], [372, 574]]</point>"}]}
{"id": 2476, "image": ["rtx_frames_success_13/49_bridge#episode_2830/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2835/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the jar of almonds."}, {"from": "gpt", "value": "<point>[[395, 481], [476, 464], [555, 479], [628, 539]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green cloth."}, {"from": "gpt", "value": "<point>[[579, 311], [498, 306], [398, 326], [312, 366]]</point>"}]}
{"id": 2478, "image": ["rtx_frames_success_13/49_bridge#episode_2880/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2882/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the stuffed animal."}, {"from": "gpt", "value": "<point>[[560, 80], [541, 135], [536, 291], [545, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ice cream scoop."}, {"from": "gpt", "value": "<point>[[222, 208], [417, 321], [596, 454], [759, 607]]</point>"}]}
{"id": 2480, "image": ["rtx_frames_success_13/49_bridge#episode_2921/frame_15.png", "rtx_frames_success_13/49_bridge#episode_2932/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the broccoli in the pot."}, {"from": "gpt", "value": "<point>[[472, 336], [440, 416], [415, 581], [421, 709]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[310, 55], [400, 145], [511, 353], [553, 501]]</point>"}]}
{"id": 2482, "image": ["rtx_frames_success_13/49_bridge#episode_2936/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2951/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue handle."}, {"from": "gpt", "value": "<point>[[398, 376], [468, 486], [538, 637], [579, 782]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[643, 308], [543, 291], [408, 298], [321, 321]]</point>"}]}
{"id": 2484, "image": ["rtx_frames_success_13/49_bridge#episode_2956/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2962/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[489, 65], [383, 185], [274, 366], [256, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow cloth."}, {"from": "gpt", "value": "<point>[[289, 208], [404, 288], [485, 376], [541, 471]]</point>"}]}
{"id": 2486, "image": ["rtx_frames_success_13/49_bridge#episode_2972/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2973/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[237, 0], [291, 123], [346, 283], [372, 406]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[321, 115], [289, 175], [265, 351], [271, 461]]</point>"}]}
{"id": 2488, "image": ["rtx_frames_success_13/49_bridge#episode_2978/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2978/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the microwave."}, {"from": "gpt", "value": "<point>[[771, 283], [609, 185], [355, 128], [145, 125]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the microwave door closed."}, {"from": "gpt", "value": "<point>[[175, 216], [318, 178], [468, 163], [590, 175]]</point>"}]}
{"id": 2490, "image": ["rtx_frames_success_13/49_bridge#episode_2988/frame_0.png", "rtx_frames_success_13/49_bridge#episode_2997/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cloth."}, {"from": "gpt", "value": "<point>[[393, 70], [474, 163], [596, 353], [639, 449]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[331, 632], [259, 667], [139, 672], [86, 617]]</point>"}]}
{"id": 2492, "image": ["rtx_frames_success_13/49_bridge#episode_3002/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3002/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red cone."}, {"from": "gpt", "value": "<point>[[383, 63], [504, 160], [639, 306], [722, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue spoon."}, {"from": "gpt", "value": "<point>[[712, 251], [532, 251], [350, 281], [220, 328]]</point>"}]}
{"id": 2494, "image": ["rtx_frames_success_13/49_bridge#episode_3005/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3008/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the metal pot on the blue towel."}, {"from": "gpt", "value": "<point>[[425, 185], [466, 253], [532, 494], [528, 589]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spice container."}, {"from": "gpt", "value": "<point>[[233, 211], [316, 306], [440, 479], [506, 591]]</point>"}]}
{"id": 2496, "image": ["rtx_frames_success_13/49_bridge#episode_3009/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3010/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[445, 216], [389, 286], [306, 411], [258, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue fork."}, {"from": "gpt", "value": "<point>[[286, 175], [427, 263], [662, 466], [773, 594]]</point>"}]}
{"id": 2498, "image": ["rtx_frames_success_13/49_bridge#episode_3018/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3019/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[164, 5], [227, 143], [387, 318], [526, 414]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the red and white item."}, {"from": "gpt", "value": "<point>[[763, 291], [684, 348], [607, 426], [530, 511]]</point>"}]}
{"id": 2500, "image": ["rtx_frames_success_13/49_bridge#episode_3026/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3026/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[350, 90], [434, 180], [626, 308], [767, 358]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to open the microwave door."}, {"from": "gpt", "value": "<point>[[720, 353], [665, 409], [570, 451], [479, 466]]</point>"}]}
{"id": 2502, "image": ["rtx_frames_success_13/49_bridge#episode_3031/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3033/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[462, 123], [560, 251], [658, 416], [726, 574]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[624, 163], [506, 246], [355, 396], [242, 521]]</point>"}]}
{"id": 2504, "image": ["rtx_frames_success_13/49_bridge#episode_3039/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3045/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[353, 143], [468, 238], [618, 414], [716, 559]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[408, 303], [415, 446], [451, 596], [491, 702]]</point>"}]}
{"id": 2506, "image": ["rtx_frames_success_13/49_bridge#episode_3047/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3047/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[284, 65], [273, 163], [303, 336], [338, 411]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the bottom right corner of the table."}, {"from": "gpt", "value": "<point>[[280, 318], [417, 386], [617, 544], [690, 662]]</point>"}]}
{"id": 2508, "image": ["rtx_frames_success_13/49_bridge#episode_3049/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3059/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush towards the napkin."}, {"from": "gpt", "value": "<point>[[348, 459], [476, 486], [620, 564], [720, 692]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the eggplant in the bowl."}, {"from": "gpt", "value": "<point>[[645, 188], [632, 281], [639, 373], [656, 481]]</point>"}]}
{"id": 2510, "image": ["rtx_frames_success_13/49_bridge#episode_3074/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3074/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[754, 143], [618, 175], [474, 248], [370, 346]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[378, 276], [481, 301], [635, 381], [712, 444]]</point>"}]}
{"id": 2512, "image": ["rtx_frames_success_13/49_bridge#episode_3080/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3105/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[406, 100], [312, 180], [211, 308], [135, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[470, 5], [504, 251], [556, 464], [647, 724]]</point>"}]}
{"id": 2514, "image": ["rtx_frames_success_13/49_bridge#episode_3109/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3114/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[553, 103], [566, 193], [564, 261], [545, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[588, 206], [553, 351], [538, 549], [547, 692]]</point>"}]}
{"id": 2516, "image": ["rtx_frames_success_13/49_bridge#episode_3116/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3126/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[55, 281], [259, 409], [489, 584], [665, 754]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red spoon."}, {"from": "gpt", "value": "<point>[[492, 175], [553, 296], [603, 499], [620, 657]]</point>"}]}
{"id": 2518, "image": ["rtx_frames_success_13/49_bridge#episode_3126/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3142/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the tomato slice."}, {"from": "gpt", "value": "<point>[[648, 564], [615, 486], [562, 406], [508, 356]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up a cucumber in the pot."}, {"from": "gpt", "value": "<point>[[470, 348], [541, 336], [594, 341], [635, 358]]</point>"}]}
{"id": 2520, "image": ["rtx_frames_success_13/49_bridge#episode_3142/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3158/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the can on the table."}, {"from": "gpt", "value": "<point>[[622, 346], [566, 296], [483, 298], [415, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn in the pan."}, {"from": "gpt", "value": "<point>[[481, 273], [400, 288], [308, 331], [224, 404]]</point>"}]}
{"id": 2522, "image": ["rtx_frames_success_13/49_bridge#episode_3158/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3197/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the corn on the towel."}, {"from": "gpt", "value": "<point>[[280, 211], [444, 248], [577, 293], [742, 383]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[602, 224], [605, 336], [605, 475], [625, 680]]</point>"}]}
{"id": 2524, "image": ["rtx_frames_success_13/49_bridge#episode_3197/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3208/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush towards the microwave."}, {"from": "gpt", "value": "<point>[[672, 432], [649, 390], [579, 301], [504, 278], [400, 270]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[355, 73], [487, 180], [624, 318], [718, 439]]</point>"}]}
{"id": 2526, "image": ["rtx_frames_success_13/49_bridge#episode_3208/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3212/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bowl between the mushroom and the brush."}, {"from": "gpt", "value": "<point>[[494, 211], [425, 251], [370, 308], [333, 391]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[506, 85], [586, 273], [639, 486], [673, 747]]</point>"}]}
{"id": 2528, "image": ["rtx_frames_success_13/49_bridge#episode_3214/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3222/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[562, 93], [582, 193], [597, 290], [608, 409], [623, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the fruits."}, {"from": "gpt", "value": "<point>[[188, 131], [353, 112], [463, 154], [582, 293], [678, 510]]</point>"}]}
{"id": 2530, "image": ["rtx_frames_success_13/49_bridge#episode_3224/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3226/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[371, 181], [394, 286], [411, 324], [460, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the strawberry."}, {"from": "gpt", "value": "<point>[[657, 394], [625, 432], [611, 502], [605, 583]]</point>"}]}
{"id": 2532, "image": ["rtx_frames_success_13/49_bridge#episode_3226/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3229/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the strawberry on the plate."}, {"from": "gpt", "value": "<point>[[686, 263], [634, 243], [556, 259], [475, 313], [434, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the green object."}, {"from": "gpt", "value": "<point>[[440, 378], [568, 402], [628, 525], [652, 629], [652, 745]]</point>"}]}
{"id": 2534, "image": ["rtx_frames_success_13/49_bridge#episode_3233/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3246/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[379, 77], [521, 162], [602, 263], [689, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pan."}, {"from": "gpt", "value": "<point>[[495, 409], [440, 340], [359, 286], [313, 324]]</point>"}]}
{"id": 2536, "image": ["rtx_frames_success_13/49_bridge#episode_3246/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3252/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan towards the cloth."}, {"from": "gpt", "value": "<point>[[342, 259], [255, 270], [180, 405], [200, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[374, 116], [400, 236], [431, 347], [460, 413]]</point>"}]}
{"id": 2538, "image": ["rtx_frames_success_13/49_bridge#episode_3252/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3256/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the orange towards the cloth."}, {"from": "gpt", "value": "<point>[[379, 193], [319, 181], [220, 208], [180, 293], [177, 386]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red spatula."}, {"from": "gpt", "value": "<point>[[568, 104], [614, 208], [646, 313], [669, 402], [692, 544], [701, 656]]</point>"}]}
{"id": 2540, "image": ["rtx_frames_success_13/49_bridge#episode_3257/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3257/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[405, 154], [542, 212], [617, 320], [689, 432], [718, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the orange cloth."}, {"from": "gpt", "value": "<point>[[721, 479], [631, 405], [533, 363], [426, 390]]</point>"}]}
{"id": 2542, "image": ["rtx_frames_success_13/49_bridge#episode_3265/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3267/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[689, 479], [623, 402], [513, 317], [368, 286], [261, 274], [185, 297]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[469, 390], [573, 425], [640, 486], [744, 606]]</point>"}]}
{"id": 2544, "image": ["rtx_frames_success_13/49_bridge#episode_3273/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3275/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow towel."}, {"from": "gpt", "value": "<point>[[346, 123], [338, 178], [353, 263], [374, 311]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[684, 60], [571, 95], [462, 190], [387, 326]]</point>"}]}
{"id": 2546, "image": ["rtx_frames_success_13/49_bridge#episode_3278/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3284/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[648, 456], [566, 466], [513, 479], [427, 511]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[564, 361], [447, 358], [308, 381], [162, 434]]</point>"}]}
{"id": 2548, "image": ["rtx_frames_success_13/49_bridge#episode_3284/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3289/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the pot."}, {"from": "gpt", "value": "<point>[[145, 366], [254, 373], [385, 356], [476, 326]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[361, 286], [395, 338], [464, 386], [524, 398]]</point>"}]}
{"id": 2550, "image": ["rtx_frames_success_13/49_bridge#episode_3294/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3298/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[484, 143], [403, 232], [350, 351], [327, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cucumber."}, {"from": "gpt", "value": "<point>[[438, 23], [474, 206], [489, 411], [479, 682]]</point>"}]}
{"id": 2552, "image": ["rtx_frames_success_13/49_bridge#episode_3304/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3304/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[261, 383], [239, 419], [201, 441], [162, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the pot."}, {"from": "gpt", "value": "<point>[[164, 404], [254, 424], [344, 471], [415, 551]]</point>"}]}
{"id": 2554, "image": ["rtx_frames_success_13/49_bridge#episode_3306/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3324/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[429, 185], [402, 271], [402, 421], [419, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[494, 125], [524, 206], [665, 388], [756, 461]]</point>"}]}
{"id": 2556, "image": ["rtx_frames_success_13/49_bridge#episode_3341/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3342/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[549, 256], [489, 351], [457, 431], [415, 634]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[455, 125], [539, 193], [609, 308], [643, 446]]</point>"}]}
{"id": 2558, "image": ["rtx_frames_success_13/49_bridge#episode_3370/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3386/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the saucepan."}, {"from": "gpt", "value": "<point>[[259, 195], [340, 268], [481, 343], [590, 373]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[265, 113], [400, 175], [568, 313], [658, 444]]</point>"}]}
{"id": 2560, "image": ["rtx_frames_success_13/49_bridge#episode_3395/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3395/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the square."}, {"from": "gpt", "value": "<point>[[842, 75], [684, 168], [444, 361], [321, 516]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the square above the fork."}, {"from": "gpt", "value": "<point>[[278, 436], [331, 391], [419, 378], [487, 391]]</point>"}]}
{"id": 2562, "image": ["rtx_frames_success_13/49_bridge#episode_3399/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3400/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[214, 145], [291, 278], [370, 479], [397, 644]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[594, 211], [513, 286], [444, 378], [393, 504]]</point>"}]}
{"id": 2564, "image": ["rtx_frames_success_13/49_bridge#episode_3406/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3406/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[395, 271], [312, 301], [226, 366], [158, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the top left edge."}, {"from": "gpt", "value": "<point>[[152, 318], [216, 318], [286, 291], [327, 251]]</point>"}]}
{"id": 2566, "image": ["rtx_frames_success_13/49_bridge#episode_3411/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3416/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[590, 125], [519, 190], [447, 321], [425, 424]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[258, 145], [419, 276], [596, 459], [767, 684]]</point>"}]}
{"id": 2568, "image": ["rtx_frames_success_13/49_bridge#episode_3424/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3438/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[850, 446], [788, 431], [727, 436], [677, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[707, 211], [524, 301], [430, 346], [299, 444]]</point>"}]}
{"id": 2570, "image": ["rtx_frames_success_13/49_bridge#episode_3440/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3445/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[312, 376], [346, 441], [353, 486], [338, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[352, 308], [464, 411], [562, 561], [624, 732]]</point>"}]}
{"id": 2572, "image": ["rtx_frames_success_13/49_bridge#episode_3451/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3454/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot on the plate."}, {"from": "gpt", "value": "<point>[[404, 251], [468, 363], [528, 531], [570, 717]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[624, 303], [536, 273], [415, 261], [331, 268]]</point>"}]}
{"id": 2574, "image": ["rtx_frames_success_13/49_bridge#episode_3457/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3470/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan containing broccoli."}, {"from": "gpt", "value": "<point>[[368, 175], [506, 276], [607, 381], [699, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[547, 206], [566, 301], [617, 386], [711, 456]]</point>"}]}
{"id": 2576, "image": ["rtx_frames_success_13/49_bridge#episode_3472/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3482/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[259, 55], [229, 188], [231, 368], [269, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave door."}, {"from": "gpt", "value": "<point>[[218, 406], [179, 534], [288, 602], [453, 506]]</point>"}]}
{"id": 2578, "image": ["rtx_frames_success_13/49_bridge#episode_3492/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3496/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move clothes from the laundry machine to the basket."}, {"from": "gpt", "value": "<point>[[368, 88], [318, 251], [274, 451], [286, 667]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the vessel."}, {"from": "gpt", "value": "<point>[[489, 8], [504, 198], [530, 336], [588, 534]]</point>"}]}
{"id": 2580, "image": ["rtx_frames_success_13/49_bridge#episode_3498/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3511/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[420, 216], [443, 290], [460, 347], [507, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the fork."}, {"from": "gpt", "value": "<point>[[553, 479], [501, 413], [434, 378], [350, 402]]</point>"}]}
{"id": 2582, "image": ["rtx_frames_success_13/49_bridge#episode_3511/frame_15.png", "rtx_frames_success_13/49_bridge#episode_3515/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the fork towards the pot."}, {"from": "gpt", "value": "<point>[[298, 336], [458, 317], [576, 344], [681, 378], [756, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[625, 355], [504, 266], [443, 263], [359, 263], [321, 347]]</point>"}]}
{"id": 2584, "image": ["rtx_frames_success_13/49_bridge#episode_3528/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3528/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[307, 394], [397, 317], [449, 274], [544, 197]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull open the oven door."}, {"from": "gpt", "value": "<point>[[423, 278], [292, 351], [194, 378], [125, 394]]</point>"}]}
{"id": 2586, "image": ["rtx_frames_success_13/49_bridge#episode_3538/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3539/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab the orange."}, {"from": "gpt", "value": "<point>[[368, 290], [330, 398], [307, 436], [269, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice container."}, {"from": "gpt", "value": "<point>[[376, 263], [426, 340], [469, 417], [510, 525], [562, 687]]</point>"}]}
{"id": 2588, "image": ["rtx_frames_success_13/49_bridge#episode_3570/frame_0.png", "rtx_frames_success_13/49_bridge#episode_3570/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the wing."}, {"from": "gpt", "value": "<point>[[229, 359], [388, 347], [513, 359], [654, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the wing behind the eggplant."}, {"from": "gpt", "value": "<point>[[515, 239], [449, 274], [394, 309], [345, 355]]</point>"}]}
{"id": 2590, "image": ["rtx_frames_success_13/49_bridge#episode_3572/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25060/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[162, 239], [226, 259], [266, 266], [339, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[703, 190], [620, 198], [521, 241], [472, 306]]</point>"}]}
{"id": 2592, "image": ["rtx_frames_success_13/49_bridge#episode_25066/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25095/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[43, 288], [162, 368], [344, 541], [457, 687]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave door."}, {"from": "gpt", "value": "<point>[[237, 253], [383, 291], [485, 346], [556, 414]]</point>"}]}
{"id": 2594, "image": ["rtx_frames_success_13/49_bridge#episode_25110/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25122/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[305, 266], [361, 371], [432, 516], [470, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover above the sushi."}, {"from": "gpt", "value": "<point>[[669, 243], [523, 256], [383, 313], [282, 388]]</point>"}]}
{"id": 2596, "image": ["rtx_frames_success_13/49_bridge#episode_25149/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25152/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[594, 105], [635, 206], [682, 363], [699, 491]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cloth."}, {"from": "gpt", "value": "<point>[[201, 95], [244, 201], [288, 268], [359, 351]]</point>"}]}
{"id": 2598, "image": ["rtx_frames_success_13/49_bridge#episode_25186/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25197/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bread."}, {"from": "gpt", "value": "<point>[[280, 125], [380, 211], [511, 381], [577, 519]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[177, 135], [367, 223], [530, 343], [669, 491]]</point>"}]}
{"id": 2600, "image": ["rtx_frames_success_13/49_bridge#episode_25201/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25203/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[506, 283], [485, 381], [474, 496], [476, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue spoon."}, {"from": "gpt", "value": "<point>[[423, 251], [421, 373], [397, 519], [348, 657]]</point>"}]}
{"id": 2602, "image": ["rtx_frames_success_13/49_bridge#episode_25220/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25220/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow utensil."}, {"from": "gpt", "value": "<point>[[532, 261], [564, 341], [603, 406], [656, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the yellow utensil towards the lower left corner of the stovetop."}, {"from": "gpt", "value": "<point>[[684, 218], [555, 316], [378, 404], [197, 476]]</point>"}]}
{"id": 2604, "image": ["rtx_frames_success_13/49_bridge#episode_25228/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25247/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[406, 193], [402, 313], [414, 431], [434, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cheese."}, {"from": "gpt", "value": "<point>[[449, 386], [361, 376], [252, 386], [139, 414]]</point>"}]}
{"id": 2606, "image": ["rtx_frames_success_13/49_bridge#episode_25255/frame_15.png", "rtx_frames_success_13/49_bridge#episode_25257/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hold the cup in the air."}, {"from": "gpt", "value": "<point>[[494, 326], [545, 454], [573, 634], [571, 762]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[102, 218], [173, 223], [252, 198], [295, 158]]</point>"}]}
{"id": 2608, "image": ["rtx_frames_success_13/49_bridge#episode_25269/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25270/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[682, 409], [648, 499], [596, 571], [523, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup on the plate."}, {"from": "gpt", "value": "<point>[[312, 281], [436, 398], [553, 586], [620, 752]]</point>"}]}
{"id": 2610, "image": ["rtx_frames_success_13/49_bridge#episode_25272/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25273/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[276, 283], [357, 371], [466, 434], [575, 469]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[528, 201], [449, 231], [383, 283], [312, 358]]</point>"}]}
{"id": 2612, "image": ["rtx_frames_success_13/49_bridge#episode_25273/frame_15.png", "rtx_frames_success_13/49_bridge#episode_25286/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon to the right."}, {"from": "gpt", "value": "<point>[[320, 213], [365, 378], [432, 511], [570, 679]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[583, 318], [583, 386], [586, 481], [603, 541]]</point>"}]}
{"id": 2614, "image": ["rtx_frames_success_13/49_bridge#episode_25286/frame_15.png", "rtx_frames_success_13/49_bridge#episode_25287/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush on the left of the pot."}, {"from": "gpt", "value": "<point>[[539, 246], [449, 266], [389, 306], [325, 378]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[216, 266], [350, 381], [508, 569], [598, 732]]</point>"}]}
{"id": 2616, "image": ["rtx_frames_success_13/49_bridge#episode_25290/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25295/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot on the plate."}, {"from": "gpt", "value": "<point>[[207, 496], [368, 529], [543, 594], [665, 662]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[624, 0], [624, 123], [673, 301], [767, 426]]</point>"}]}
{"id": 2618, "image": ["rtx_frames_success_13/49_bridge#episode_25311/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25320/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[320, 246], [346, 356], [359, 431], [365, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[496, 78], [504, 283], [549, 546], [605, 734]]</point>"}]}
{"id": 2620, "image": ["rtx_frames_success_13/49_bridge#episode_25339/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25345/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[577, 8], [590, 163], [630, 323], [680, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to upright the spice bottle."}, {"from": "gpt", "value": "<point>[[329, 216], [385, 175], [430, 208], [459, 286]]</point>"}]}
{"id": 2622, "image": ["rtx_frames_success_13/49_bridge#episode_25349/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25349/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[415, 276], [385, 316], [355, 388], [340, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the pot."}, {"from": "gpt", "value": "<point>[[336, 331], [444, 441], [562, 581], [648, 729]]</point>"}]}
{"id": 2624, "image": ["rtx_frames_success_13/49_bridge#episode_25350/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25362/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[545, 429], [474, 406], [383, 398], [312, 409]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[387, 130], [410, 281], [466, 471], [530, 604]]</point>"}]}
{"id": 2626, "image": ["rtx_frames_success_13/49_bridge#episode_25364/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25366/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[258, 216], [329, 306], [389, 426], [432, 556]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[320, 60], [408, 173], [517, 356], [575, 501]]</point>"}]}
{"id": 2628, "image": ["rtx_frames_success_13/49_bridge#episode_25366/frame_15.png", "rtx_frames_success_13/49_bridge#episode_25373/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pot on the green pot."}, {"from": "gpt", "value": "<point>[[577, 311], [483, 378], [402, 461], [346, 554]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the rag."}, {"from": "gpt", "value": "<point>[[496, 143], [461, 308], [440, 496], [436, 657]]</point>"}]}
{"id": 2630, "image": ["rtx_frames_success_13/49_bridge#episode_25396/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25416/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the oven door."}, {"from": "gpt", "value": "<point>[[372, 414], [293, 371], [184, 396], [71, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[530, 331], [609, 469], [673, 642], [722, 799]]</point>"}]}
{"id": 2632, "image": ["rtx_frames_success_13/49_bridge#episode_25419/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25423/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[680, 70], [564, 193], [434, 338], [338, 481]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[524, 120], [397, 246], [273, 409], [216, 541]]</point>"}]}
{"id": 2634, "image": ["rtx_frames_success_13/49_bridge#episode_25423/frame_15.png", "rtx_frames_success_13/49_bridge#episode_25435/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon to the right."}, {"from": "gpt", "value": "<point>[[141, 398], [312, 411], [600, 506], [739, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[363, 123], [423, 236], [472, 386], [502, 529]]</point>"}]}
{"id": 2636, "image": ["rtx_frames_success_13/49_bridge#episode_25442/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25442/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cardboard fence."}, {"from": "gpt", "value": "<point>[[246, 10], [177, 150], [105, 313], [56, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cardboard fence."}, {"from": "gpt", "value": "<point>[[81, 388], [141, 336], [244, 306], [305, 326]]</point>"}]}
{"id": 2638, "image": ["rtx_frames_success_13/49_bridge#episode_25450/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25455/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[143, 135], [190, 256], [261, 404], [335, 516]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[521, 612], [385, 564], [173, 566], [34, 622]]</point>"}]}
{"id": 2640, "image": ["rtx_frames_success_13/49_bridge#episode_25459/frame_0.png", "rtx_frames_success_13/49_bridge#episode_25459/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the can."}, {"from": "gpt", "value": "<point>[[135, 198], [256, 198], [423, 233], [511, 271]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the can on the table."}, {"from": "gpt", "value": "<point>[[541, 316], [560, 411], [568, 539], [547, 649]]</point>"}]}
{"id": 2642, "image": ["rtx_frames_success_14/49_bridge#episode_49/frame_15.png", "rtx_frames_success_14/49_bridge#episode_378/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cucumber towards the bowl."}, {"from": "gpt", "value": "<point>[[624, 702], [539, 534], [447, 365], [367, 185]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blueberries."}, {"from": "gpt", "value": "<point>[[459, 618], [493, 539], [522, 433], [581, 275]]</point>"}]}
{"id": 2644, "image": ["rtx_frames_success_14/49_bridge#episode_397/frame_0.png", "rtx_frames_success_14/49_bridge#episode_416/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[632, 753], [607, 539], [569, 320], [514, 73]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the play food."}, {"from": "gpt", "value": "<point>[[683, 360], [548, 320], [426, 292], [329, 197]]</point>"}]}
{"id": 2646, "image": ["rtx_frames_success_14/49_bridge#episode_416/frame_15.png", "rtx_frames_success_14/49_bridge#episode_429/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the play food towards the area beneath the microwave."}, {"from": "gpt", "value": "<point>[[320, 247], [430, 247], [556, 270], [674, 219]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the container with sushi."}, {"from": "gpt", "value": "<point>[[249, 494], [346, 433], [447, 298], [485, 124]]</point>"}]}
{"id": 2648, "image": ["rtx_frames_success_14/49_bridge#episode_441/frame_0.png", "rtx_frames_success_14/49_bridge#episode_441/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red bottle."}, {"from": "gpt", "value": "<point>[[177, 404], [282, 455], [388, 461], [501, 478]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the plate towards the spice container."}, {"from": "gpt", "value": "<point>[[362, 691], [291, 534], [228, 404], [135, 213]]</point>"}]}
{"id": 2650, "image": ["rtx_frames_success_14/49_bridge#episode_472/frame_0.png", "rtx_frames_success_14/49_bridge#episode_474/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[590, 534], [455, 461], [320, 382], [194, 264]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[341, 343], [455, 287], [535, 174], [577, 17]]</point>"}]}
{"id": 2652, "image": ["rtx_frames_success_14/49_bridge#episode_474/frame_15.png", "rtx_frames_success_14/49_bridge#episode_477/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife."}, {"from": "gpt", "value": "<point>[[746, 556], [683, 438], [594, 315], [522, 157]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli in the bowl."}, {"from": "gpt", "value": "<point>[[324, 489], [388, 416], [438, 287], [447, 152]]</point>"}]}
{"id": 2654, "image": ["rtx_frames_success_14/49_bridge#episode_488/frame_0.png", "rtx_frames_success_14/49_bridge#episode_488/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the broccoli from the bowl."}, {"from": "gpt", "value": "<point>[[354, 287], [472, 337], [598, 337], [729, 253]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the broccoli towards the yellow cloth."}, {"from": "gpt", "value": "<point>[[716, 702], [531, 534], [421, 382], [295, 163]]</point>"}]}
{"id": 2656, "image": ["rtx_frames_success_14/49_bridge#episode_500/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3610/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[451, 545], [388, 393], [350, 275], [287, 129]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[506, 680], [506, 556], [472, 427], [455, 309]]</point>"}]}
{"id": 2658, "image": ["rtx_frames_success_14/49_bridge#episode_3618/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3628/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pumpkin."}, {"from": "gpt", "value": "<point>[[291, 416], [472, 343], [632, 275], [860, 152]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[404, 320], [463, 242], [514, 118], [552, 22]]</point>"}]}
{"id": 2660, "image": ["rtx_frames_success_14/49_bridge#episode_3634/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3641/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot containing the beet."}, {"from": "gpt", "value": "<point>[[796, 500], [674, 365], [573, 242], [434, 67]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[409, 326], [409, 247], [404, 152], [392, 28]]</point>"}]}
{"id": 2662, "image": ["rtx_frames_success_14/49_bridge#episode_3650/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3655/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the potato."}, {"from": "gpt", "value": "<point>[[708, 466], [611, 388], [506, 298], [375, 185]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[581, 399], [510, 309], [447, 197], [371, 84]]</point>"}]}
{"id": 2664, "image": ["rtx_frames_success_14/49_bridge#episode_3673/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3688/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[754, 444], [670, 371], [531, 281], [379, 163]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[312, 539], [379, 376], [404, 270], [413, 140]]</point>"}]}
{"id": 2666, "image": ["rtx_frames_success_14/49_bridge#episode_3688/frame_15.png", "rtx_frames_success_14/49_bridge#episode_3715/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush towards the cloth."}, {"from": "gpt", "value": "<point>[[716, 348], [632, 421], [539, 444], [447, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the croissant in the bowl."}, {"from": "gpt", "value": "<point>[[531, 652], [430, 511], [329, 360], [202, 253]]</point>"}]}
{"id": 2668, "image": ["rtx_frames_success_14/49_bridge#episode_3738/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3744/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the chicken wing."}, {"from": "gpt", "value": "<point>[[624, 708], [527, 556], [447, 416], [341, 247]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move pot towards stove."}, {"from": "gpt", "value": "<point>[[518, 197], [628, 202], [716, 208], [817, 129]]</point>"}]}
{"id": 2670, "image": ["rtx_frames_success_14/49_bridge#episode_3745/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3745/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the tomato."}, {"from": "gpt", "value": "<point>[[826, 489], [838, 393], [851, 281], [885, 152]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the red pepper."}, {"from": "gpt", "value": "<point>[[270, 427], [367, 309], [438, 213], [531, 79]]</point>"}]}
{"id": 2672, "image": ["rtx_frames_success_14/49_bridge#episode_3758/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3763/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[581, 601], [581, 455], [590, 264], [594, 73]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[354, 522], [438, 438], [527, 309], [581, 152]]</point>"}]}
{"id": 2674, "image": ["rtx_frames_success_14/49_bridge#episode_3777/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3781/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[291, 449], [367, 444], [476, 433], [598, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bread."}, {"from": "gpt", "value": "<point>[[413, 461], [510, 433], [628, 382], [729, 320]]</point>"}]}
{"id": 2676, "image": ["rtx_frames_success_14/49_bridge#episode_3789/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3790/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the bowl."}, {"from": "gpt", "value": "<point>[[611, 343], [514, 264], [438, 185], [362, 73]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green cloth."}, {"from": "gpt", "value": "<point>[[463, 590], [430, 421], [379, 287], [358, 112]]</point>"}]}
{"id": 2678, "image": ["rtx_frames_success_14/49_bridge#episode_3797/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3798/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[657, 646], [522, 545], [375, 455], [244, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cup on the plate."}, {"from": "gpt", "value": "<point>[[379, 421], [312, 376], [244, 298], [185, 185]]</point>"}]}
{"id": 2680, "image": ["rtx_frames_success_14/49_bridge#episode_3800/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3800/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[750, 365], [708, 270], [666, 101], [632, 0]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[396, 522], [341, 343], [312, 202], [320, 39]]</point>"}]}
{"id": 2682, "image": ["rtx_frames_success_14/49_bridge#episode_3831/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3873/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[590, 635], [649, 522], [683, 404], [725, 197]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[463, 376], [316, 360], [236, 309], [126, 292]]</point>"}]}
{"id": 2684, "image": ["rtx_frames_success_14/49_bridge#episode_3873/frame_15.png", "rtx_frames_success_14/49_bridge#episode_3882/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the oven door closed."}, {"from": "gpt", "value": "<point>[[472, 264], [632, 309], [822, 376], [927, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the crescent roll."}, {"from": "gpt", "value": "<point>[[636, 719], [573, 579], [510, 472], [426, 253]]</point>"}]}
{"id": 2686, "image": ["rtx_frames_success_14/49_bridge#episode_3947/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3950/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[404, 494], [438, 421], [438, 309], [438, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the vessel."}, {"from": "gpt", "value": "<point>[[295, 326], [371, 331], [451, 326], [510, 242]]</point>"}]}
{"id": 2688, "image": ["rtx_frames_success_14/49_bridge#episode_3950/frame_15.png", "rtx_frames_success_14/49_bridge#episode_3959/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the red ball towards the bottom right of the table."}, {"from": "gpt", "value": "<point>[[611, 702], [510, 590], [455, 455], [400, 292]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the tomato."}, {"from": "gpt", "value": "<point>[[270, 421], [350, 388], [413, 331], [430, 219]]</point>"}]}
{"id": 2690, "image": ["rtx_frames_success_14/49_bridge#episode_3963/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3964/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[371, 258], [497, 326], [607, 320], [721, 264]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[295, 534], [282, 410], [270, 303], [249, 185]]</point>"}]}
{"id": 2692, "image": ["rtx_frames_success_14/49_bridge#episode_3965/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3983/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[619, 517], [699, 444], [754, 354], [826, 219]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[666, 624], [522, 494], [379, 421], [249, 287]]</point>"}]}
{"id": 2694, "image": ["rtx_frames_success_14/49_bridge#episode_3984/frame_0.png", "rtx_frames_success_14/49_bridge#episode_3992/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[603, 691], [569, 534], [539, 388], [514, 174]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[695, 629], [556, 455], [434, 287], [341, 84]]</point>"}]}
{"id": 2696, "image": ["rtx_frames_success_14/49_bridge#episode_3997/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4012/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[110, 449], [261, 421], [388, 388], [489, 315]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the blueberries."}, {"from": "gpt", "value": "<point>[[809, 713], [860, 629], [876, 556], [919, 466]]</point>"}]}
{"id": 2698, "image": ["rtx_frames_success_14/49_bridge#episode_4012/frame_15.png", "rtx_frames_success_14/49_bridge#episode_4013/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the blueberries on the plate."}, {"from": "gpt", "value": "<point>[[548, 275], [577, 208], [598, 163], [624, 79]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the grapes in the pan."}, {"from": "gpt", "value": "<point>[[607, 416], [489, 376], [362, 331], [240, 264]]</point>"}]}
{"id": 2700, "image": ["rtx_frames_success_14/49_bridge#episode_4021/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4041/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[388, 331], [303, 354], [219, 354], [101, 354]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven handle."}, {"from": "gpt", "value": "<point>[[708, 247], [590, 236], [463, 124], [337, 67]]</point>"}]}
{"id": 2702, "image": ["rtx_frames_success_14/49_bridge#episode_4041/frame_15.png", "rtx_frames_success_14/49_bridge#episode_4055/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the oven door open."}, {"from": "gpt", "value": "<point>[[156, 399], [270, 343], [333, 270], [421, 174]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[666, 466], [472, 399], [278, 320], [84, 242]]</point>"}]}
{"id": 2704, "image": ["rtx_frames_success_14/49_bridge#episode_4079/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4079/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[712, 388], [590, 331], [480, 292], [379, 197]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the mushroom in the pot."}, {"from": "gpt", "value": "<point>[[261, 242], [371, 258], [497, 264], [662, 287]]</point>"}]}
{"id": 2706, "image": ["rtx_frames_success_14/49_bridge#episode_4080/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4131/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red spoon."}, {"from": "gpt", "value": "<point>[[514, 607], [430, 511], [329, 399], [219, 225]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the fish to the right of the pot."}, {"from": "gpt", "value": "<point>[[708, 500], [598, 393], [480, 253], [400, 107]]</point>"}]}
{"id": 2708, "image": ["rtx_frames_success_14/49_bridge#episode_4145/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4145/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[708, 393], [577, 348], [472, 315], [358, 191]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot to the bottom left corner of the table."}, {"from": "gpt", "value": "<point>[[291, 511], [434, 455], [514, 337], [539, 191]]</point>"}]}
{"id": 2710, "image": ["rtx_frames_success_14/49_bridge#episode_4152/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4159/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green pepper."}, {"from": "gpt", "value": "<point>[[742, 28], [598, 135], [447, 163], [299, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[278, 511], [379, 421], [447, 309], [480, 140]]</point>"}]}
{"id": 2712, "image": ["rtx_frames_success_14/49_bridge#episode_4169/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4173/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green toy."}, {"from": "gpt", "value": "<point>[[510, 607], [535, 444], [548, 264], [581, 51]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[270, 455], [278, 382], [303, 303], [312, 202]]</point>"}]}
{"id": 2714, "image": ["rtx_frames_success_14/49_bridge#episode_4177/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4193/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up a piece of clothing."}, {"from": "gpt", "value": "<point>[[489, 112], [404, 191], [299, 270], [118, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[565, 343], [497, 236], [430, 152], [371, 17]]</point>"}]}
{"id": 2716, "image": ["rtx_frames_success_14/49_bridge#episode_4193/frame_15.png", "rtx_frames_success_14/49_bridge#episode_4203/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the spoon on the left corner of the table."}, {"from": "gpt", "value": "<point>[[190, 399], [324, 354], [417, 287], [544, 140]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[160, 388], [244, 382], [329, 343], [413, 213]]</point>"}]}
{"id": 2718, "image": ["rtx_frames_success_14/49_bridge#episode_4209/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4232/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the blue spoon."}, {"from": "gpt", "value": "<point>[[548, 657], [560, 500], [565, 331], [573, 174]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up a piece of clothing from the basket."}, {"from": "gpt", "value": "<point>[[447, 635], [375, 511], [320, 399], [282, 247]]</point>"}]}
{"id": 2720, "image": ["rtx_frames_success_14/49_bridge#episode_4236/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4240/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[400, 360], [472, 410], [548, 421], [607, 348]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[531, 646], [442, 478], [375, 298], [274, 118]]</point>"}]}
{"id": 2722, "image": ["rtx_frames_success_14/49_bridge#episode_4246/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4251/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[565, 264], [615, 315], [662, 365], [733, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[657, 466], [750, 433], [779, 309], [817, 163]]</point>"}]}
{"id": 2724, "image": ["rtx_frames_success_14/49_bridge#episode_4252/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4261/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the hot sauce bottle."}, {"from": "gpt", "value": "<point>[[421, 478], [548, 466], [645, 466], [767, 365]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[270, 489], [354, 371], [379, 275], [392, 140]]</point>"}]}
{"id": 2726, "image": ["rtx_frames_success_14/49_bridge#episode_4266/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4269/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the yellow cloth."}, {"from": "gpt", "value": "<point>[[228, 500], [426, 449], [544, 326], [590, 96]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[312, 545], [303, 427], [270, 298], [244, 163]]</point>"}]}
{"id": 2728, "image": ["rtx_frames_success_14/49_bridge#episode_4272/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4288/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[657, 669], [501, 556], [346, 455], [164, 242]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[598, 404], [489, 309], [409, 152], [295, 17]]</point>"}]}
{"id": 2730, "image": ["rtx_frames_success_14/49_bridge#episode_4295/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4306/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[291, 73], [219, 84], [143, 79], [63, 73]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the eggplant."}, {"from": "gpt", "value": "<point>[[447, 596], [354, 500], [253, 399], [194, 253]]</point>"}]}
{"id": 2732, "image": ["rtx_frames_success_14/49_bridge#episode_4319/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4319/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[708, 365], [598, 219], [522, 129], [438, 6]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[581, 601], [645, 455], [699, 331], [750, 219]]</point>"}]}
{"id": 2734, "image": ["rtx_frames_success_14/49_bridge#episode_4331/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4331/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[657, 298], [527, 230], [417, 157], [320, 39]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pot next to the towel."}, {"from": "gpt", "value": "<point>[[244, 506], [396, 438], [485, 337], [581, 202]]</point>"}]}
{"id": 2736, "image": ["rtx_frames_success_14/49_bridge#episode_4342/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4342/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[522, 444], [573, 444], [624, 438], [691, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot towards the towel."}, {"from": "gpt", "value": "<point>[[611, 331], [501, 298], [413, 258], [303, 213]]</point>"}]}
{"id": 2738, "image": ["rtx_frames_success_14/49_bridge#episode_4348/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4358/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[371, 680], [421, 517], [426, 343], [447, 79]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[653, 393], [522, 309], [396, 219], [249, 118]]</point>"}]}
{"id": 2740, "image": ["rtx_frames_success_14/49_bridge#episode_4358/frame_15.png", "rtx_frames_success_14/49_bridge#episode_4362/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to close the microwave door."}, {"from": "gpt", "value": "<point>[[666, 309], [581, 236], [501, 140], [434, 39]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position above the orange cloth."}, {"from": "gpt", "value": "<point>[[434, 466], [329, 404], [223, 331], [181, 213]]</point>"}]}
{"id": 2742, "image": ["rtx_frames_success_14/49_bridge#episode_4376/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4383/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red spoon."}, {"from": "gpt", "value": "<point>[[337, 433], [371, 303], [375, 174], [396, 39]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow bell pepper."}, {"from": "gpt", "value": "<point>[[177, 309], [303, 298], [430, 236], [510, 129]]</point>"}]}
{"id": 2744, "image": ["rtx_frames_success_14/49_bridge#episode_4389/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4391/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[198, 466], [337, 388], [455, 275], [607, 101]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the red object."}, {"from": "gpt", "value": "<point>[[303, 337], [375, 376], [430, 371], [506, 253]]</point>"}]}
{"id": 2746, "image": ["rtx_frames_success_14/49_bridge#episode_4398/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4398/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal container with the grey cube."}, {"from": "gpt", "value": "<point>[[152, 500], [270, 416], [392, 264], [518, 39]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal container with the grey cube."}, {"from": "gpt", "value": "<point>[[767, 483], [586, 455], [404, 410], [177, 354]]</point>"}]}
{"id": 2748, "image": ["rtx_frames_success_14/49_bridge#episode_4406/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4407/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot containing broccoli."}, {"from": "gpt", "value": "<point>[[603, 522], [459, 416], [362, 303], [278, 163]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[607, 433], [493, 382], [388, 326], [249, 253]]</point>"}]}
{"id": 2750, "image": ["rtx_frames_success_14/49_bridge#episode_4418/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4418/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange fruit."}, {"from": "gpt", "value": "<point>[[278, 478], [371, 399], [430, 253], [442, 79]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the orange fruit towards the bowl."}, {"from": "gpt", "value": "<point>[[518, 635], [506, 567], [489, 466], [438, 331]]</point>"}]}
{"id": 2752, "image": ["rtx_frames_success_14/49_bridge#episode_4438/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4438/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon from the bowl."}, {"from": "gpt", "value": "<point>[[775, 466], [611, 427], [472, 331], [278, 219]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon over the cloth."}, {"from": "gpt", "value": "<point>[[324, 410], [472, 455], [607, 455], [758, 421]]</point>"}]}
{"id": 2754, "image": ["rtx_frames_success_14/49_bridge#episode_4441/frame_15.png", "rtx_frames_success_14/49_bridge#episode_4445/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[506, 511], [438, 433], [375, 382], [312, 309]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the desired location."}, {"from": "gpt", "value": "<point>[[548, 624], [514, 489], [480, 382], [447, 236]]</point>"}]}
{"id": 2756, "image": ["rtx_frames_success_14/49_bridge#episode_4468/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4472/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[779, 343], [742, 242], [733, 146], [733, 56]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[826, 635], [678, 567], [510, 506], [274, 388]]</point>"}]}
{"id": 2758, "image": ["rtx_frames_success_14/49_bridge#episode_4475/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4491/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[754, 803], [683, 624], [615, 416], [548, 157]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the food across the surface."}, {"from": "gpt", "value": "<point>[[548, 584], [497, 444], [463, 337], [447, 197]]</point>"}]}
{"id": 2760, "image": ["rtx_frames_success_14/49_bridge#episode_4499/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4521/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[126, 506], [295, 393], [421, 270], [493, 96]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the stuffed animal."}, {"from": "gpt", "value": "<point>[[312, 309], [438, 287], [518, 219], [628, 73]]</point>"}]}
{"id": 2762, "image": ["rtx_frames_success_14/49_bridge#episode_4528/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4534/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the measuring cup."}, {"from": "gpt", "value": "<point>[[324, 517], [308, 376], [274, 264], [261, 118]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the plate."}, {"from": "gpt", "value": "<point>[[303, 399], [421, 416], [548, 438], [750, 410]]</point>"}]}
{"id": 2764, "image": ["rtx_frames_success_14/49_bridge#episode_4534/frame_15.png", "rtx_frames_success_14/49_bridge#episode_4536/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the apple on the plate."}, {"from": "gpt", "value": "<point>[[282, 455], [295, 315], [291, 202], [278, 67]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[683, 466], [662, 320], [632, 208], [666, 39]]</point>"}]}
{"id": 2766, "image": ["rtx_frames_success_14/49_bridge#episode_4577/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4579/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[573, 522], [506, 404], [434, 287], [346, 101]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the metal pot on the purple towel."}, {"from": "gpt", "value": "<point>[[649, 669], [657, 511], [678, 365], [699, 208]]</point>"}]}
{"id": 2768, "image": ["rtx_frames_success_14/49_bridge#episode_4585/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4588/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[758, 472], [763, 360], [775, 242], [775, 96]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cheese."}, {"from": "gpt", "value": "<point>[[173, 522], [253, 478], [265, 376], [244, 225]]</point>"}]}
{"id": 2770, "image": ["rtx_frames_success_14/49_bridge#episode_4601/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4608/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[261, 466], [367, 365], [434, 264], [472, 107]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[278, 522], [278, 444], [278, 343], [274, 208]]</point>"}]}
{"id": 2772, "image": ["rtx_frames_success_14/49_bridge#episode_4610/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4613/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[438, 596], [497, 455], [510, 309], [539, 118]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[228, 348], [371, 309], [518, 219], [628, 34]]</point>"}]}
{"id": 2774, "image": ["rtx_frames_success_14/49_bridge#episode_4642/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4647/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[489, 393], [421, 365], [346, 309], [312, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[666, 545], [632, 365], [603, 230], [556, 62]]</point>"}]}
{"id": 2776, "image": ["rtx_frames_success_14/49_bridge#episode_4657/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4657/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[341, 427], [404, 348], [442, 247], [472, 118]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position over the sushi."}, {"from": "gpt", "value": "<point>[[640, 478], [548, 478], [451, 444], [354, 410]]</point>"}]}
{"id": 2778, "image": ["rtx_frames_success_14/49_bridge#episode_4662/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4689/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[657, 393], [657, 264], [653, 146], [653, 39]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the spatula."}, {"from": "gpt", "value": "<point>[[312, 551], [270, 421], [223, 298], [219, 152]]</point>"}]}
{"id": 2780, "image": ["rtx_frames_success_14/49_bridge#episode_4736/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4759/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the hot dog bun."}, {"from": "gpt", "value": "<point>[[240, 382], [295, 388], [346, 388], [358, 275]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange measuring cup."}, {"from": "gpt", "value": "<point>[[581, 449], [548, 354], [514, 253], [485, 152]]</point>"}]}
{"id": 2782, "image": ["rtx_frames_success_14/49_bridge#episode_4779/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4798/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[472, 618], [446, 459], [410, 294], [393, 118]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to close the microwave door."}, {"from": "gpt", "value": "<point>[[169, 410], [346, 388], [552, 298], [763, 96]]</point>"}]}
{"id": 2784, "image": ["rtx_frames_success_14/49_bridge#episode_4808/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4808/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cloth and the pot."}, {"from": "gpt", "value": "<point>[[581, 517], [463, 326], [379, 107], [341, 11]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot in front of the spoon."}, {"from": "gpt", "value": "<point>[[544, 506], [421, 500], [278, 466], [169, 455]]</point>"}]}
{"id": 2786, "image": ["rtx_frames_success_14/49_bridge#episode_4846/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4847/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[206, 461], [337, 410], [447, 343], [586, 185]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the strawberry."}, {"from": "gpt", "value": "<point>[[261, 393], [194, 331], [177, 219], [160, 101]]</point>"}]}
{"id": 2788, "image": ["rtx_frames_success_14/49_bridge#episode_4847/frame_15.png", "rtx_frames_success_14/49_bridge#episode_4872/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the strawberry on the purple towel."}, {"from": "gpt", "value": "<point>[[742, 489], [607, 399], [476, 287], [346, 157]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple cloth."}, {"from": "gpt", "value": "<point>[[581, 466], [480, 360], [400, 258], [337, 73]]</point>"}]}
{"id": 2790, "image": ["rtx_frames_success_14/49_bridge#episode_4889/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4894/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[742, 545], [733, 427], [716, 298], [716, 163]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the towel."}, {"from": "gpt", "value": "<point>[[249, 500], [375, 444], [501, 354], [565, 191]]</point>"}]}
{"id": 2792, "image": ["rtx_frames_success_14/49_bridge#episode_4899/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4912/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[413, 303], [501, 393], [586, 472], [699, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[640, 719], [678, 506], [708, 287], [712, 39]]</point>"}]}
{"id": 2794, "image": ["rtx_frames_success_14/49_bridge#episode_4913/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4945/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red spoon."}, {"from": "gpt", "value": "<point>[[721, 702], [560, 584], [379, 461], [198, 354]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[944, 635], [809, 596], [670, 534], [539, 478]]</point>"}]}
{"id": 2796, "image": ["rtx_frames_success_14/49_bridge#episode_4948/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4980/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[539, 393], [598, 287], [624, 180], [590, 22]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the vessel."}, {"from": "gpt", "value": "<point>[[303, 466], [434, 427], [556, 354], [645, 163]]</point>"}]}
{"id": 2798, "image": ["rtx_frames_success_14/49_bridge#episode_4991/frame_0.png", "rtx_frames_success_14/49_bridge#episode_4993/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[396, 354], [493, 287], [527, 163], [518, 28]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow spoon."}, {"from": "gpt", "value": "<point>[[590, 607], [573, 433], [556, 258], [544, 96]]</point>"}]}
{"id": 2800, "image": ["rtx_frames_success_14/49_bridge#episode_4993/frame_15.png", "rtx_frames_success_14/49_bridge#episode_5006/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the yellow spoon."}, {"from": "gpt", "value": "<point>[[468, 410], [527, 438], [598, 461], [666, 461]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[573, 427], [468, 348], [371, 247], [299, 135]]</point>"}]}
{"id": 2802, "image": ["rtx_frames_success_14/49_bridge#episode_5017/frame_0.png", "rtx_frames_success_15/49_bridge#episode_55/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot in the pot."}, {"from": "gpt", "value": "<point>[[447, 562], [413, 410], [358, 258], [287, 107]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[657, 229], [593, 580], [648, 646]]</point>"}]}
{"id": 2804, "image": ["rtx_frames_success_15/49_bridge#episode_58/frame_0.png", "rtx_frames_success_15/49_bridge#episode_64/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[313, 410], [529, 310], [609, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red fork."}, {"from": "gpt", "value": "<point>[[472, 81], [495, 243], [536, 336], [553, 386]]</point>"}]}
{"id": 2806, "image": ["rtx_frames_success_15/49_bridge#episode_516/frame_0.png", "rtx_frames_success_15/49_bridge#episode_524/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[404, 41], [496, 554], [562, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[307, 151], [454, 661], [593, 713]]</point>"}]}
{"id": 2808, "image": ["rtx_frames_success_15/49_bridge#episode_546/frame_0.png", "rtx_frames_success_15/49_bridge#episode_575/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[560, 203], [402, 417], [407, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[349, 288], [302, 462], [252, 495]]</point>"}]}
{"id": 2810, "image": ["rtx_frames_success_15/49_bridge#episode_582/frame_0.png", "rtx_frames_success_15/49_bridge#episode_582/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[202, 539], [452, 462], [546, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to close the microwave door."}, {"from": "gpt", "value": "<point>[[634, 414], [709, 188], [662, 70]]</point>"}]}
{"id": 2812, "image": ["rtx_frames_success_15/49_bridge#episode_585/frame_0.png", "rtx_frames_success_15/49_bridge#episode_593/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[418, 30], [302, 417], [402, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[446, 96], [482, 233], [421, 332]]</point>"}]}
{"id": 2814, "image": ["rtx_frames_success_15/49_bridge#episode_594/frame_0.png", "rtx_frames_success_15/49_bridge#episode_617/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the hot dog."}, {"from": "gpt", "value": "<point>[[504, 295], [471, 569], [385, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[554, 262], [435, 211], [404, 266]]</point>"}]}
{"id": 2816, "image": ["rtx_frames_success_15/49_bridge#episode_638/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5039/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[472, 286], [423, 351], [368, 456], [336, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door."}, {"from": "gpt", "value": "<point>[[169, 558], [235, 717], [332, 746]]</point>"}]}
{"id": 2818, "image": ["rtx_frames_success_15/49_bridge#episode_5039/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5068/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the cabinet door closed."}, {"from": "gpt", "value": "<point>[[537, 635], [792, 657], [859, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[452, 103], [488, 565], [551, 643]]</point>"}]}
{"id": 2820, "image": ["rtx_frames_success_15/49_bridge#episode_5076/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5076/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[382, 776], [180, 558], [224, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the center of the table."}, {"from": "gpt", "value": "<point>[[249, 484], [490, 532], [590, 635]]</point>"}]}
{"id": 2822, "image": ["rtx_frames_success_15/49_bridge#episode_5095/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5116/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[476, 299], [521, 473], [598, 550]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[648, 414], [482, 310], [429, 343]]</point>"}]}
{"id": 2824, "image": ["rtx_frames_success_15/49_bridge#episode_5122/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5126/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red fork."}, {"from": "gpt", "value": "<point>[[573, 332], [648, 532], [717, 550]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[219, 310], [529, 332], [637, 392]]</point>"}]}
{"id": 2826, "image": ["rtx_frames_success_15/49_bridge#episode_5129/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5137/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the rag."}, {"from": "gpt", "value": "<point>[[593, 59], [443, 403], [504, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[299, 262], [571, 521], [648, 473]]</point>"}]}
{"id": 2828, "image": ["rtx_frames_success_15/49_bridge#episode_5137/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5147/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to knock over the metal pot."}, {"from": "gpt", "value": "<point>[[665, 469], [299, 476], [202, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[571, 218], [424, 399], [427, 513]]</point>"}]}
{"id": 2830, "image": ["rtx_frames_success_15/49_bridge#episode_5155/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5159/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[283, 159], [648, 388], [720, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pear."}, {"from": "gpt", "value": "<point>[[521, 251], [476, 746], [579, 846]]</point>"}]}
{"id": 2832, "image": ["rtx_frames_success_15/49_bridge#episode_5174/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5174/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[349, 451], [255, 491], [188, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drop the broccoli into the pot."}, {"from": "gpt", "value": "<point>[[188, 491], [310, 410], [343, 351]]</point>"}]}
{"id": 2834, "image": ["rtx_frames_success_15/49_bridge#episode_5193/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5198/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[324, 295], [380, 595], [454, 643]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the brush."}, {"from": "gpt", "value": "<point>[[321, 225], [427, 421], [499, 443]]</point>"}]}
{"id": 2836, "image": ["rtx_frames_success_15/49_bridge#episode_5198/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5212/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the brush towards the towel."}, {"from": "gpt", "value": "<point>[[476, 355], [291, 373], [211, 428]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[429, 114], [463, 665], [501, 746]]</point>"}]}
{"id": 2838, "image": ["rtx_frames_success_15/49_bridge#episode_5214/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5216/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[330, 140], [427, 351], [416, 447]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower the arm towards the spatula."}, {"from": "gpt", "value": "<point>[[418, 314], [496, 742], [576, 816]]</point>"}]}
{"id": 2840, "image": ["rtx_frames_success_15/49_bridge#episode_5221/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5223/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot to the right of the purple towel."}, {"from": "gpt", "value": "<point>[[416, 358], [587, 502], [684, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pear."}, {"from": "gpt", "value": "<point>[[175, 144], [438, 436], [560, 436]]</point>"}]}
{"id": 2842, "image": ["rtx_frames_success_15/49_bridge#episode_5255/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5267/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[393, 779], [560, 310], [501, 126]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[643, 199], [446, 247], [366, 332]]</point>"}]}
{"id": 2844, "image": ["rtx_frames_success_15/49_bridge#episode_5268/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5281/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[607, 211], [485, 340], [457, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the box cutter."}, {"from": "gpt", "value": "<point>[[114, 89], [806, 761], [914, 720]]</point>"}]}
{"id": 2846, "image": ["rtx_frames_success_15/49_bridge#episode_5281/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5283/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the box cutter in the drawer."}, {"from": "gpt", "value": "<point>[[889, 746], [416, 517], [388, 403]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice container."}, {"from": "gpt", "value": "<point>[[227, 447], [438, 262], [521, 277]]</point>"}]}
{"id": 2848, "image": ["rtx_frames_success_15/49_bridge#episode_5299/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5299/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the corn."}, {"from": "gpt", "value": "<point>[[620, 310], [222, 473], [141, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the brush between the pan and the original location of the corn."}, {"from": "gpt", "value": "<point>[[305, 259], [474, 310], [546, 388]]</point>"}]}
{"id": 2850, "image": ["rtx_frames_success_15/49_bridge#episode_5305/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5311/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spatula."}, {"from": "gpt", "value": "<point>[[438, 528], [573, 362], [665, 388]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[155, 299], [338, 403], [355, 355]]</point>"}]}
{"id": 2852, "image": ["rtx_frames_success_15/49_bridge#episode_5313/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5321/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[202, 539], [355, 543], [413, 473]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[424, 369], [471, 576], [529, 613]]</point>"}]}
{"id": 2854, "image": ["rtx_frames_success_15/49_bridge#episode_5334/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5337/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[396, 251], [496, 698], [571, 705]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the fork."}, {"from": "gpt", "value": "<point>[[515, 81], [576, 661], [640, 717]]</point>"}]}
{"id": 2856, "image": ["rtx_frames_success_15/49_bridge#episode_5348/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5352/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[404, 351], [657, 451], [748, 580]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[429, 321], [404, 513], [346, 547]]</point>"}]}
{"id": 2858, "image": ["rtx_frames_success_15/49_bridge#episode_5359/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5384/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the right side of the table."}, {"from": "gpt", "value": "<point>[[374, 229], [560, 495], [648, 528]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[158, 107], [155, 399], [219, 458]]</point>"}]}
{"id": 2860, "image": ["rtx_frames_success_15/49_bridge#episode_5386/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5388/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[258, 222], [684, 277], [742, 384]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[219, 203], [371, 318], [407, 388]]</point>"}]}
{"id": 2862, "image": ["rtx_frames_success_15/49_bridge#episode_5391/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5407/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pepper."}, {"from": "gpt", "value": "<point>[[241, 107], [576, 543], [670, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[374, 366], [623, 624], [726, 591]]</point>"}]}
{"id": 2864, "image": ["rtx_frames_success_15/49_bridge#episode_5408/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5414/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[172, 351], [607, 281], [698, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the corn."}, {"from": "gpt", "value": "<point>[[479, 295], [330, 325], [271, 403]]</point>"}]}
{"id": 2866, "image": ["rtx_frames_success_15/49_bridge#episode_5414/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5418/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the corn on the stove."}, {"from": "gpt", "value": "<point>[[288, 377], [596, 395], [659, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[368, 185], [335, 284], [380, 347]]</point>"}]}
{"id": 2868, "image": ["rtx_frames_success_15/49_bridge#episode_5474/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5476/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana in the bowl."}, {"from": "gpt", "value": "<point>[[341, 140], [302, 480], [380, 550]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[435, 388], [554, 539], [645, 558]]</point>"}]}
{"id": 2870, "image": ["rtx_frames_success_15/49_bridge#episode_5494/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5509/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[474, 406], [665, 380], [715, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet handle."}, {"from": "gpt", "value": "<point>[[357, 569], [596, 561], [676, 613]]</point>"}]}
{"id": 2872, "image": ["rtx_frames_success_15/49_bridge#episode_5513/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5524/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[86, 100], [416, 447], [524, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the green object."}, {"from": "gpt", "value": "<point>[[360, 244], [551, 661], [665, 735]]</point>"}]}
{"id": 2874, "image": ["rtx_frames_success_15/49_bridge#episode_5542/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5544/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[418, 59], [582, 332], [587, 428]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[235, 318], [410, 561], [391, 632]]</point>"}]}
{"id": 2876, "image": ["rtx_frames_success_15/49_bridge#episode_5547/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5560/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[305, 170], [460, 646], [529, 694]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bottle on the table."}, {"from": "gpt", "value": "<point>[[717, 126], [809, 395], [801, 495]]</point>"}]}
{"id": 2878, "image": ["rtx_frames_success_15/49_bridge#episode_5562/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5569/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[175, 292], [371, 632], [515, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[175, 262], [615, 362], [701, 440]]</point>"}]}
{"id": 2880, "image": ["rtx_frames_success_15/49_bridge#episode_5570/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5600/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[601, 222], [449, 362], [371, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[269, 81], [219, 380], [269, 484]]</point>"}]}
{"id": 2882, "image": ["rtx_frames_success_15/49_bridge#episode_5610/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5616/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan containing the broccoli."}, {"from": "gpt", "value": "<point>[[393, 140], [488, 717], [562, 798]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bread."}, {"from": "gpt", "value": "<point>[[377, 310], [355, 561], [424, 609]]</point>"}]}
{"id": 2884, "image": ["rtx_frames_success_15/49_bridge#episode_5618/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5623/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pear."}, {"from": "gpt", "value": "<point>[[407, 281], [449, 358], [515, 369]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the wok."}, {"from": "gpt", "value": "<point>[[340, 120], [471, 546], [549, 590]]</point>"}]}
{"id": 2886, "image": ["rtx_frames_success_15/49_bridge#episode_5638/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5650/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the ear of corn across the table."}, {"from": "gpt", "value": "<point>[[343, 183], [512, 602], [601, 640]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[217, 251], [480, 459], [582, 454]]</point>"}]}
{"id": 2888, "image": ["rtx_frames_success_15/49_bridge#episode_5654/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5656/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spice jar."}, {"from": "gpt", "value": "<point>[[359, 155], [365, 667], [430, 754]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal bowl."}, {"from": "gpt", "value": "<point>[[343, 38], [444, 280], [452, 360]]</point>"}]}
{"id": 2890, "image": ["rtx_frames_success_15/49_bridge#episode_5656/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5659/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal bowl."}, {"from": "gpt", "value": "<point>[[427, 360], [242, 376], [185, 441]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue spoon."}, {"from": "gpt", "value": "<point>[[508, 392], [593, 457], [669, 403]]</point>"}]}
{"id": 2892, "image": ["rtx_frames_success_15/49_bridge#episode_5659/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5680/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the blue spoon on the left of the towel."}, {"from": "gpt", "value": "<point>[[629, 215], [274, 414], [165, 495]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[452, 237], [556, 586], [633, 624]]</point>"}]}
{"id": 2894, "image": ["rtx_frames_success_15/49_bridge#episode_5680/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5720/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pot of the table."}, {"from": "gpt", "value": "<point>[[371, 285], [270, 446], [266, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cucumber."}, {"from": "gpt", "value": "<point>[[500, 382], [403, 387], [351, 441]]</point>"}]}
{"id": 2896, "image": ["rtx_frames_success_15/49_bridge#episode_5724/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5729/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[471, 281], [299, 425], [277, 524]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[604, 155], [521, 284], [446, 310]]</point>"}]}
{"id": 2898, "image": ["rtx_frames_success_15/49_bridge#episode_5745/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5748/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[435, 155], [399, 310], [418, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[468, 425], [360, 539], [357, 617]]</point>"}]}
{"id": 2900, "image": ["rtx_frames_success_15/49_bridge#episode_5748/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5772/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot with the broccoli."}, {"from": "gpt", "value": "<point>[[352, 628], [479, 462], [482, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower the green capsicum towards the table."}, {"from": "gpt", "value": "<point>[[385, 225], [496, 414], [571, 443]]</point>"}]}
{"id": 2902, "image": ["rtx_frames_success_15/49_bridge#episode_5775/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5804/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[407, 329], [413, 613], [468, 676]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mushroom towards the pot."}, {"from": "gpt", "value": "<point>[[224, 129], [432, 680], [604, 717]]</point>"}]}
{"id": 2904, "image": ["rtx_frames_success_15/49_bridge#episode_5815/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5825/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cheese wedge."}, {"from": "gpt", "value": "<point>[[296, 403], [150, 392], [83, 458]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[385, 192], [402, 561], [488, 620]]</point>"}]}
{"id": 2906, "image": ["rtx_frames_success_15/49_bridge#episode_5825/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5865/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the corn towards the pot."}, {"from": "gpt", "value": "<point>[[499, 639], [709, 484], [679, 421], [643, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[676, 362], [421, 321], [360, 358]]</point>"}]}
{"id": 2908, "image": ["rtx_frames_success_15/49_bridge#episode_5865/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5870/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan towards the upper right corner."}, {"from": "gpt", "value": "<point>[[380, 288], [695, 513], [809, 513]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot lid."}, {"from": "gpt", "value": "<point>[[416, 314], [535, 318], [535, 377]]</point>"}]}
{"id": 2910, "image": ["rtx_frames_success_15/49_bridge#episode_5872/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5878/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[648, 30], [291, 307], [222, 303]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[393, 225], [321, 465], [255, 480]]</point>"}]}
{"id": 2912, "image": ["rtx_frames_success_15/49_bridge#episode_5880/frame_15.png", "rtx_frames_success_15/49_bridge#episode_5892/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the broccoli towards the pot."}, {"from": "gpt", "value": "<point>[[346, 355], [191, 373], [144, 458]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spatula on the plate."}, {"from": "gpt", "value": "<point>[[537, 255], [457, 332], [465, 428]]</point>"}]}
{"id": 2914, "image": ["rtx_frames_success_15/49_bridge#episode_5894/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5894/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[258, 273], [681, 506], [762, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the can to the left."}, {"from": "gpt", "value": "<point>[[543, 351], [371, 347], [324, 273]]</point>"}]}
{"id": 2916, "image": ["rtx_frames_success_15/49_bridge#episode_5902/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5914/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[562, 266], [665, 580], [756, 639]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[260, 22], [263, 528], [332, 587]]</point>"}]}
{"id": 2918, "image": ["rtx_frames_success_15/49_bridge#episode_5916/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5926/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the fork."}, {"from": "gpt", "value": "<point>[[352, 78], [645, 403], [715, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the vessel."}, {"from": "gpt", "value": "<point>[[715, 244], [338, 303], [288, 362]]</point>"}]}
{"id": 2920, "image": ["rtx_frames_success_15/49_bridge#episode_5931/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5949/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[496, 288], [607, 632], [695, 654]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue fruit."}, {"from": "gpt", "value": "<point>[[305, 244], [208, 343], [202, 432]]</point>"}]}
{"id": 2922, "image": ["rtx_frames_success_15/49_bridge#episode_5975/frame_0.png", "rtx_frames_success_15/49_bridge#episode_5994/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue object."}, {"from": "gpt", "value": "<point>[[573, 251], [607, 657], [676, 776]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bowl."}, {"from": "gpt", "value": "<point>[[294, 236], [366, 347], [443, 388]]</point>"}]}
{"id": 2924, "image": ["rtx_frames_success_15/49_bridge#episode_5994/frame_15.png", "rtx_frames_success_15/49_bridge#episode_6001/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bowl."}, {"from": "gpt", "value": "<point>[[388, 288], [471, 469], [548, 532]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[493, 196], [271, 403], [213, 488]]</point>"}]}
{"id": 2926, "image": ["rtx_frames_success_15/49_bridge#episode_6006/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6020/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[546, 192], [233, 447], [125, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spice bottle."}, {"from": "gpt", "value": "<point>[[571, 347], [474, 351], [468, 447]]</point>"}]}
{"id": 2928, "image": ["rtx_frames_success_15/49_bridge#episode_6038/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6048/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[357, 358], [504, 321], [579, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[288, 343], [332, 506], [399, 554]]</point>"}]}
{"id": 2930, "image": ["rtx_frames_success_15/49_bridge#episode_6056/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6057/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pickle."}, {"from": "gpt", "value": "<point>[[454, 639], [668, 532], [648, 443]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pan containing the broccoli."}, {"from": "gpt", "value": "<point>[[421, 497], [559, 379], [599, 430]]</point>"}]}
{"id": 2932, "image": ["rtx_frames_success_15/49_bridge#episode_6057/frame_15.png", "rtx_frames_success_15/49_bridge#episode_6059/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan containing the broccoli."}, {"from": "gpt", "value": "<point>[[632, 432], [529, 547], [476, 513]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the salt shaker towards the back of the bowl."}, {"from": "gpt", "value": "<point>[[393, 399], [596, 414], [723, 499]]</point>"}]}
{"id": 2934, "image": ["rtx_frames_success_15/49_bridge#episode_6061/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6084/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[343, 218], [327, 484], [391, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[496, 199], [327, 528], [258, 517]]</point>"}]}
{"id": 2936, "image": ["rtx_frames_success_15/49_bridge#episode_6097/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6103/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cup on the plate."}, {"from": "gpt", "value": "<point>[[460, 270], [291, 314], [233, 266]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[637, 369], [440, 432], [402, 499]]</point>"}]}
{"id": 2938, "image": ["rtx_frames_success_15/49_bridge#episode_6105/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6108/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal vessel."}, {"from": "gpt", "value": "<point>[[438, 303], [537, 561], [593, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cheese."}, {"from": "gpt", "value": "<point>[[590, 222], [654, 432], [740, 414]]</point>"}]}
{"id": 2940, "image": ["rtx_frames_success_15/49_bridge#episode_6108/frame_15.png", "rtx_frames_success_15/49_bridge#episode_6109/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cheese."}, {"from": "gpt", "value": "<point>[[695, 366], [571, 661], [626, 813]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the eggplant."}, {"from": "gpt", "value": "<point>[[629, 602], [233, 469], [183, 373]]</point>"}]}
{"id": 2942, "image": ["rtx_frames_success_15/49_bridge#episode_6109/frame_15.png", "rtx_frames_success_15/49_bridge#episode_6123/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the can."}, {"from": "gpt", "value": "<point>[[388, 377], [604, 425], [681, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[271, 369], [609, 469], [704, 543]]</point>"}]}
{"id": 2944, "image": ["rtx_frames_success_15/49_bridge#episode_6139/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6139/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the eggplant from the pan."}, {"from": "gpt", "value": "<point>[[429, 295], [626, 362], [695, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the eggplant to the left corner."}, {"from": "gpt", "value": "<point>[[632, 406], [596, 646], [623, 724]]</point>"}]}
{"id": 2946, "image": ["rtx_frames_success_15/49_bridge#episode_6166/frame_15.png", "rtx_frames_success_15/49_bridge#episode_6169/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the lemon in the pot."}, {"from": "gpt", "value": "<point>[[510, 288], [659, 742], [731, 798]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the vessel."}, {"from": "gpt", "value": "<point>[[452, 458], [640, 414], [690, 476]]</point>"}]}
{"id": 2948, "image": ["rtx_frames_success_15/49_bridge#episode_6169/frame_15.png", "rtx_frames_success_15/49_bridge#episode_6172/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the vessel to another location on the table."}, {"from": "gpt", "value": "<point>[[512, 347], [277, 440], [216, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the towel."}, {"from": "gpt", "value": "<point>[[302, 255], [565, 513], [607, 628]]</point>"}]}
{"id": 2950, "image": ["rtx_frames_success_15/49_bridge#episode_6183/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6183/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the scoop."}, {"from": "gpt", "value": "<point>[[543, 236], [313, 499], [230, 524]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the scoop towards the pot."}, {"from": "gpt", "value": "<point>[[241, 377], [471, 329], [571, 377]]</point>"}]}
{"id": 2952, "image": ["rtx_frames_success_15/49_bridge#episode_6262/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6265/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the blue towel."}, {"from": "gpt", "value": "<point>[[551, 44], [413, 380], [307, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[294, 358], [485, 484], [529, 550]]</point>"}]}
{"id": 2954, "image": ["rtx_frames_success_15/49_bridge#episode_6284/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6291/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[391, 203], [343, 495], [377, 580]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[61, 151], [327, 425], [399, 417]]</point>"}]}
{"id": 2956, "image": ["rtx_frames_success_15/49_bridge#episode_6294/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6294/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[551, 536], [130, 329], [64, 340]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[371, 251], [471, 196], [479, 140]]</point>"}]}
{"id": 2958, "image": ["rtx_frames_success_15/49_bridge#episode_6295/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6297/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[266, 211], [371, 447], [446, 495]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[249, 196], [543, 502], [618, 476]]</point>"}]}
{"id": 2960, "image": ["rtx_frames_success_15/49_bridge#episode_6299/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6299/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[668, 325], [668, 495], [720, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the sushi to the other side of the table."}, {"from": "gpt", "value": "<point>[[706, 447], [499, 384], [438, 403]]</point>"}]}
{"id": 2962, "image": ["rtx_frames_success_15/49_bridge#episode_6320/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6322/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the overturned pot."}, {"from": "gpt", "value": "<point>[[526, 233], [485, 458], [521, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[457, 244], [548, 510], [632, 576]]</point>"}]}
{"id": 2964, "image": ["rtx_frames_success_15/49_bridge#episode_6338/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6345/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue object."}, {"from": "gpt", "value": "<point>[[665, 207], [460, 528], [490, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[156, 278], [243, 320], [313, 351], [371, 402]]</point>"}]}
{"id": 2966, "image": ["rtx_frames_success_15/49_bridge#episode_6349/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6362/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[443, 124], [429, 239], [423, 367], [414, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[886, 622], [819, 571], [715, 568], [591, 568], [469, 583]]</point>"}]}
{"id": 2968, "image": ["rtx_frames_success_15/49_bridge#episode_6392/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6430/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[307, 81], [403, 212], [469, 351], [550, 583]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[136, 375], [385, 324], [521, 363], [712, 456]]</point>"}]}
{"id": 2970, "image": ["rtx_frames_success_15/49_bridge#episode_6447/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6452/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the knife."}, {"from": "gpt", "value": "<point>[[171, 270], [240, 355], [330, 417], [429, 467], [478, 490]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the goldfish cracker."}, {"from": "gpt", "value": "<point>[[652, 317], [582, 290], [472, 263], [391, 247]]</point>"}]}
{"id": 2972, "image": ["rtx_frames_success_15/49_bridge#episode_6459/frame_0.png", "rtx_frames_success_15/49_bridge#episode_6476/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[724, 336], [721, 471], [738, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the napkin."}, {"from": "gpt", "value": "<point>[[704, 402], [683, 521], [660, 664]]</point>"}]}
{"id": 2974, "image": ["rtx_frames_success_15/49_bridge#episode_6481/frame_0.png", "rtx_frames_success_16/49_bridge#episode_68/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[449, 270], [475, 390], [489, 506], [498, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the potato on the right corner of the table."}, {"from": "gpt", "value": "<point>[[550, 286], [625, 347], [709, 479], [773, 772]]</point>"}]}
{"id": 2976, "image": ["rtx_frames_success_16/49_bridge#episode_70/frame_0.png", "rtx_frames_success_16/49_bridge#episode_70/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[460, 313], [527, 398], [582, 444], [669, 490]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spatula towards the bottom left corner."}, {"from": "gpt", "value": "<point>[[420, 313], [330, 378], [252, 429], [162, 425]]</point>"}]}
{"id": 2978, "image": ["rtx_frames_success_16/49_bridge#episode_72/frame_0.png", "rtx_frames_success_16/49_bridge#episode_72/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[324, 255], [255, 305], [211, 355], [182, 421]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the towel."}, {"from": "gpt", "value": "<point>[[130, 421], [217, 425], [330, 413], [431, 336]]</point>"}]}
{"id": 2980, "image": ["rtx_frames_success_16/49_bridge#episode_650/frame_0.png", "rtx_frames_success_16/49_bridge#episode_665/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[582, 69], [582, 135], [585, 236], [576, 351], [559, 448], [515, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[492, 46], [443, 139], [397, 236], [394, 413]]</point>"}]}
{"id": 2982, "image": ["rtx_frames_success_16/49_bridge#episode_671/frame_15.png", "rtx_frames_success_16/49_bridge#episode_694/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the potato away."}, {"from": "gpt", "value": "<point>[[489, 602], [420, 533], [371, 475], [321, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white object."}, {"from": "gpt", "value": "<point>[[81, 205], [243, 181], [374, 193], [507, 216], [718, 359]]</point>"}]}
{"id": 2984, "image": ["rtx_frames_success_16/49_bridge#episode_701/frame_0.png", "rtx_frames_success_16/49_bridge#episode_703/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[449, 247], [411, 290], [330, 344], [272, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the vegetable."}, {"from": "gpt", "value": "<point>[[611, 23], [565, 100], [481, 162], [408, 224], [353, 255]]</point>"}]}
{"id": 2986, "image": ["rtx_frames_success_16/49_bridge#episode_715/frame_0.png", "rtx_frames_success_16/49_bridge#episode_717/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position over the spatula."}, {"from": "gpt", "value": "<point>[[434, 174], [420, 236], [449, 293], [486, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[365, 193], [449, 274], [518, 328], [539, 402]]</point>"}]}
{"id": 2988, "image": ["rtx_frames_success_16/49_bridge#episode_739/frame_0.png", "rtx_frames_success_16/49_bridge#episode_739/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue spoon."}, {"from": "gpt", "value": "<point>[[510, 266], [539, 347], [576, 425], [588, 517], [597, 726]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the blue spoon towards the pot."}, {"from": "gpt", "value": "<point>[[359, 390], [301, 402], [258, 390], [235, 347]]</point>"}]}
{"id": 2990, "image": ["rtx_frames_success_16/49_bridge#episode_751/frame_0.png", "rtx_frames_success_16/49_bridge#episode_758/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[229, 205], [298, 239], [420, 270], [504, 317], [539, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[562, 147], [495, 228], [475, 336], [469, 448], [472, 568]]</point>"}]}
{"id": 2992, "image": ["rtx_frames_success_16/49_bridge#episode_760/frame_0.png", "rtx_frames_success_16/49_bridge#episode_760/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the mushroom."}, {"from": "gpt", "value": "<point>[[521, 525], [347, 452], [258, 417], [96, 386]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place into bowl."}, {"from": "gpt", "value": "<point>[[122, 347], [208, 336], [292, 409], [359, 525]]</point>"}]}
{"id": 2994, "image": ["rtx_frames_success_16/49_bridge#episode_764/frame_0.png", "rtx_frames_success_16/49_bridge#episode_768/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[264, 170], [278, 228], [301, 309], [298, 382], [266, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red object."}, {"from": "gpt", "value": "<point>[[446, 162], [492, 251], [568, 355], [628, 421], [695, 486], [744, 579]]</point>"}]}
{"id": 2996, "image": ["rtx_frames_success_16/49_bridge#episode_775/frame_0.png", "rtx_frames_success_16/49_bridge#episode_789/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cloth."}, {"from": "gpt", "value": "<point>[[324, 243], [388, 247], [475, 286], [513, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[359, 166], [452, 220], [515, 270], [550, 328], [579, 382], [599, 425]]</point>"}]}
{"id": 2998, "image": ["rtx_frames_success_16/49_bridge#episode_6515/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6522/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the blue spoon."}, {"from": "gpt", "value": "<point>[[562, 224], [484, 208], [385, 228], [304, 255], [206, 313]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[683, 432], [614, 452], [550, 475], [504, 560]]</point>"}]}
{"id": 3000, "image": ["rtx_frames_success_16/49_bridge#episode_6534/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6534/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[556, 421], [605, 440], [652, 471], [724, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[681, 521], [605, 486], [539, 486], [437, 490], [342, 552]]</point>"}]}
{"id": 3002, "image": ["rtx_frames_success_16/49_bridge#episode_6548/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6548/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli."}, {"from": "gpt", "value": "<point>[[223, 340], [347, 305], [417, 320], [489, 378]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move broccoli towards the pot."}, {"from": "gpt", "value": "<point>[[460, 336], [408, 328], [330, 336], [232, 378]]</point>"}]}
{"id": 3004, "image": ["rtx_frames_success_16/49_bridge#episode_6556/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6559/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the plush toy from the table."}, {"from": "gpt", "value": "<point>[[565, 390], [521, 336], [472, 282], [414, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the cloth."}, {"from": "gpt", "value": "<point>[[333, 181], [301, 247], [290, 347], [292, 490]]</point>"}]}
{"id": 3006, "image": ["rtx_frames_success_16/49_bridge#episode_6571/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6571/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the knife from the stove."}, {"from": "gpt", "value": "<point>[[510, 263], [449, 270], [339, 309], [249, 378], [188, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the pan."}, {"from": "gpt", "value": "<point>[[194, 405], [281, 448], [336, 467], [443, 417]]</point>"}]}
{"id": 3008, "image": ["rtx_frames_success_16/49_bridge#episode_6572/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6576/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spatula."}, {"from": "gpt", "value": "<point>[[472, 81], [556, 89], [620, 170], [654, 243], [681, 359], [698, 483]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[498, 479], [397, 479], [350, 475], [321, 521]]</point>"}]}
{"id": 3010, "image": ["rtx_frames_success_16/49_bridge#episode_6579/frame_15.png", "rtx_frames_success_16/49_bridge#episode_6597/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[814, 448], [738, 440], [634, 456], [568, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[356, 286], [440, 305], [466, 347], [489, 378]]</point>"}]}
{"id": 3012, "image": ["rtx_frames_success_16/49_bridge#episode_6635/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6640/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red object."}, {"from": "gpt", "value": "<point>[[327, 517], [466, 533], [573, 552], [672, 672]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[750, 386], [654, 320], [550, 313], [426, 344]]</point>"}]}
{"id": 3014, "image": ["rtx_frames_success_16/49_bridge#episode_6648/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6648/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the brush."}, {"from": "gpt", "value": "<point>[[376, 185], [319, 228], [237, 317], [208, 483]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the brush towards the corner."}, {"from": "gpt", "value": "<point>[[226, 456], [290, 479], [382, 490], [536, 475]]</point>"}]}
{"id": 3016, "image": ["rtx_frames_success_16/49_bridge#episode_6653/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6660/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[0, 290], [98, 243], [252, 232], [397, 247], [573, 293]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the mushroom."}, {"from": "gpt", "value": "<point>[[553, 510], [486, 459], [368, 440], [284, 448], [200, 483]]</point>"}]}
{"id": 3018, "image": ["rtx_frames_success_16/49_bridge#episode_6681/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6681/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[785, 413], [652, 525], [559, 595], [481, 718]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the towel."}, {"from": "gpt", "value": "<point>[[582, 548], [666, 533], [727, 479], [782, 425]]</point>"}]}
{"id": 3020, "image": ["rtx_frames_success_16/49_bridge#episode_6689/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6691/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[704, 151], [579, 193], [472, 255], [431, 340], [400, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow cloth."}, {"from": "gpt", "value": "<point>[[269, 127], [359, 170], [414, 216], [455, 290], [481, 347]]</point>"}]}
{"id": 3022, "image": ["rtx_frames_success_16/49_bridge#episode_6707/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6708/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[266, 66], [365, 127], [507, 228], [669, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[539, 139], [429, 205], [327, 297], [264, 552]]</point>"}]}
{"id": 3024, "image": ["rtx_frames_success_16/49_bridge#episode_6708/frame_15.png", "rtx_frames_success_16/49_bridge#episode_6732/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon across the table."}, {"from": "gpt", "value": "<point>[[223, 506], [403, 560], [539, 602], [666, 707]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[347, 193], [405, 243], [469, 324], [513, 398], [550, 533], [562, 629]]</point>"}]}
{"id": 3026, "image": ["rtx_frames_success_16/49_bridge#episode_6740/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6744/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom in the pot."}, {"from": "gpt", "value": "<point>[[307, 208], [379, 220], [452, 247], [533, 344]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[536, 232], [504, 259], [492, 324], [513, 363]]</point>"}]}
{"id": 3028, "image": ["rtx_frames_success_16/49_bridge#episode_6762/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6769/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the corn."}, {"from": "gpt", "value": "<point>[[313, 378], [391, 425], [484, 432], [614, 417], [692, 363]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[556, 402], [481, 436], [397, 467], [287, 490]]</point>"}]}
{"id": 3030, "image": ["rtx_frames_success_16/49_bridge#episode_6770/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6785/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[301, 124], [290, 224], [278, 367], [281, 467], [304, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to release the spatula and pipe."}, {"from": "gpt", "value": "<point>[[440, 429], [542, 432], [594, 479], [640, 602]]</point>"}]}
{"id": 3032, "image": ["rtx_frames_success_16/49_bridge#episode_6785/frame_15.png", "rtx_frames_success_16/49_bridge#episode_6808/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pepper in the pot."}, {"from": "gpt", "value": "<point>[[657, 521], [498, 444], [397, 448], [321, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the can."}, {"from": "gpt", "value": "<point>[[599, 475], [466, 521], [391, 537], [284, 517]]</point>"}]}
{"id": 3034, "image": ["rtx_frames_success_16/49_bridge#episode_6808/frame_15.png", "rtx_frames_success_16/49_bridge#episode_6811/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the can on the table."}, {"from": "gpt", "value": "<point>[[336, 471], [449, 452], [542, 432], [663, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange toy animal."}, {"from": "gpt", "value": "<point>[[301, 50], [304, 116], [304, 185], [321, 255]]</point>"}]}
{"id": 3036, "image": ["rtx_frames_success_16/49_bridge#episode_6811/frame_15.png", "rtx_frames_success_16/49_bridge#episode_6815/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the orange toy animal towards the right side of the purple cloth."}, {"from": "gpt", "value": "<point>[[313, 166], [414, 247], [536, 390], [649, 718]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[67, 58], [208, 69], [310, 131], [368, 197], [440, 328]]</point>"}]}
{"id": 3038, "image": ["rtx_frames_success_16/49_bridge#episode_6819/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6833/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the can."}, {"from": "gpt", "value": "<point>[[762, 699], [599, 660], [513, 533], [434, 386], [440, 301]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the bowl."}, {"from": "gpt", "value": "<point>[[324, 147], [394, 205], [423, 290], [449, 429], [475, 680]]</point>"}]}
{"id": 3040, "image": ["rtx_frames_success_16/49_bridge#episode_6850/frame_15.png", "rtx_frames_success_16/49_bridge#episode_6865/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mushroom towards the towel."}, {"from": "gpt", "value": "<point>[[492, 139], [365, 189], [243, 263], [177, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[597, 556], [481, 514], [376, 471], [243, 417], [127, 417]]</point>"}]}
{"id": 3042, "image": ["rtx_frames_success_16/49_bridge#episode_6865/frame_15.png", "rtx_frames_success_16/49_bridge#episode_6866/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the towel."}, {"from": "gpt", "value": "<point>[[93, 448], [200, 402], [362, 359], [475, 359], [597, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[489, 154], [437, 263], [385, 344], [356, 529], [356, 571]]</point>"}]}
{"id": 3044, "image": ["rtx_frames_success_16/49_bridge#episode_6882/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6886/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red object."}, {"from": "gpt", "value": "<point>[[49, 193], [87, 239], [133, 324], [177, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[597, 143], [559, 297], [533, 425], [475, 649]]</point>"}]}
{"id": 3046, "image": ["rtx_frames_success_16/49_bridge#episode_6894/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6899/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[576, 533], [478, 486], [359, 486], [290, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[738, 301], [625, 290], [495, 328], [434, 378], [359, 494]]</point>"}]}
{"id": 3048, "image": ["rtx_frames_success_16/49_bridge#episode_6908/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6910/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[463, 12], [510, 81], [576, 181], [628, 282], [669, 417], [695, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[489, 340], [611, 336], [695, 371], [782, 498]]</point>"}]}
{"id": 3050, "image": ["rtx_frames_success_16/49_bridge#episode_6916/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6919/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[258, 355], [324, 429], [405, 521], [469, 653], [521, 849]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the vessel."}, {"from": "gpt", "value": "<point>[[258, 263], [379, 320], [486, 402], [559, 510], [591, 629]]</point>"}]}
{"id": 3052, "image": ["rtx_frames_success_16/49_bridge#episode_6928/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6930/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave handle."}, {"from": "gpt", "value": "<point>[[443, 301], [481, 347], [579, 355], [738, 305]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[617, 301], [527, 297], [414, 301], [301, 340], [206, 440]]</point>"}]}
{"id": 3054, "image": ["rtx_frames_success_16/49_bridge#episode_6938/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6945/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[585, 205], [527, 259], [489, 328], [463, 394], [440, 467], [434, 544], [429, 649]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the apple."}, {"from": "gpt", "value": "<point>[[67, 120], [127, 162], [174, 247], [214, 402]]</point>"}]}
{"id": 3056, "image": ["rtx_frames_success_16/49_bridge#episode_6966/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6967/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[235, 154], [295, 162], [394, 193], [492, 243], [582, 309], [614, 363]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull open the oven door."}, {"from": "gpt", "value": "<point>[[510, 606], [527, 703], [568, 780], [637, 884]]</point>"}]}
{"id": 3058, "image": ["rtx_frames_success_16/49_bridge#episode_6971/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6976/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[371, 139], [423, 185], [449, 220], [484, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pear."}, {"from": "gpt", "value": "<point>[[501, 290], [553, 301], [602, 336], [631, 367]]</point>"}]}
{"id": 3060, "image": ["rtx_frames_success_16/49_bridge#episode_6984/frame_0.png", "rtx_frames_success_16/49_bridge#episode_6985/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[197, 386], [353, 363], [469, 351], [597, 378], [753, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the spoon on the left the pot."}, {"from": "gpt", "value": "<point>[[226, 402], [194, 440], [148, 421], [136, 378]]</point>"}]}
{"id": 3062, "image": ["rtx_frames_success_16/49_bridge#episode_6985/frame_15.png", "rtx_frames_success_16/49_bridge#episode_6995/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the spoon between the napkin and the stove."}, {"from": "gpt", "value": "<point>[[133, 320], [237, 324], [394, 390], [510, 463], [611, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[481, 93], [437, 162], [382, 228], [339, 297], [301, 417]]</point>"}]}
{"id": 3064, "image": ["rtx_frames_success_16/49_bridge#episode_7004/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7023/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[235, 309], [214, 386], [217, 448], [240, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[391, 313], [356, 367], [330, 394], [278, 459]]</point>"}]}
{"id": 3066, "image": ["rtx_frames_success_16/49_bridge#episode_7027/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7028/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[533, 228], [510, 278], [466, 375], [460, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[837, 255], [793, 286], [736, 324], [660, 336]]</point>"}]}
{"id": 3068, "image": ["rtx_frames_success_16/49_bridge#episode_7028/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7028/frame_27.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pot on the stove."}, {"from": "gpt", "value": "<point>[[374, 228], [359, 317], [374, 413], [417, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to put the potato in the pot."}, {"from": "gpt", "value": "<point>[[458, 270], [484, 320], [518, 328], [542, 309]]</point>"}]}
{"id": 3070, "image": ["rtx_frames_success_16/49_bridge#episode_7052/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7052/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[403, 236], [486, 301], [576, 405], [654, 533], [675, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spoon down."}, {"from": "gpt", "value": "<point>[[709, 575], [625, 591], [492, 579], [359, 521], [290, 390]]</point>"}]}
{"id": 3072, "image": ["rtx_frames_success_16/49_bridge#episode_7064/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7069/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[515, 162], [481, 224], [440, 324], [437, 398], [443, 498]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the jar of almonds over the pot."}, {"from": "gpt", "value": "<point>[[400, 452], [455, 425], [515, 432], [568, 452], [599, 506]]</point>"}]}
{"id": 3074, "image": ["rtx_frames_success_16/49_bridge#episode_7080/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7082/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[194, 205], [261, 247], [356, 328], [420, 417], [515, 668]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the oven door handle."}, {"from": "gpt", "value": "<point>[[174, 444], [301, 471], [466, 448], [579, 367], [657, 239]]</point>"}]}
{"id": 3076, "image": ["rtx_frames_success_16/49_bridge#episode_7082/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7098/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull open the oven door."}, {"from": "gpt", "value": "<point>[[510, 320], [495, 386], [429, 444], [336, 494], [217, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pepper."}, {"from": "gpt", "value": "<point>[[307, 27], [353, 73], [388, 127], [405, 216]]</point>"}]}
{"id": 3078, "image": ["rtx_frames_success_16/49_bridge#episode_7098/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7104/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the pepper in the pot."}, {"from": "gpt", "value": "<point>[[339, 77], [292, 166], [266, 247], [264, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drumstick."}, {"from": "gpt", "value": "<point>[[278, 93], [255, 154], [232, 243], [235, 359], [240, 471]]</point>"}]}
{"id": 3080, "image": ["rtx_frames_success_16/49_bridge#episode_7106/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7140/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[408, 147], [475, 224], [550, 328], [617, 425], [669, 560], [678, 710]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[443, 278], [379, 340], [342, 378], [281, 367]]</point>"}]}
{"id": 3082, "image": ["rtx_frames_success_16/49_bridge#episode_7152/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7152/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[782, 394], [753, 421], [704, 448], [649, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the corn in the bowl."}, {"from": "gpt", "value": "<point>[[652, 409], [585, 394], [544, 405], [515, 448]]</point>"}]}
{"id": 3084, "image": ["rtx_frames_success_16/49_bridge#episode_7169/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7174/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn in the pan."}, {"from": "gpt", "value": "<point>[[168, 239], [266, 305], [359, 367], [423, 456], [530, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[513, 239], [440, 266], [362, 305], [319, 398]]</point>"}]}
{"id": 3086, "image": ["rtx_frames_success_16/49_bridge#episode_7180/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7203/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the basket of clothes."}, {"from": "gpt", "value": "<point>[[507, 429], [547, 521], [582, 598], [608, 745]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the chicken leg."}, {"from": "gpt", "value": "<point>[[501, 429], [617, 479], [715, 525], [805, 741]]</point>"}]}
{"id": 3088, "image": ["rtx_frames_success_16/49_bridge#episode_7215/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7224/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[246, 208], [350, 232], [420, 270], [495, 328], [588, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[371, 243], [478, 278], [562, 313], [681, 398]]</point>"}]}
{"id": 3090, "image": ["rtx_frames_success_16/49_bridge#episode_7245/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7246/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[162, 309], [362, 347], [472, 440], [582, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[701, 440], [614, 471], [513, 479], [423, 483], [307, 463], [208, 432]]</point>"}]}
{"id": 3092, "image": ["rtx_frames_success_16/49_bridge#episode_7248/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7248/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[408, 502], [507, 475], [614, 440], [764, 375]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull down the microwave handle."}, {"from": "gpt", "value": "<point>[[753, 332], [701, 402], [625, 463], [565, 486], [443, 506]]</point>"}]}
{"id": 3094, "image": ["rtx_frames_success_16/49_bridge#episode_7273/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7273/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[313, 85], [261, 170], [223, 259], [197, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the desired location."}, {"from": "gpt", "value": "<point>[[136, 371], [284, 398], [434, 417], [608, 394]]</point>"}]}
{"id": 3096, "image": ["rtx_frames_success_16/49_bridge#episode_7284/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7321/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow towel."}, {"from": "gpt", "value": "<point>[[426, 328], [542, 344], [628, 386], [663, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the mushroom."}, {"from": "gpt", "value": "<point>[[542, 394], [597, 452], [643, 529], [666, 672]]</point>"}]}
{"id": 3098, "image": ["rtx_frames_success_16/49_bridge#episode_7321/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7344/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move down towards the microwave."}, {"from": "gpt", "value": "<point>[[408, 270], [521, 332], [605, 386], [733, 490]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the potato."}, {"from": "gpt", "value": "<point>[[368, 483], [295, 452], [180, 425], [96, 456]]</point>"}]}
{"id": 3100, "image": ["rtx_frames_success_16/49_bridge#episode_7352/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7352/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[310, 236], [295, 297], [295, 386], [327, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the cutting board."}, {"from": "gpt", "value": "<point>[[269, 405], [330, 467], [385, 514], [501, 660]]</point>"}]}
{"id": 3102, "image": ["rtx_frames_success_16/49_bridge#episode_7355/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7356/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave door."}, {"from": "gpt", "value": "<point>[[43, 266], [168, 278], [292, 301], [423, 359], [544, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[469, 15], [446, 66], [434, 143], [423, 259], [434, 452]]</point>"}]}
{"id": 3104, "image": ["rtx_frames_success_16/49_bridge#episode_7369/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7369/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cucumber."}, {"from": "gpt", "value": "<point>[[681, 378], [585, 371], [475, 386], [382, 394], [223, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cucumber."}, {"from": "gpt", "value": "<point>[[206, 351], [307, 336], [368, 344], [501, 398]]</point>"}]}
{"id": 3106, "image": ["rtx_frames_success_16/49_bridge#episode_7370/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7370/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[405, 108], [350, 178], [301, 266], [290, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the right of the pot."}, {"from": "gpt", "value": "<point>[[675, 494], [660, 579], [663, 656], [689, 710]]</point>"}]}
{"id": 3108, "image": ["rtx_frames_success_16/49_bridge#episode_7376/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7382/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow pepper."}, {"from": "gpt", "value": "<point>[[492, 143], [533, 162], [620, 224], [698, 328], [744, 440], [767, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom in the pot."}, {"from": "gpt", "value": "<point>[[646, 212], [625, 266], [579, 355], [515, 517]]</point>"}]}
{"id": 3110, "image": ["rtx_frames_success_16/49_bridge#episode_7383/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7401/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the purple cloth."}, {"from": "gpt", "value": "<point>[[374, 228], [475, 243], [550, 293], [605, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn in the pot."}, {"from": "gpt", "value": "<point>[[405, 286], [405, 371], [420, 444], [452, 556]]</point>"}]}
{"id": 3112, "image": ["rtx_frames_success_16/49_bridge#episode_7401/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7414/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the corn towards the middle right of the table."}, {"from": "gpt", "value": "<point>[[924, 166], [869, 205], [814, 278], [773, 367], [759, 444], [759, 544], [759, 614]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[411, 189], [513, 208], [625, 236], [724, 286], [808, 371]]</point>"}]}
{"id": 3114, "image": ["rtx_frames_success_16/49_bridge#episode_7428/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7434/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[423, 170], [385, 208], [353, 263], [321, 320], [292, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[339, 69], [287, 162], [269, 270], [275, 425]]</point>"}]}
{"id": 3116, "image": ["rtx_frames_success_16/49_bridge#episode_7434/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7446/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the mushroom in the pot."}, {"from": "gpt", "value": "<point>[[266, 394], [333, 375], [417, 394], [504, 436], [623, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[301, 116], [362, 170], [405, 263], [446, 347], [469, 421], [489, 533], [486, 660]]</point>"}]}
{"id": 3118, "image": ["rtx_frames_success_16/49_bridge#episode_7446/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7464/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the cloth."}, {"from": "gpt", "value": "<point>[[394, 645], [429, 533], [452, 440], [463, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cup."}, {"from": "gpt", "value": "<point>[[759, 440], [698, 421], [637, 367], [582, 247]]</point>"}]}
{"id": 3120, "image": ["rtx_frames_success_16/49_bridge#episode_7464/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7469/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cup towards the plate."}, {"from": "gpt", "value": "<point>[[602, 243], [518, 251], [408, 263], [339, 305]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blueberries."}, {"from": "gpt", "value": "<point>[[617, 305], [553, 293], [449, 320], [423, 452]]</point>"}]}
{"id": 3122, "image": ["rtx_frames_success_16/49_bridge#episode_7494/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7506/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the sushi."}, {"from": "gpt", "value": "<point>[[521, 243], [431, 247], [330, 263], [252, 324], [156, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the metal pot."}, {"from": "gpt", "value": "<point>[[423, 243], [400, 336], [359, 386], [278, 432]]</point>"}]}
{"id": 3124, "image": ["rtx_frames_success_16/49_bridge#episode_7506/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7509/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot to the right side of the table."}, {"from": "gpt", "value": "<point>[[738, 382], [724, 440], [718, 486], [733, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[347, 147], [400, 201], [429, 270], [469, 351], [486, 425], [489, 529]]</point>"}]}
{"id": 3126, "image": ["rtx_frames_success_16/49_bridge#episode_7509/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7527/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the bowl on the towel."}, {"from": "gpt", "value": "<point>[[521, 456], [446, 425], [388, 378], [359, 274]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave door."}, {"from": "gpt", "value": "<point>[[730, 672], [675, 672], [582, 649], [486, 571], [385, 467], [290, 324], [191, 135]]</point>"}]}
{"id": 3128, "image": ["rtx_frames_success_16/49_bridge#episode_7527/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7529/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the door closed."}, {"from": "gpt", "value": "<point>[[185, 382], [272, 413], [484, 440], [707, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[510, 266], [544, 313], [579, 390], [611, 606]]</point>"}]}
{"id": 3130, "image": ["rtx_frames_success_16/49_bridge#episode_7543/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7554/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[203, 286], [290, 317], [356, 359], [397, 425], [446, 544]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[460, 162], [429, 243], [408, 293], [400, 371]]</point>"}]}
{"id": 3132, "image": ["rtx_frames_success_16/49_bridge#episode_7554/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7568/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bowl above the cloth."}, {"from": "gpt", "value": "<point>[[553, 266], [620, 305], [689, 371], [727, 479], [744, 544]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[295, 320], [417, 320], [504, 332], [582, 405]]</point>"}]}
{"id": 3134, "image": ["rtx_frames_success_16/49_bridge#episode_7580/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7589/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[171, 475], [307, 471], [400, 517], [440, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[156, 46], [252, 85], [324, 166], [400, 255], [452, 332], [504, 425], [559, 521], [585, 583], [599, 699]]</point>"}]}
{"id": 3136, "image": ["rtx_frames_success_16/49_bridge#episode_7589/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7596/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the lid."}, {"from": "gpt", "value": "<point>[[579, 676], [452, 625], [411, 517], [455, 359]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[142, 263], [243, 266], [374, 344], [446, 398], [521, 475], [559, 564]]</point>"}]}
{"id": 3138, "image": ["rtx_frames_success_16/49_bridge#episode_7598/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7599/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the salmon."}, {"from": "gpt", "value": "<point>[[240, 228], [269, 290], [307, 413], [304, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[550, 216], [550, 293], [582, 371], [646, 490]]</point>"}]}
{"id": 3140, "image": ["rtx_frames_success_16/49_bridge#episode_7599/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7609/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the pot."}, {"from": "gpt", "value": "<point>[[542, 185], [452, 247], [362, 293], [249, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the purple towel."}, {"from": "gpt", "value": "<point>[[443, 46], [429, 127], [423, 212], [440, 328], [486, 490]]</point>"}]}
{"id": 3142, "image": ["rtx_frames_success_16/49_bridge#episode_7620/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7644/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the broccoli in the pan."}, {"from": "gpt", "value": "<point>[[489, 290], [420, 359], [347, 409], [208, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[510, 351], [388, 394], [310, 413], [269, 371]]</point>"}]}
{"id": 3144, "image": ["rtx_frames_success_16/49_bridge#episode_7671/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7688/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[295, 266], [408, 278], [504, 320], [576, 386], [634, 486]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[278, 224], [249, 266], [235, 320], [232, 355]]</point>"}]}
{"id": 3146, "image": ["rtx_frames_success_16/49_bridge#episode_7688/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7695/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi roll."}, {"from": "gpt", "value": "<point>[[211, 340], [301, 328], [400, 320], [486, 328], [623, 375]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[388, 162], [374, 278], [388, 355], [400, 429], [423, 587]]</point>"}]}
{"id": 3148, "image": ["rtx_frames_success_16/49_bridge#episode_7719/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7728/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the chocolate bar."}, {"from": "gpt", "value": "<point>[[539, 378], [504, 471], [466, 506], [394, 537], [324, 544]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[657, 228], [637, 328], [611, 425], [602, 548], [608, 660]]</point>"}]}
{"id": 3150, "image": ["rtx_frames_success_16/49_bridge#episode_7737/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7745/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[211, 243], [304, 270], [405, 301], [466, 320], [533, 378], [591, 432], [669, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[304, 232], [403, 297], [481, 363], [553, 432], [605, 583]]</point>"}]}
{"id": 3152, "image": ["rtx_frames_success_16/49_bridge#episode_7750/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7759/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[570, 514], [498, 510], [434, 444], [405, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the cauliflower."}, {"from": "gpt", "value": "<point>[[536, 147], [527, 212], [527, 282], [568, 440]]</point>"}]}
{"id": 3154, "image": ["rtx_frames_success_16/49_bridge#episode_7759/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7766/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot with cauliflower onto the stove."}, {"from": "gpt", "value": "<point>[[611, 46], [643, 112], [678, 178], [704, 243], [686, 293], [565, 286], [492, 286]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the food item."}, {"from": "gpt", "value": "<point>[[374, 313], [365, 378], [333, 405], [275, 425], [217, 432], [90, 429]]</point>"}]}
{"id": 3156, "image": ["rtx_frames_success_16/49_bridge#episode_7766/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7773/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the food item."}, {"from": "gpt", "value": "<point>[[168, 363], [292, 317], [397, 297], [466, 293], [524, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to open the gripper above the broccoli in the bowl."}, {"from": "gpt", "value": "<point>[[408, 205], [397, 263], [391, 301], [411, 340]]</point>"}]}
{"id": 3158, "image": ["rtx_frames_success_16/49_bridge#episode_7776/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7776/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position above the pot."}, {"from": "gpt", "value": "<point>[[515, 259], [492, 290], [458, 351], [411, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pot on the right corner of the table."}, {"from": "gpt", "value": "<point>[[452, 266], [472, 363], [504, 459], [536, 533], [591, 672]]</point>"}]}
{"id": 3160, "image": ["rtx_frames_success_16/49_bridge#episode_7778/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7789/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the refrigerator door closed."}, {"from": "gpt", "value": "<point>[[316, 529], [417, 525], [492, 514], [594, 467]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[411, 309], [405, 386], [429, 456], [484, 564]]</point>"}]}
{"id": 3162, "image": ["rtx_frames_success_16/49_bridge#episode_7810/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7812/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[458, 413], [440, 471], [403, 494], [356, 510], [301, 510], [272, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[229, 328], [403, 336], [536, 351], [631, 382], [707, 429]]</point>"}]}
{"id": 3164, "image": ["rtx_frames_success_16/49_bridge#episode_7812/frame_15.png", "rtx_frames_success_16/49_bridge#episode_7814/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the sushi towards the cherry can."}, {"from": "gpt", "value": "<point>[[307, 270], [272, 324], [243, 355], [203, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[715, 375], [643, 425], [530, 432], [365, 429], [261, 405]]</point>"}]}
{"id": 3166, "image": ["rtx_frames_success_16/49_bridge#episode_7819/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7821/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[391, 216], [403, 293], [429, 351], [498, 405], [623, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the rag."}, {"from": "gpt", "value": "<point>[[269, 124], [252, 220], [255, 301], [266, 398], [290, 490]]</point>"}]}
{"id": 3168, "image": ["rtx_frames_success_16/49_bridge#episode_7828/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7833/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow cloth."}, {"from": "gpt", "value": "<point>[[194, 317], [191, 375], [191, 417], [203, 483]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[701, 178], [628, 224], [475, 278], [342, 270]]</point>"}]}
{"id": 3170, "image": ["rtx_frames_success_16/49_bridge#episode_7841/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7841/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the sushi from the red cloth."}, {"from": "gpt", "value": "<point>[[579, 282], [542, 328], [515, 363], [489, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the sushi to the left of the red cloth."}, {"from": "gpt", "value": "<point>[[492, 332], [452, 382], [403, 378], [359, 336]]</point>"}]}
{"id": 3172, "image": ["rtx_frames_success_16/49_bridge#episode_7843/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7845/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the vessel."}, {"from": "gpt", "value": "<point>[[715, 178], [715, 236], [701, 297], [678, 355], [646, 405], [608, 456], [573, 510], [527, 556], [504, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[553, 328], [565, 421], [547, 494], [510, 544]]</point>"}]}
{"id": 3174, "image": ["rtx_frames_success_16/49_bridge#episode_7858/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7871/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[510, 247], [530, 402], [530, 541], [530, 633], [521, 749]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[336, 181], [313, 278], [275, 371], [226, 479], [197, 517]]</point>"}]}
{"id": 3176, "image": ["rtx_frames_success_16/49_bridge#episode_7893/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7895/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[359, 112], [339, 166], [313, 247], [295, 313], [298, 456]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[536, 193], [536, 278], [536, 347], [527, 440], [527, 506], [510, 579], [475, 629]]</point>"}]}
{"id": 3178, "image": ["rtx_frames_success_16/49_bridge#episode_7900/frame_0.png", "rtx_frames_success_16/49_bridge#episode_7909/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[258, 100], [287, 131], [327, 154], [353, 185], [394, 228], [423, 278], [472, 340], [495, 378], [513, 432], [518, 483], [524, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[395, 308], [455, 371], [568, 411], [701, 426]]</point>"}]}
{"id": 3180, "image": ["rtx_frames_success_17/49_bridge#episode_86/frame_0.png", "rtx_frames_success_17/49_bridge#episode_812/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[466, 19], [422, 115], [304, 271], [281, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the small metal vessel."}, {"from": "gpt", "value": "<point>[[141, 169], [228, 224], [341, 273], [395, 362]]</point>"}]}
{"id": 3182, "image": ["rtx_frames_success_17/49_bridge#episode_827/frame_0.png", "rtx_frames_success_17/49_bridge#episode_827/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the washing machine."}, {"from": "gpt", "value": "<point>[[245, 565], [311, 468], [386, 372], [499, 280]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the clothing out of the machine."}, {"from": "gpt", "value": "<point>[[443, 247], [353, 360], [298, 414], [258, 682]]</point>"}]}
{"id": 3184, "image": ["rtx_frames_success_17/49_bridge#episode_853/frame_0.png", "rtx_frames_success_17/49_bridge#episode_853/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[581, 9], [454, 146], [286, 226], [169, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the knife towards the towel."}, {"from": "gpt", "value": "<point>[[187, 339], [360, 456], [572, 534], [581, 654]]</point>"}]}
{"id": 3186, "image": ["rtx_frames_success_17/49_bridge#episode_854/frame_0.png", "rtx_frames_success_17/49_bridge#episode_854/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[429, 186], [521, 271], [574, 339], [591, 438]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the spoon on the left of the towel."}, {"from": "gpt", "value": "<point>[[625, 233], [533, 299], [334, 376], [259, 464]]</point>"}]}
{"id": 3188, "image": ["rtx_frames_success_17/49_bridge#episode_856/frame_0.png", "rtx_frames_success_17/49_bridge#episode_859/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[598, 106], [558, 202], [510, 372], [512, 574]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the green towel."}, {"from": "gpt", "value": "<point>[[487, 494], [386, 459], [236, 421], [187, 478]]</point>"}]}
{"id": 3190, "image": ["rtx_frames_success_17/49_bridge#episode_862/frame_0.png", "rtx_frames_success_17/49_bridge#episode_878/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[446, 73], [466, 202], [494, 402], [466, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[388, 198], [478, 268], [596, 332], [598, 421]]</point>"}]}
{"id": 3192, "image": ["rtx_frames_success_17/49_bridge#episode_880/frame_0.png", "rtx_frames_success_17/49_bridge#episode_880/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[249, 155], [245, 224], [226, 296], [282, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cloth to the back of the table."}, {"from": "gpt", "value": "<point>[[208, 294], [411, 435], [568, 567], [561, 682]]</point>"}]}
{"id": 3194, "image": ["rtx_frames_success_17/49_bridge#episode_893/frame_0.png", "rtx_frames_success_17/49_bridge#episode_911/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[256, 2], [302, 64], [392, 176], [409, 360]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the knife."}, {"from": "gpt", "value": "<point>[[401, 31], [552, 207], [598, 318], [630, 673]]</point>"}]}
{"id": 3196, "image": ["rtx_frames_success_17/49_bridge#episode_911/frame_15.png", "rtx_frames_success_17/49_bridge#episode_913/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife towards the sushi."}, {"from": "gpt", "value": "<point>[[570, 449], [431, 327], [351, 254], [314, 304]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[311, 89], [348, 169], [418, 327], [381, 553]]</point>"}]}
{"id": 3198, "image": ["rtx_frames_success_17/49_bridge#episode_914/frame_0.png", "rtx_frames_success_17/49_bridge#episode_917/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cloth."}, {"from": "gpt", "value": "<point>[[471, 254], [581, 395], [626, 513], [611, 668]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the mushroom."}, {"from": "gpt", "value": "<point>[[141, 247], [226, 233], [279, 228], [351, 266]]</point>"}]}
{"id": 3200, "image": ["rtx_frames_success_17/49_bridge#episode_917/frame_15.png", "rtx_frames_success_17/49_bridge#episode_924/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mushroom over the pot."}, {"from": "gpt", "value": "<point>[[295, 148], [259, 214], [192, 329], [203, 461]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bowl."}, {"from": "gpt", "value": "<point>[[249, 12], [376, 146], [519, 391], [514, 628]]</point>"}]}
{"id": 3202, "image": ["rtx_frames_success_17/49_bridge#episode_7950/frame_0.png", "rtx_frames_success_17/49_bridge#episode_7953/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[382, 343], [537, 388], [632, 543]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[424, 225], [319, 395], [346, 495]]</point>"}]}
{"id": 3204, "image": ["rtx_frames_success_17/49_bridge#episode_7963/frame_0.png", "rtx_frames_success_17/49_bridge#episode_7964/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the vessel."}, {"from": "gpt", "value": "<point>[[740, 691], [427, 288], [316, 266]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the refrigerator handle."}, {"from": "gpt", "value": "<point>[[199, 646], [418, 362], [565, 392]]</point>"}]}
{"id": 3206, "image": ["rtx_frames_success_17/49_bridge#episode_7966/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8009/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[717, 395], [313, 336], [233, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the spoon."}, {"from": "gpt", "value": "<point>[[512, 358], [374, 355], [360, 307]]</point>"}]}
{"id": 3208, "image": ["rtx_frames_success_17/49_bridge#episode_8009/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8017/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move right."}, {"from": "gpt", "value": "<point>[[568, 251], [693, 325], [759, 447]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[648, 362], [679, 576], [590, 705]]</point>"}]}
{"id": 3210, "image": ["rtx_frames_success_17/49_bridge#episode_8029/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8040/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the corn."}, {"from": "gpt", "value": "<point>[[450, 155], [528, 311], [593, 409], [692, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[406, 127], [360, 266], [251, 386], [162, 454]]</point>"}]}
{"id": 3212, "image": ["rtx_frames_success_17/49_bridge#episode_8045/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8051/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry in the bowl."}, {"from": "gpt", "value": "<point>[[499, 122], [579, 669], [715, 761]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[596, 321], [449, 432], [357, 499]]</point>"}]}
{"id": 3214, "image": ["rtx_frames_success_17/49_bridge#episode_8065/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8073/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[759, 329], [655, 440], [533, 496], [480, 544]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[482, 216], [501, 336], [526, 461], [528, 671]]</point>"}]}
{"id": 3216, "image": ["rtx_frames_success_17/49_bridge#episode_8085/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8103/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[339, 282], [318, 332], [288, 367], [251, 412]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[145, 47], [155, 224], [169, 360], [231, 480]]</point>"}]}
{"id": 3218, "image": ["rtx_frames_success_17/49_bridge#episode_8103/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8106/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[415, 381], [476, 421], [554, 471], [658, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[861, 315], [678, 325], [533, 334], [344, 334]]</point>"}]}
{"id": 3220, "image": ["rtx_frames_success_17/49_bridge#episode_8113/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8115/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the can."}, {"from": "gpt", "value": "<point>[[35, 242], [194, 304], [402, 358], [577, 388]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[365, 104], [376, 212], [409, 351], [386, 471]]</point>"}]}
{"id": 3222, "image": ["rtx_frames_success_17/49_bridge#episode_8143/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8150/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the small metal pan."}, {"from": "gpt", "value": "<point>[[381, 89], [436, 224], [484, 348], [454, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[305, 85], [431, 296], [637, 464], [819, 600]]</point>"}]}
{"id": 3224, "image": ["rtx_frames_success_17/49_bridge#episode_8155/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8157/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[231, 139], [323, 247], [424, 376], [459, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[729, 120], [665, 245], [581, 287], [475, 346]]</point>"}]}
{"id": 3226, "image": ["rtx_frames_success_17/49_bridge#episode_8167/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8175/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[176, 127], [335, 219], [427, 315], [473, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the can."}, {"from": "gpt", "value": "<point>[[78, 205], [176, 247], [224, 268], [236, 315]]</point>"}]}
{"id": 3228, "image": ["rtx_frames_success_17/49_bridge#episode_8175/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8185/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to slide the can towards the pan."}, {"from": "gpt", "value": "<point>[[228, 205], [256, 299], [318, 438], [355, 562]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot on the shelf."}, {"from": "gpt", "value": "<point>[[399, 132], [379, 282], [378, 398], [402, 515]]</point>"}]}
{"id": 3230, "image": ["rtx_frames_success_17/49_bridge#episode_8193/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8198/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cucumber."}, {"from": "gpt", "value": "<point>[[858, 327], [796, 374], [702, 398], [614, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bowl from the stove top."}, {"from": "gpt", "value": "<point>[[416, 195], [512, 271], [581, 320], [648, 388]]</point>"}]}
{"id": 3232, "image": ["rtx_frames_success_17/49_bridge#episode_8199/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8199/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green vegetable."}, {"from": "gpt", "value": "<point>[[491, 16], [466, 120], [443, 252], [441, 419]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the orange vegetable."}, {"from": "gpt", "value": "<point>[[469, 353], [358, 315], [284, 311], [258, 391]]</point>"}]}
{"id": 3234, "image": ["rtx_frames_success_17/49_bridge#episode_8204/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8210/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[454, 111], [522, 200], [600, 275], [595, 325]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange cloth."}, {"from": "gpt", "value": "<point>[[259, 75], [388, 172], [506, 252], [551, 433]]</point>"}]}
{"id": 3236, "image": ["rtx_frames_success_17/49_bridge#episode_8230/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8230/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pot."}, {"from": "gpt", "value": "<point>[[379, 240], [316, 294], [272, 344], [256, 447]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the pot next to the carrot."}, {"from": "gpt", "value": "<point>[[353, 311], [443, 353], [512, 391], [586, 478]]</point>"}]}
{"id": 3238, "image": ["rtx_frames_success_17/49_bridge#episode_8234/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8235/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[632, 264], [468, 344], [304, 395], [275, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position over the red ball inside the pot."}, {"from": "gpt", "value": "<point>[[572, 202], [519, 228], [466, 266], [443, 372]]</point>"}]}
{"id": 3240, "image": ["rtx_frames_success_17/49_bridge#episode_8235/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8241/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the red ball to the left."}, {"from": "gpt", "value": "<point>[[256, 238], [215, 299], [185, 348], [191, 438]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[168, 233], [291, 301], [369, 334], [586, 362]]</point>"}]}
{"id": 3242, "image": ["rtx_frames_success_17/49_bridge#episode_8255/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8255/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the potato."}, {"from": "gpt", "value": "<point>[[549, 376], [491, 384], [404, 398], [395, 308]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the potato towards the blue cloth."}, {"from": "gpt", "value": "<point>[[427, 247], [496, 362], [556, 440], [561, 626]]</point>"}]}
{"id": 3244, "image": ["rtx_frames_success_17/49_bridge#episode_8259/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8267/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[570, 416], [439, 405], [277, 393], [235, 322]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[275, 247], [351, 306], [454, 311], [551, 235]]</point>"}]}
{"id": 3246, "image": ["rtx_frames_success_17/49_bridge#episode_8301/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8310/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lemon cob in the pot."}, {"from": "gpt", "value": "<point>[[180, 254], [309, 336], [454, 381], [664, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the bowl."}, {"from": "gpt", "value": "<point>[[422, 216], [446, 311], [484, 405], [471, 609]]</point>"}]}
{"id": 3248, "image": ["rtx_frames_success_17/49_bridge#episode_8316/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8330/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the spoon."}, {"from": "gpt", "value": "<point>[[468, 33], [408, 162], [325, 334], [328, 640]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[572, 327], [457, 388], [335, 435], [295, 548]]</point>"}]}
{"id": 3250, "image": ["rtx_frames_success_17/49_bridge#episode_8332/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8352/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot in the bowl."}, {"from": "gpt", "value": "<point>[[341, 134], [309, 176], [256, 242], [247, 372]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the item."}, {"from": "gpt", "value": "<point>[[394, 136], [348, 212], [288, 294], [274, 442]]</point>"}]}
{"id": 3252, "image": ["rtx_frames_success_17/49_bridge#episode_8363/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8363/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the pot."}, {"from": "gpt", "value": "<point>[[155, 327], [235, 329], [346, 329], [420, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the pot to the right of the blue towel."}, {"from": "gpt", "value": "<point>[[420, 264], [469, 355], [517, 499], [476, 659]]</point>"}]}
{"id": 3254, "image": ["rtx_frames_success_17/49_bridge#episode_8373/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8380/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[672, 259], [551, 332], [397, 416], [291, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[58, 247], [203, 325], [418, 449], [454, 567]]</point>"}]}
{"id": 3256, "image": ["rtx_frames_success_17/49_bridge#episode_8387/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8391/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container with the carrot."}, {"from": "gpt", "value": "<point>[[498, 259], [450, 339], [413, 471], [455, 692]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[686, 212], [540, 296], [311, 372], [157, 435]]</point>"}]}
{"id": 3258, "image": ["rtx_frames_success_17/49_bridge#episode_8401/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8409/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[540, 28], [476, 139], [392, 325], [441, 513]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl with corn."}, {"from": "gpt", "value": "<point>[[628, 240], [559, 275], [459, 327], [411, 419]]</point>"}]}
{"id": 3260, "image": ["rtx_frames_success_17/49_bridge#episode_8420/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8420/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[180, 327], [235, 275], [274, 226], [298, 259]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[293, 228], [386, 294], [469, 379], [450, 522]]</point>"}]}
{"id": 3262, "image": ["rtx_frames_success_17/49_bridge#episode_8436/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8445/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the mushroom in the bowl."}, {"from": "gpt", "value": "<point>[[637, 228], [577, 308], [503, 452], [510, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the scrubber."}, {"from": "gpt", "value": "<point>[[263, 73], [335, 129], [397, 205], [386, 332]]</point>"}]}
{"id": 3264, "image": ["rtx_frames_success_17/49_bridge#episode_8455/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8482/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[745, 346], [634, 369], [482, 405], [349, 355]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the red spoon."}, {"from": "gpt", "value": "<point>[[526, 459], [503, 518], [466, 598], [494, 661]]</point>"}]}
{"id": 3266, "image": ["rtx_frames_success_17/49_bridge#episode_8483/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8487/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal pot."}, {"from": "gpt", "value": "<point>[[284, 87], [378, 266], [487, 414], [485, 600]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[450, 2], [480, 129], [547, 379], [556, 612]]</point>"}]}
{"id": 3268, "image": ["rtx_frames_success_17/49_bridge#episode_8490/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8516/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[323, 33], [455, 184], [572, 405], [577, 619]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the sushi container."}, {"from": "gpt", "value": "<point>[[445, 118], [355, 174], [240, 240], [206, 412]]</point>"}]}
{"id": 3270, "image": ["rtx_frames_success_17/49_bridge#episode_8516/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8524/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the sushi on the middle of the table."}, {"from": "gpt", "value": "<point>[[206, 386], [334, 388], [429, 398], [436, 518]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the toy leopard."}, {"from": "gpt", "value": "<point>[[286, 212], [201, 268], [152, 304], [187, 431]]</point>"}]}
{"id": 3272, "image": ["rtx_frames_success_17/49_bridge#episode_8525/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8536/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[584, 5], [616, 141], [646, 372], [588, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[716, 414], [641, 402], [547, 405], [554, 452]]</point>"}]}
{"id": 3274, "image": ["rtx_frames_success_17/49_bridge#episode_8542/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8559/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[545, 52], [496, 108], [431, 176], [401, 318]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[196, 87], [281, 167], [371, 301], [371, 447]]</point>"}]}
{"id": 3276, "image": ["rtx_frames_success_17/49_bridge#episode_8561/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8581/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[326, 374], [429, 393], [540, 391], [635, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the blue spoon."}, {"from": "gpt", "value": "<point>[[491, 144], [422, 235], [302, 313], [203, 494]]</point>"}]}
{"id": 3278, "image": ["rtx_frames_success_17/49_bridge#episode_8581/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8583/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the silver pot."}, {"from": "gpt", "value": "<point>[[282, 275], [404, 332], [540, 372], [692, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the broccoli."}, {"from": "gpt", "value": "<point>[[397, 24], [365, 87], [321, 162], [305, 294]]</point>"}]}
{"id": 3280, "image": ["rtx_frames_success_17/49_bridge#episode_8583/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8630/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the broccoli on the orange towel."}, {"from": "gpt", "value": "<point>[[517, 414], [589, 511], [648, 612], [623, 687]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[360, 235], [487, 369], [618, 499], [669, 715]]</point>"}]}
{"id": 3282, "image": ["rtx_frames_success_17/49_bridge#episode_8644/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8652/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[379, 306], [404, 358], [413, 433], [402, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the beet."}, {"from": "gpt", "value": "<point>[[229, 191], [328, 268], [411, 332], [448, 393]]</point>"}]}
{"id": 3284, "image": ["rtx_frames_success_17/49_bridge#episode_8668/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8682/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[233, 146], [295, 235], [397, 351], [432, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the metal bowl."}, {"from": "gpt", "value": "<point>[[535, 141], [461, 174], [330, 240], [341, 384]]</point>"}]}
{"id": 3286, "image": ["rtx_frames_success_17/49_bridge#episode_8686/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8688/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[588, 205], [448, 231], [236, 311], [208, 400]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[466, 191], [408, 318], [342, 452], [378, 546]]</point>"}]}
{"id": 3288, "image": ["rtx_frames_success_17/49_bridge#episode_8688/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8700/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pan towards the bottle."}, {"from": "gpt", "value": "<point>[[529, 558], [614, 624], [656, 694], [660, 772]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the blue spoon."}, {"from": "gpt", "value": "<point>[[514, 12], [399, 127], [265, 249], [169, 456]]</point>"}]}
{"id": 3290, "image": ["rtx_frames_success_17/49_bridge#episode_8713/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8716/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the metal pot."}, {"from": "gpt", "value": "<point>[[171, 259], [304, 294], [478, 311], [596, 365]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[591, 129], [485, 195], [335, 273], [217, 449]]</point>"}]}
{"id": 3292, "image": ["rtx_frames_success_17/49_bridge#episode_8717/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8717/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cucumber."}, {"from": "gpt", "value": "<point>[[480, 9], [422, 99], [323, 209], [314, 482]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cucumber towards the cup."}, {"from": "gpt", "value": "<point>[[353, 341], [462, 348], [551, 351], [596, 433]]</point>"}]}
{"id": 3294, "image": ["rtx_frames_success_17/49_bridge#episode_8718/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8751/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[655, 28], [683, 188], [695, 409], [667, 807]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cucumber."}, {"from": "gpt", "value": "<point>[[328, 19], [432, 144], [556, 346], [665, 753]]</point>"}]}
{"id": 3296, "image": ["rtx_frames_success_17/49_bridge#episode_8753/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8755/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the saucepan."}, {"from": "gpt", "value": "<point>[[485, 198], [427, 256], [399, 315], [418, 346]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the corn cob."}, {"from": "gpt", "value": "<point>[[335, 66], [259, 179], [173, 304], [173, 464]]</point>"}]}
{"id": 3298, "image": ["rtx_frames_success_17/49_bridge#episode_8758/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8758/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the dish."}, {"from": "gpt", "value": "<point>[[258, 336], [335, 369], [508, 459], [630, 407]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bowl behind the towel."}, {"from": "gpt", "value": "<point>[[510, 231], [432, 242], [321, 275], [284, 355]]</point>"}]}
{"id": 3300, "image": ["rtx_frames_success_17/49_bridge#episode_8761/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8763/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the basket of clothes."}, {"from": "gpt", "value": "<point>[[145, 73], [342, 193], [660, 456], [791, 748]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the pot."}, {"from": "gpt", "value": "<point>[[602, 115], [420, 205], [236, 259], [238, 384]]</point>"}]}
{"id": 3302, "image": ["rtx_frames_success_17/49_bridge#episode_8769/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8779/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the item."}, {"from": "gpt", "value": "<point>[[418, 235], [526, 264], [586, 261], [628, 311]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pickle."}, {"from": "gpt", "value": "<point>[[630, 162], [452, 247], [191, 367], [122, 511]]</point>"}]}
{"id": 3304, "image": ["rtx_frames_success_17/49_bridge#episode_8810/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8814/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[291, 59], [404, 193], [545, 315], [561, 508]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[351, 174], [307, 235], [233, 304], [214, 395]]</point>"}]}
{"id": 3306, "image": ["rtx_frames_success_17/49_bridge#episode_8852/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8865/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi piece."}, {"from": "gpt", "value": "<point>[[355, 113], [332, 245], [284, 372], [296, 518]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brinjal."}, {"from": "gpt", "value": "<point>[[307, 80], [268, 224], [242, 381], [298, 572]]</point>"}]}
{"id": 3308, "image": ["rtx_frames_success_17/49_bridge#episode_8881/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8887/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the cup."}, {"from": "gpt", "value": "<point>[[600, 146], [533, 179], [395, 235], [379, 344]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[397, 405], [471, 461], [544, 520], [519, 645]]</point>"}]}
{"id": 3310, "image": ["rtx_frames_success_17/49_bridge#episode_8903/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8904/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the spoon."}, {"from": "gpt", "value": "<point>[[397, 129], [473, 195], [549, 268], [552, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[514, 289], [409, 301], [268, 313], [208, 351]]</point>"}]}
{"id": 3312, "image": ["rtx_frames_success_17/49_bridge#episode_8904/frame_15.png", "rtx_frames_success_17/49_bridge#episode_8912/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the knife towards the blue towel."}, {"from": "gpt", "value": "<point>[[194, 301], [312, 440], [356, 518], [328, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the food item."}, {"from": "gpt", "value": "<point>[[427, 129], [446, 252], [459, 424], [371, 602]]</point>"}]}
{"id": 3314, "image": ["rtx_frames_success_17/49_bridge#episode_8920/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8932/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cloth."}, {"from": "gpt", "value": "<point>[[318, 207], [385, 235], [482, 301], [489, 355]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red fork."}, {"from": "gpt", "value": "<point>[[222, 266], [342, 322], [434, 412], [492, 553]]</point>"}]}
{"id": 3316, "image": ["rtx_frames_success_17/49_bridge#episode_8962/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8982/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow towel."}, {"from": "gpt", "value": "<point>[[221, 198], [245, 282], [305, 424], [295, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[427, 122], [501, 285], [581, 520], [540, 689]]</point>"}]}
{"id": 3318, "image": ["rtx_frames_success_17/49_bridge#episode_8985/frame_0.png", "rtx_frames_success_17/49_bridge#episode_8985/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the spoon."}, {"from": "gpt", "value": "<point>[[401, 96], [372, 162], [339, 214], [346, 287]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon over the orange square."}, {"from": "gpt", "value": "<point>[[288, 61], [439, 264], [628, 433], [642, 631]]</point>"}]}
{"id": 3320, "image": ["rtx_frames_success_17/49_bridge#episode_9001/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9005/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[422, 193], [545, 280], [664, 384], [750, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the food item."}, {"from": "gpt", "value": "<point>[[284, 89], [381, 195], [476, 322], [574, 645]]</point>"}]}
{"id": 3322, "image": ["rtx_frames_success_17/49_bridge#episode_9039/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9057/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[420, 5], [388, 75], [372, 184], [397, 325]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower over the pot."}, {"from": "gpt", "value": "<point>[[464, 165], [401, 299], [383, 386], [406, 527]]</point>"}]}
{"id": 3324, "image": ["rtx_frames_success_17/49_bridge#episode_9066/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9084/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the cloth."}, {"from": "gpt", "value": "<point>[[78, 344], [169, 386], [263, 459], [369, 449]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spoon."}, {"from": "gpt", "value": "<point>[[418, 388], [286, 407], [173, 426], [125, 471]]</point>"}]}
{"id": 3326, "image": ["rtx_frames_success_17/49_bridge#episode_9085/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9114/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[612, 184], [542, 280], [466, 400], [466, 515]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[734, 278], [626, 358], [473, 421], [462, 591]]</point>"}]}
{"id": 3328, "image": ["rtx_frames_success_17/49_bridge#episode_9123/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9125/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the bread loaf."}, {"from": "gpt", "value": "<point>[[231, 219], [378, 353], [496, 449], [544, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[392, 188], [494, 249], [600, 346], [674, 504]]</point>"}]}
{"id": 3330, "image": ["rtx_frames_success_17/49_bridge#episode_9132/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9136/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[284, 132], [397, 212], [538, 393], [600, 675]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[750, 26], [686, 118], [595, 228], [556, 466]]</point>"}]}
{"id": 3332, "image": ["rtx_frames_success_17/49_bridge#episode_9139/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9142/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the knife."}, {"from": "gpt", "value": "<point>[[296, 12], [263, 80], [228, 254], [258, 459]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the cloth."}, {"from": "gpt", "value": "<point>[[362, 181], [522, 346], [628, 496], [625, 666]]</point>"}]}
{"id": 3334, "image": ["rtx_frames_success_17/49_bridge#episode_9157/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9163/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[420, 16], [452, 101], [503, 261], [484, 442]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[476, 151], [528, 280], [561, 424], [544, 656]]</point>"}]}
{"id": 3336, "image": ["rtx_frames_success_17/49_bridge#episode_9163/frame_15.png", "rtx_frames_success_17/49_bridge#episode_9164/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon towards the front left corner."}, {"from": "gpt", "value": "<point>[[506, 628], [390, 553], [245, 454], [139, 487]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the potato."}, {"from": "gpt", "value": "<point>[[304, 136], [358, 165], [409, 188], [394, 287]]</point>"}]}
{"id": 3338, "image": ["rtx_frames_success_17/49_bridge#episode_9169/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9170/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sushi."}, {"from": "gpt", "value": "<point>[[589, 129], [641, 296], [656, 445], [641, 649]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[471, 42], [575, 160], [656, 353], [676, 494]]</point>"}]}
{"id": 3340, "image": ["rtx_frames_success_17/49_bridge#episode_9170/frame_15.png", "rtx_frames_success_17/49_bridge#episode_9171/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spatula towards the towel."}, {"from": "gpt", "value": "<point>[[302, 287], [242, 336], [168, 414], [182, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the can of cherries."}, {"from": "gpt", "value": "<point>[[454, 12], [386, 64], [309, 141], [256, 339]]</point>"}]}
{"id": 3342, "image": ["rtx_frames_success_17/49_bridge#episode_9171/frame_15.png", "rtx_frames_success_17/49_bridge#episode_9176/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the sushi roll in the pot."}, {"from": "gpt", "value": "<point>[[314, 191], [418, 219], [568, 268], [724, 414]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[499, 282], [422, 325], [326, 381], [307, 520]]</point>"}]}
{"id": 3344, "image": ["rtx_frames_success_17/49_bridge#episode_9213/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9216/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[445, 9], [482, 158], [508, 334], [459, 513]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the sandwich."}, {"from": "gpt", "value": "<point>[[563, 344], [424, 353], [304, 367], [221, 414]]</point>"}]}
{"id": 3346, "image": ["rtx_frames_success_17/49_bridge#episode_9216/frame_15.png", "rtx_frames_success_17/49_bridge#episode_9218/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the sandwich over the yellow cloth."}, {"from": "gpt", "value": "<point>[[212, 351], [404, 456], [565, 551], [681, 755]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the cheese."}, {"from": "gpt", "value": "<point>[[621, 101], [538, 165], [392, 268], [325, 372]]</point>"}]}
{"id": 3348, "image": ["rtx_frames_success_17/49_bridge#episode_9218/frame_15.png", "rtx_frames_success_17/49_bridge#episode_9221/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the pan."}, {"from": "gpt", "value": "<point>[[319, 144], [238, 216], [155, 318], [171, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[642, 348], [572, 407], [494, 515], [514, 605]]</point>"}]}
{"id": 3350, "image": ["rtx_frames_success_17/49_bridge#episode_9225/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9225/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up a piece of sushi."}, {"from": "gpt", "value": "<point>[[378, 325], [304, 348], [194, 412], [178, 482]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the brush on the left of the pan."}, {"from": "gpt", "value": "<point>[[206, 426], [332, 431], [429, 454], [461, 496]]</point>"}]}
{"id": 3352, "image": ["rtx_frames_success_17/49_bridge#episode_9244/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9252/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red fork."}, {"from": "gpt", "value": "<point>[[454, 365], [559, 426], [621, 475], [644, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot containing the broccoli."}, {"from": "gpt", "value": "<point>[[376, 256], [510, 322], [595, 391], [630, 487]]</point>"}]}
{"id": 3354, "image": ["rtx_frames_success_17/49_bridge#episode_9256/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9257/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the piece of food."}, {"from": "gpt", "value": "<point>[[247, 325], [379, 348], [547, 358], [635, 400]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[286, 240], [404, 292], [545, 353], [646, 520]]</point>"}]}
{"id": 3356, "image": ["rtx_frames_success_17/49_bridge#episode_9268/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9269/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[526, 115], [584, 219], [639, 325], [635, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot."}, {"from": "gpt", "value": "<point>[[420, 353], [386, 386], [323, 440], [298, 541]]</point>"}]}
{"id": 3358, "image": ["rtx_frames_success_17/49_bridge#episode_9282/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9288/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue fork."}, {"from": "gpt", "value": "<point>[[549, 92], [655, 320], [718, 471], [701, 755]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the metal pot."}, {"from": "gpt", "value": "<point>[[566, 169], [612, 271], [672, 421], [626, 781]]</point>"}]}
{"id": 3360, "image": ["rtx_frames_success_17/49_bridge#episode_9288/frame_15.png", "rtx_frames_success_17/49_bridge#episode_9290/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the metal pot on the table."}, {"from": "gpt", "value": "<point>[[609, 529], [512, 442], [418, 402], [309, 414]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the pot."}, {"from": "gpt", "value": "<point>[[194, 308], [266, 306], [349, 308], [399, 336]]</point>"}]}
{"id": 3362, "image": ["rtx_frames_success_17/49_bridge#episode_9291/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9300/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[536, 313], [581, 362], [611, 421], [591, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[478, 325], [545, 313], [628, 261], [688, 254]]</point>"}]}
{"id": 3364, "image": ["rtx_frames_success_17/49_bridge#episode_9304/frame_0.png", "rtx_frames_success_17/49_bridge#episode_9313/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[304, 212], [390, 292], [531, 442], [593, 659]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the yellow brush."}, {"from": "gpt", "value": "<point>[[390, 296], [429, 386], [475, 492], [459, 598]]</point>"}]}
{"id": 3366, "image": ["rtx_frames_success_17/49_bridge#episode_9345/frame_0.png", "rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_5/frame_0.png"], "height": [480, 128], "width": [640, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position above the pot."}, {"from": "gpt", "value": "<point>[[596, 172], [496, 200], [418, 221], [372, 287]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[263, 959], [246, 769], [248, 487], [278, 310]]</point>"}]}
{"id": 3368, "image": ["rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_12/frame_0.png", "rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_13/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the filing cabinet."}, {"from": "gpt", "value": "<point>[[252, 972], [235, 795], [250, 513], [289, 320]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle of the top drawer."}, {"from": "gpt", "value": "<point>[[242, 953], [242, 808], [259, 511], [278, 336]]</point>"}]}
{"id": 3370, "image": ["rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_15/frame_0.png", "rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_19/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the drawer."}, {"from": "gpt", "value": "<point>[[274, 953], [263, 744], [265, 547], [280, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle of the drawer."}, {"from": "gpt", "value": "<point>[[267, 955], [258, 778], [265, 532], [282, 320]]</point>"}]}
{"id": 3372, "image": ["rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_20/frame_0.png", "rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_28/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the filing cabinet."}, {"from": "gpt", "value": "<point>[[258, 951], [248, 750], [256, 498], [288, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle of the filing cabinet."}, {"from": "gpt", "value": "<point>[[248, 955], [239, 795], [259, 513], [282, 350]]</point>"}]}
{"id": 3374, "image": ["rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_29/frame_0.png", "rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_32/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the drawer."}, {"from": "gpt", "value": "<point>[[276, 945], [242, 750], [244, 519], [278, 305]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle of the filing cabinet drawer."}, {"from": "gpt", "value": "<point>[[252, 962], [229, 716], [237, 455], [263, 203]]</point>"}]}
{"id": 3376, "image": ["rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_39/frame_0.png", "rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_47/frame_0.png"], "height": [128, 128], "width": [128, 128], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the drawer."}, {"from": "gpt", "value": "<point>[[242, 1000], [224, 782], [235, 472], [271, 293]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[254, 979], [244, 791], [259, 583], [299, 367]]</point>"}]}
{"id": 3378, "image": ["rtx_frames_success_18/6_utokyo_pr2_opening_fridge_converted_externally_to_rlds#episode_53/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_3/frame_0.png"], "height": [128, 224], "width": [128, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the drawer."}, {"from": "gpt", "value": "<point>[[261, 959], [258, 742], [267, 553], [289, 350]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the bag of chips."}, {"from": "gpt", "value": "<point>[[555, 489], [489, 477], [421, 479], [352, 496]]</point>"}]}
{"id": 3380, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_3/frame_15.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_5/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the left bread."}, {"from": "gpt", "value": "<point>[[370, 496], [340, 524], [321, 549], [297, 583]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red can."}, {"from": "gpt", "value": "<point>[[560, 511], [508, 502], [455, 532], [429, 581]]</point>"}]}
{"id": 3382, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_8/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_21/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mustard bottle."}, {"from": "gpt", "value": "<point>[[549, 504], [481, 477], [410, 470], [338, 481]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the coke can."}, {"from": "gpt", "value": "<point>[[564, 485], [526, 492], [498, 511], [477, 543]]</point>"}]}
{"id": 3384, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_24/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_24/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red block."}, {"from": "gpt", "value": "<point>[[566, 477], [509, 468], [455, 464], [397, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the red block."}, {"from": "gpt", "value": "<point>[[397, 461], [374, 472], [357, 494], [348, 528], [344, 575]]</point>"}]}
{"id": 3386, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_40/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_40/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the juice box."}, {"from": "gpt", "value": "<point>[[504, 440], [568, 477], [613, 519], [648, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the juice box."}, {"from": "gpt", "value": "<point>[[654, 549], [669, 585], [671, 620], [669, 671]]</point>"}]}
{"id": 3388, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_44/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_61/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the white carton."}, {"from": "gpt", "value": "<point>[[568, 487], [583, 530], [594, 575], [528, 455]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the milk carton."}, {"from": "gpt", "value": "<point>[[543, 492], [517, 509], [504, 534], [502, 573]]</point>"}]}
{"id": 3390, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_69/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_78/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[562, 485], [577, 504], [585, 526], [586, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the milk carton."}, {"from": "gpt", "value": "<point>[[566, 498], [511, 492], [445, 504], [398, 530]]</point>"}]}
{"id": 3392, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_81/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_87/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the milk carton."}, {"from": "gpt", "value": "<point>[[586, 492], [579, 523], [583, 551], [594, 575]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the red can."}, {"from": "gpt", "value": "<point>[[570, 417], [519, 412], [457, 414], [404, 425]]</point>"}]}
{"id": 3394, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_87/frame_15.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_94/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red can."}, {"from": "gpt", "value": "<point>[[374, 398], [352, 417], [333, 447], [325, 481]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the milk carton."}, {"from": "gpt", "value": "<point>[[564, 491], [628, 491], [677, 508], [727, 545]]</point>"}]}
{"id": 3396, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_96/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_99/frame_0.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red block."}, {"from": "gpt", "value": "<point>[[558, 487], [517, 509], [502, 558], [511, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bread package."}, {"from": "gpt", "value": "<point>[[560, 489], [494, 496], [417, 519], [365, 555]]</point>"}]}
{"id": 3398, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_100/frame_0.png", "rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_100/frame_15.png"], "height": [224, 224], "width": [224, 224], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the glass bottle."}, {"from": "gpt", "value": "<point>[[566, 494], [506, 496], [455, 491], [398, 470]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the glass bottle."}, {"from": "gpt", "value": "<point>[[395, 459], [367, 474], [348, 511], [350, 560]]</point>"}]}
{"id": 3400, "image": ["rtx_frames_success_18/9_asu_table_top_converted_externally_to_rlds#episode_107/frame_0.png", "rtx_frames_success_18/49_bridge#episode_95/frame_0.png"], "height": [224, 480], "width": [224, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green bottle."}, {"from": "gpt", "value": "<point>[[571, 502], [479, 496], [387, 506], [289, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the white object."}, {"from": "gpt", "value": "<point>[[412, 0], [532, 105], [632, 266], [697, 431]]</point>"}]}
{"id": 3402, "image": ["rtx_frames_success_18/49_bridge#episode_96/frame_0.png", "rtx_frames_success_18/49_bridge#episode_96/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the metal bowl."}, {"from": "gpt", "value": "<point>[[402, 3], [560, 145], [664, 286], [744, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bowl towards the purple cloth."}, {"from": "gpt", "value": "<point>[[686, 241], [680, 371], [656, 534], [626, 657]]</point>"}]}
{"id": 3404, "image": ["rtx_frames_success_18/49_bridge#episode_97/frame_0.png", "rtx_frames_success_18/49_bridge#episode_98/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the tilted pot."}, {"from": "gpt", "value": "<point>[[241, 185], [274, 256], [308, 301], [365, 311]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[573, 401], [462, 388], [333, 396], [250, 426]]</point>"}]}
{"id": 3406, "image": ["rtx_frames_success_18/49_bridge#episode_935/frame_0.png", "rtx_frames_success_18/49_bridge#episode_944/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[607, 180], [645, 303], [686, 519], [701, 682]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brush."}, {"from": "gpt", "value": "<point>[[485, 68], [346, 183], [231, 313], [141, 471]]</point>"}]}
{"id": 3408, "image": ["rtx_frames_success_18/49_bridge#episode_953/frame_0.png", "rtx_frames_success_18/49_bridge#episode_963/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[333, 3], [404, 206], [536, 471], [656, 662]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[111, 271], [207, 409], [314, 506], [423, 569]]</point>"}]}
{"id": 3410, "image": ["rtx_frames_success_18/49_bridge#episode_970/frame_0.png", "rtx_frames_success_18/49_bridge#episode_976/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the towel."}, {"from": "gpt", "value": "<point>[[481, 10], [483, 143], [470, 321], [445, 476]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to open the cabinet door."}, {"from": "gpt", "value": "<point>[[352, 561], [472, 604], [545, 614], [645, 617]]</point>"}]}
{"id": 3412, "image": ["rtx_frames_success_18/49_bridge#episode_986/frame_0.png", "rtx_frames_success_18/49_bridge#episode_990/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[647, 684], [735, 551], [767, 396], [752, 246]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[335, 143], [440, 236], [571, 461], [624, 652]]</point>"}]}
{"id": 3414, "image": ["rtx_frames_success_18/49_bridge#episode_9354/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9354/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the jar of almonds."}, {"from": "gpt", "value": "<point>[[363, 296], [342, 363], [353, 451], [387, 524]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to tilt the jar over the pot to pour the almonds."}, {"from": "gpt", "value": "<point>[[320, 333], [370, 388], [425, 486], [447, 576]]</point>"}]}
{"id": 3416, "image": ["rtx_frames_success_18/49_bridge#episode_9377/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9403/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[498, 180], [583, 323], [660, 479], [718, 644]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[199, 358], [312, 351], [429, 361], [556, 393]]</point>"}]}
{"id": 3418, "image": ["rtx_frames_success_18/49_bridge#episode_9409/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9422/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[233, 251], [306, 308], [385, 414], [429, 531]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the fork."}, {"from": "gpt", "value": "<point>[[203, 128], [273, 273], [321, 451], [340, 591]]</point>"}]}
{"id": 3420, "image": ["rtx_frames_success_18/49_bridge#episode_9424/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9424/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to hover over the fruit."}, {"from": "gpt", "value": "<point>[[538, 271], [492, 261], [453, 283], [429, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the fruit across the table."}, {"from": "gpt", "value": "<point>[[585, 226], [647, 293], [692, 404], [703, 514]]</point>"}]}
{"id": 3422, "image": ["rtx_frames_success_18/49_bridge#episode_9432/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9488/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan with broccoli."}, {"from": "gpt", "value": "<point>[[214, 396], [329, 376], [444, 366], [568, 386]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[406, 115], [461, 195], [523, 336], [556, 431]]</point>"}]}
{"id": 3424, "image": ["rtx_frames_success_18/49_bridge#episode_9489/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9505/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the cloth."}, {"from": "gpt", "value": "<point>[[226, 293], [231, 393], [248, 466], [286, 554]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[173, 5], [261, 90], [342, 221], [385, 336]]</point>"}]}
{"id": 3426, "image": ["rtx_frames_success_18/49_bridge#episode_9506/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9517/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[291, 150], [278, 298], [301, 516], [327, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[487, 183], [502, 288], [530, 393], [577, 476]]</point>"}]}
{"id": 3428, "image": ["rtx_frames_success_18/49_bridge#episode_9519/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9536/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[327, 10], [380, 135], [496, 291], [618, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[438, 125], [541, 221], [669, 409], [722, 526]]</point>"}]}
{"id": 3430, "image": ["rtx_frames_success_18/49_bridge#episode_9541/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9543/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[182, 125], [258, 348], [348, 516], [477, 672]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[457, 211], [348, 188], [242, 241], [132, 409]]</point>"}]}
{"id": 3432, "image": ["rtx_frames_success_18/49_bridge#episode_9583/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9588/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[336, 45], [306, 190], [306, 436], [335, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[586, 416], [556, 464], [515, 471], [455, 446]]</point>"}]}
{"id": 3434, "image": ["rtx_frames_success_18/49_bridge#episode_9593/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9593/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[600, 431], [536, 506], [470, 624], [436, 719]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the spoon towards the pot."}, {"from": "gpt", "value": "<point>[[291, 436], [340, 436], [391, 424], [444, 391]]</point>"}]}
{"id": 3436, "image": ["rtx_frames_success_18/49_bridge#episode_9596/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9596/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the pan containing food."}, {"from": "gpt", "value": "<point>[[192, 40], [293, 190], [434, 333], [575, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the food towards the cloth."}, {"from": "gpt", "value": "<point>[[457, 65], [395, 143], [344, 238], [299, 358]]</point>"}]}
{"id": 3438, "image": ["rtx_frames_success_18/49_bridge#episode_9613/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9623/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the potato."}, {"from": "gpt", "value": "<point>[[468, 145], [385, 236], [303, 366], [263, 481]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[400, 135], [500, 288], [600, 554], [628, 749]]</point>"}]}
{"id": 3440, "image": ["rtx_frames_success_18/49_bridge#episode_9623/frame_15.png", "rtx_frames_success_18/49_bridge#episode_9634/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the spoon across the table."}, {"from": "gpt", "value": "<point>[[306, 371], [263, 406], [231, 469], [214, 531]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the carrot in the container."}, {"from": "gpt", "value": "<point>[[620, 444], [592, 479], [530, 504], [466, 494]]</point>"}]}
{"id": 3442, "image": ["rtx_frames_success_18/49_bridge#episode_9646/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9663/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[229, 138], [310, 293], [383, 486], [430, 632]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[432, 218], [404, 268], [346, 333], [297, 366]]</point>"}]}
{"id": 3444, "image": ["rtx_frames_success_18/49_bridge#episode_9670/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9670/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the can."}, {"from": "gpt", "value": "<point>[[605, 672], [543, 479], [423, 311], [263, 223]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the can towards the spoon."}, {"from": "gpt", "value": "<point>[[312, 253], [415, 373], [504, 516], [592, 662]]</point>"}]}
{"id": 3446, "image": ["rtx_frames_success_18/49_bridge#episode_9672/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9710/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spatula."}, {"from": "gpt", "value": "<point>[[291, 125], [368, 261], [421, 396], [496, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the rag."}, {"from": "gpt", "value": "<point>[[406, 386], [508, 351], [594, 368], [694, 436]]</point>"}]}
{"id": 3448, "image": ["rtx_frames_success_18/49_bridge#episode_9710/frame_15.png", "rtx_frames_success_18/49_bridge#episode_9711/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to slide the rag towards the can."}, {"from": "gpt", "value": "<point>[[524, 414], [496, 356], [427, 316], [363, 316]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pan containing the sushi roll."}, {"from": "gpt", "value": "<point>[[195, 48], [344, 140], [496, 288], [585, 421]]</point>"}]}
{"id": 3450, "image": ["rtx_frames_success_18/49_bridge#episode_9711/frame_15.png", "rtx_frames_success_18/49_bridge#episode_9713/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the green cloth with the sushi roll."}, {"from": "gpt", "value": "<point>[[453, 40], [385, 115], [346, 211], [314, 311]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the butter stick."}, {"from": "gpt", "value": "<point>[[581, 115], [498, 145], [406, 211], [323, 303]]</point>"}]}
{"id": 3452, "image": ["rtx_frames_success_18/49_bridge#episode_9715/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9723/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the carrot on the plate."}, {"from": "gpt", "value": "<point>[[357, 238], [470, 343], [564, 466], [639, 604]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pot."}, {"from": "gpt", "value": "<point>[[160, 160], [192, 253], [252, 336], [327, 378]]</point>"}]}
{"id": 3454, "image": ["rtx_frames_success_18/49_bridge#episode_9724/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9729/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the potato from the pan."}, {"from": "gpt", "value": "<point>[[658, 55], [564, 211], [502, 341], [436, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green spatula."}, {"from": "gpt", "value": "<point>[[461, 105], [494, 308], [558, 504], [652, 679]]</point>"}]}
{"id": 3456, "image": ["rtx_frames_success_18/49_bridge#episode_9729/frame_15.png", "rtx_frames_success_18/49_bridge#episode_9754/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the green spatula towards the upper left corner of the board."}, {"from": "gpt", "value": "<point>[[553, 368], [487, 331], [414, 308], [353, 296]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spoon."}, {"from": "gpt", "value": "<point>[[519, 216], [543, 263], [558, 326], [551, 401]]</point>"}]}
{"id": 3458, "image": ["rtx_frames_success_18/49_bridge#episode_9761/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9770/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[372, 60], [336, 175], [306, 246], [241, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up clothes from a basket."}, {"from": "gpt", "value": "<point>[[485, 524], [434, 581], [365, 599], [291, 591]]</point>"}]}
{"id": 3460, "image": ["rtx_frames_success_18/49_bridge#episode_9770/frame_15.png", "rtx_frames_success_18/49_bridge#episode_9776/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the clothes into the washing machine."}, {"from": "gpt", "value": "<point>[[267, 496], [335, 421], [417, 371], [523, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the item in the bowl."}, {"from": "gpt", "value": "<point>[[250, 150], [383, 308], [496, 484], [562, 649]]</point>"}]}
{"id": 3462, "image": ["rtx_frames_success_18/49_bridge#episode_9780/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9783/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[414, 386], [427, 479], [459, 526], [508, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pot."}, {"from": "gpt", "value": "<point>[[620, 419], [541, 489], [436, 541], [331, 534]]</point>"}]}
{"id": 3464, "image": ["rtx_frames_success_18/49_bridge#episode_9783/frame_15.png", "rtx_frames_success_18/49_bridge#episode_9825/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot over the cloth."}, {"from": "gpt", "value": "<point>[[408, 446], [485, 446], [564, 439], [643, 401]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the microwave handle."}, {"from": "gpt", "value": "<point>[[570, 524], [496, 446], [438, 343], [410, 211]]</point>"}]}
{"id": 3466, "image": ["rtx_frames_success_18/49_bridge#episode_9826/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9826/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red fork."}, {"from": "gpt", "value": "<point>[[461, 105], [395, 226], [344, 373], [310, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the fork to the right side of the table."}, {"from": "gpt", "value": "<point>[[269, 426], [397, 466], [534, 539], [630, 627]]</point>"}]}
{"id": 3468, "image": ["rtx_frames_success_18/49_bridge#episode_9862/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9900/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the corn."}, {"from": "gpt", "value": "<point>[[556, 401], [543, 308], [489, 251], [408, 293]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spatula."}, {"from": "gpt", "value": "<point>[[336, 241], [305, 376], [288, 536], [289, 682]]</point>"}]}
{"id": 3470, "image": ["rtx_frames_success_18/49_bridge#episode_9903/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9968/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mushroom."}, {"from": "gpt", "value": "<point>[[436, 216], [470, 316], [485, 439], [483, 574]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[355, 73], [417, 160], [477, 276], [504, 381]]</point>"}]}
{"id": 3472, "image": ["rtx_frames_success_18/49_bridge#episode_9980/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9985/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the vessel."}, {"from": "gpt", "value": "<point>[[489, 218], [464, 298], [417, 343], [376, 363]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the spice container."}, {"from": "gpt", "value": "<point>[[414, 291], [436, 358], [444, 441], [444, 511]]</point>"}]}
{"id": 3474, "image": ["rtx_frames_success_18/49_bridge#episode_9986/frame_0.png", "rtx_frames_success_18/49_bridge#episode_9993/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the spice container."}, {"from": "gpt", "value": "<point>[[169, 283], [229, 326], [312, 346], [406, 351]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the carrot."}, {"from": "gpt", "value": "<point>[[564, 140], [509, 278], [474, 481], [468, 647]]</point>"}]}
{"id": 3476, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_0/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_11/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[418, 532], [269, 565], [197, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth across the table."}, {"from": "gpt", "value": "<point>[[449, 499], [302, 554], [205, 624]]</point>"}]}
{"id": 3478, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_19/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_19/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[651, 406], [615, 447], [598, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[479, 543], [341, 617], [277, 713]]</point>"}]}
{"id": 3480, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_23/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_26/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower towards the pot."}, {"from": "gpt", "value": "<point>[[540, 517], [263, 572], [188, 620]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bottle towards the pot."}, {"from": "gpt", "value": "<point>[[529, 410], [449, 469], [429, 539]]</point>"}]}
{"id": 3482, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_32/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_34/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the object towards the second container."}, {"from": "gpt", "value": "<point>[[571, 425], [454, 506], [371, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[515, 462], [485, 539], [499, 606]]</point>"}]}
{"id": 3484, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_38/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_44/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[540, 517], [324, 650], [280, 705]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the ranch bottle over the pot."}, {"from": "gpt", "value": "<point>[[510, 421], [457, 462], [457, 543]]</point>"}]}
{"id": 3486, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_48/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_48/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the brown cup."}, {"from": "gpt", "value": "<point>[[524, 510], [366, 451], [280, 495]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[302, 432], [374, 432], [407, 480]]</point>"}]}
{"id": 3488, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_54/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_59/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth to the left."}, {"from": "gpt", "value": "<point>[[479, 524], [338, 561], [288, 643]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object in the first container."}, {"from": "gpt", "value": "<point>[[546, 500], [445, 496], [376, 526]]</point>"}]}
{"id": 3490, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_83/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_104/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[532, 495], [357, 536], [271, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[568, 550], [427, 454], [371, 480]]</point>"}]}
{"id": 3492, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_107/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_110/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[518, 480], [307, 543], [216, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[507, 543], [457, 495], [449, 462]]</point>"}]}
{"id": 3494, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_118/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_121/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the object over the second container."}, {"from": "gpt", "value": "<point>[[471, 425], [341, 465], [316, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the item over the second container."}, {"from": "gpt", "value": "<point>[[499, 417], [363, 558], [385, 735]]</point>"}]}
{"id": 3496, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_137/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_145/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[529, 358], [429, 384], [404, 447]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[540, 506], [391, 665], [388, 735]]</point>"}]}
{"id": 3498, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_152/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_168/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left."}, {"from": "gpt", "value": "<point>[[463, 650], [299, 643], [219, 776]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to release the pot."}, {"from": "gpt", "value": "<point>[[540, 513], [360, 587], [313, 650]]</point>"}]}
{"id": 3500, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_168/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_169/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place a bottle in a pot."}, {"from": "gpt", "value": "<point>[[357, 488], [418, 499], [446, 547]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[562, 443], [510, 425], [488, 462]]</point>"}]}
{"id": 3502, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_169/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_184/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left."}, {"from": "gpt", "value": "<point>[[377, 550], [230, 595], [163, 650]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[537, 506], [321, 513], [247, 539]]</point>"}]}
{"id": 3504, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_199/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_224/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[590, 417], [532, 495], [485, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to stack the blue cup in the brown cup."}, {"from": "gpt", "value": "<point>[[573, 351], [410, 447], [357, 517]]</point>"}]}
{"id": 3506, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_231/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_231/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[546, 406], [529, 469], [568, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[465, 591], [321, 620], [241, 724]]</point>"}]}
{"id": 3508, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_233/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_250/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[507, 373], [512, 484], [551, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object in the first container."}, {"from": "gpt", "value": "<point>[[543, 517], [382, 543], [288, 598]]</point>"}]}
{"id": 3510, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_252/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_252/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[590, 654], [529, 524], [557, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[526, 502], [357, 587], [247, 705]]</point>"}]}
{"id": 3512, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_255/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_256/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bottle towards the pot."}, {"from": "gpt", "value": "<point>[[596, 465], [479, 488], [446, 550]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the cloth."}, {"from": "gpt", "value": "<point>[[562, 388], [546, 488], [515, 521]]</point>"}]}
{"id": 3514, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_256/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_257/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[388, 617], [283, 635], [227, 720]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[501, 369], [521, 473], [568, 517]]</point>"}]}
{"id": 3516, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_257/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_258/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[407, 572], [296, 617], [183, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bottle towards the pot."}, {"from": "gpt", "value": "<point>[[463, 417], [427, 484], [438, 561]]</point>"}]}
{"id": 3518, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_269/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_272/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[551, 451], [535, 517], [568, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[565, 510], [507, 543], [496, 602]]</point>"}]}
{"id": 3520, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_285/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_286/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower the pot towards the bottle."}, {"from": "gpt", "value": "<point>[[537, 506], [357, 654], [324, 739]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object in the first container."}, {"from": "gpt", "value": "<point>[[543, 495], [427, 580], [421, 654]]</point>"}]}
{"id": 3522, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_286/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_294/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the object towards the second container."}, {"from": "gpt", "value": "<point>[[526, 340], [510, 406], [551, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cloth left."}, {"from": "gpt", "value": "<point>[[385, 521], [247, 591], [175, 654]]</point>"}]}
{"id": 3524, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_310/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_310/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the bottle."}, {"from": "gpt", "value": "<point>[[540, 517], [623, 517], [640, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bottle towards the pot."}, {"from": "gpt", "value": "<point>[[612, 458], [496, 473], [449, 539]]</point>"}]}
{"id": 3526, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_312/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_317/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the brown cup."}, {"from": "gpt", "value": "<point>[[524, 506], [416, 440], [355, 428]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[546, 491], [357, 665], [321, 724]]</point>"}]}
{"id": 3528, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_330/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_330/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[584, 532], [471, 524], [463, 473]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[327, 561], [213, 602], [158, 646]]</point>"}]}
{"id": 3530, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_331/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_335/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the item towards the second container."}, {"from": "gpt", "value": "<point>[[546, 399], [341, 576], [313, 720]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the smaller cup."}, {"from": "gpt", "value": "<point>[[560, 476], [582, 469], [612, 495]]</point>"}]}
{"id": 3532, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_348/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_357/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the item in the first container."}, {"from": "gpt", "value": "<point>[[548, 502], [393, 513], [335, 580]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to stack the blue cup in the brown cup."}, {"from": "gpt", "value": "<point>[[512, 273], [501, 425], [548, 499]]</point>"}]}
{"id": 3534, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_363/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_363/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lift the pot."}, {"from": "gpt", "value": "<point>[[540, 506], [338, 680], [274, 757]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the bottle into the pot."}, {"from": "gpt", "value": "<point>[[307, 661], [460, 609], [457, 536]]</point>"}]}
{"id": 3536, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_364/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_372/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[554, 665], [485, 521], [512, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the object towards the second container."}, {"from": "gpt", "value": "<point>[[576, 414], [438, 465], [396, 547]]</point>"}]}
{"id": 3538, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_375/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_377/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the object towards the second container."}, {"from": "gpt", "value": "<point>[[463, 417], [305, 595], [288, 705]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[529, 502], [374, 606], [357, 680]]</point>"}]}
{"id": 3540, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_382/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_383/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[543, 491], [321, 591], [230, 709]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the item in the first container."}, {"from": "gpt", "value": "<point>[[546, 499], [374, 536], [346, 587]]</point>"}]}
{"id": 3542, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_393/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_393/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[590, 358], [535, 414], [515, 476], [554, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[427, 528], [280, 569], [172, 639]]</point>"}]}
{"id": 3544, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_397/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_397/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[543, 343], [515, 436], [557, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left."}, {"from": "gpt", "value": "<point>[[454, 521], [269, 580], [172, 650]]</point>"}]}
{"id": 3546, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_402/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_411/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the brown cup."}, {"from": "gpt", "value": "<point>[[554, 458], [488, 539], [457, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[557, 425], [463, 458], [446, 550]]</point>"}]}
{"id": 3548, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_418/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_423/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object in the first container."}, {"from": "gpt", "value": "<point>[[540, 513], [474, 480], [440, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object in the first container."}, {"from": "gpt", "value": "<point>[[582, 466], [452, 512], [409, 541]]</point>"}]}
{"id": 3550, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_427/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_430/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[460, 510], [421, 620], [274, 657]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[532, 510], [374, 650], [355, 750]]</point>"}]}
{"id": 3552, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_434/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_436/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the item in the first container."}, {"from": "gpt", "value": "<point>[[551, 476], [465, 506], [399, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[518, 414], [554, 451]]</point>"}]}
{"id": 3554, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_440/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_441/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[537, 469], [435, 432], [368, 476]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[535, 458], [551, 532]]</point>"}]}
{"id": 3556, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_443/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_449/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[540, 524], [418, 517], [305, 606]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the pot."}, {"from": "gpt", "value": "<point>[[526, 488], [391, 480], [260, 569]]</point>"}]}
{"id": 3558, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_449/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_460/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the pot towards the bottle."}, {"from": "gpt", "value": "<point>[[252, 554], [346, 499], [438, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to sweep the cloth across the table to the left."}, {"from": "gpt", "value": "<point>[[468, 591], [374, 584], [283, 739]]</point>"}]}
{"id": 3560, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_463/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_468/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[432, 558], [285, 595], [227, 680]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[551, 443], [493, 406], [474, 454]]</point>"}]}
{"id": 3562, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_470/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_471/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object in the first container."}, {"from": "gpt", "value": "<point>[[548, 484], [465, 491], [368, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the bottle."}, {"from": "gpt", "value": "<point>[[540, 465], [535, 414], [463, 414]]</point>"}]}
{"id": 3564, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_471/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_475/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bottle towards the pot."}, {"from": "gpt", "value": "<point>[[465, 443], [413, 488], [432, 547]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cloth to the left."}, {"from": "gpt", "value": "<point>[[510, 617], [388, 635], [288, 809]]</point>"}]}
{"id": 3566, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_478/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_479/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[537, 536], [479, 484], [429, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to sweep the cloth across the table to the left."}, {"from": "gpt", "value": "<point>[[515, 572], [402, 587], [319, 753]]</point>"}]}
{"id": 3568, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_483/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_487/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the tiger towards the grey bowl."}, {"from": "gpt", "value": "<point>[[496, 480], [396, 569], [391, 654]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[557, 480], [501, 469], [493, 539]]</point>"}]}
{"id": 3570, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_487/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_488/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left."}, {"from": "gpt", "value": "<point>[[385, 632], [288, 617], [224, 728]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[568, 414], [645, 451], [654, 517]]</point>"}]}
{"id": 3572, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_490/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_492/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[512, 539], [380, 536], [263, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grip the cloth."}, {"from": "gpt", "value": "<point>[[629, 443], [590, 465], [568, 532]]</point>"}]}
{"id": 3574, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_492/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_498/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth towards the left edge of the table."}, {"from": "gpt", "value": "<point>[[418, 635], [296, 628], [241, 720]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[540, 476], [410, 669], [366, 742]]</point>"}]}
{"id": 3576, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_503/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_506/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the bottle into the pot."}, {"from": "gpt", "value": "<point>[[526, 392], [454, 414], [452, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[557, 510], [294, 499], [183, 510]]</point>"}]}
{"id": 3578, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_510/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_516/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the item in the first container."}, {"from": "gpt", "value": "<point>[[546, 458], [421, 454], [310, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[521, 355], [474, 469], [540, 462]]</point>"}]}
{"id": 3580, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_524/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_532/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth to the left."}, {"from": "gpt", "value": "<point>[[488, 558], [324, 635], [291, 750]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[485, 414], [493, 510], [562, 506]]</point>"}]}
{"id": 3582, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_532/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_541/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth to the left."}, {"from": "gpt", "value": "<point>[[416, 595], [310, 654], [288, 698]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the brown cup."}, {"from": "gpt", "value": "<point>[[540, 484], [463, 587], [460, 643]]</point>"}]}
{"id": 3584, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_543/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_547/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[532, 528], [341, 628], [263, 761]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the item in the first container."}, {"from": "gpt", "value": "<point>[[543, 502], [388, 491], [294, 524]]</point>"}]}
{"id": 3586, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_551/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_555/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the item in the first container."}, {"from": "gpt", "value": "<point>[[548, 465], [402, 584], [296, 687]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the cloth."}, {"from": "gpt", "value": "<point>[[604, 406], [535, 417], [529, 447]]</point>"}]}
{"id": 3588, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_555/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_560/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[449, 565], [332, 598], [258, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the item in the first container."}, {"from": "gpt", "value": "<point>[[546, 511], [518, 446], [482, 424]]</point>"}]}
{"id": 3590, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_565/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_566/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[540, 440], [460, 458], [443, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the marker."}, {"from": "gpt", "value": "<point>[[515, 502], [391, 495], [321, 528]]</point>"}]}
{"id": 3592, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_566/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_571/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the bowl."}, {"from": "gpt", "value": "<point>[[352, 499], [399, 458], [452, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[596, 646], [507, 554], [543, 473]]</point>"}]}
{"id": 3594, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_571/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_579/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the cloth to the left."}, {"from": "gpt", "value": "<point>[[493, 480], [346, 536], [255, 613]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[546, 428], [479, 447], [457, 499]]</point>"}]}
{"id": 3596, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_585/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_605/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[554, 432], [493, 454], [446, 414]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[532, 510], [360, 639], [302, 746]]</point>"}]}
{"id": 3598, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_606/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_614/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[499, 502], [402, 506], [382, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[529, 550], [341, 650], [321, 794]]</point>"}]}
{"id": 3600, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_615/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_617/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the blue cup."}, {"from": "gpt", "value": "<point>[[584, 550], [427, 536], [374, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[510, 417], [454, 510]]</point>"}]}
{"id": 3602, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_620/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_620/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[637, 358], [546, 417], [548, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left edge of the table."}, {"from": "gpt", "value": "<point>[[454, 558], [296, 587], [247, 661]]</point>"}]}
{"id": 3604, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_623/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_625/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bottle towards the pot."}, {"from": "gpt", "value": "<point>[[554, 469], [457, 458], [443, 550]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the item in the first container."}, {"from": "gpt", "value": "<point>[[544, 513], [394, 561], [384, 653]]</point>"}]}
{"id": 3606, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_627/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_630/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[551, 373], [504, 428], [543, 473]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the bottle."}, {"from": "gpt", "value": "<point>[[543, 521], [404, 617], [341, 705]]</point>"}]}
{"id": 3608, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_637/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_641/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[535, 524], [413, 624], [380, 720]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the object over the second container."}, {"from": "gpt", "value": "<point>[[432, 436], [324, 451], [341, 550]]</point>"}]}
{"id": 3610, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_657/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_660/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[529, 510], [349, 657], [266, 739]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[427, 491], [319, 561], [244, 628]]</point>"}]}
{"id": 3612, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_667/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_669/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the item in the first container."}, {"from": "gpt", "value": "<point>[[547, 514], [414, 535], [265, 635]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[410, 506], [283, 580], [222, 639]]</point>"}]}
{"id": 3614, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_709/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_709/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[584, 406], [460, 440], [399, 476]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to stack the blue cup in the brown cup."}, {"from": "gpt", "value": "<point>[[346, 358], [468, 436], [465, 536]]</point>"}]}
{"id": 3616, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_721/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_726/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the object in the first container."}, {"from": "gpt", "value": "<point>[[543, 511], [355, 576], [266, 619]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the item towards the second container."}, {"from": "gpt", "value": "<point>[[493, 513], [388, 510], [357, 657]]</point>"}]}
{"id": 3618, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_730/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_733/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object in the first container."}, {"from": "gpt", "value": "<point>[[548, 505], [466, 462], [362, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the blue cup towards the brown cup."}, {"from": "gpt", "value": "<point>[[440, 421], [424, 524], [490, 528]]</point>"}]}
{"id": 3620, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_735/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_737/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the item in the first container."}, {"from": "gpt", "value": "<point>[[544, 511], [503, 514], [465, 518]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cloth to the left."}, {"from": "gpt", "value": "<point>[[424, 609], [296, 643], [266, 761]]</point>"}]}
{"id": 3622, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_738/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_738/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cloth."}, {"from": "gpt", "value": "<point>[[529, 351], [518, 458], [573, 484]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[440, 521], [296, 565], [211, 698]]</point>"}]}
{"id": 3624, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_739/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_772/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to release the bottle into the pot."}, {"from": "gpt", "value": "<point>[[540, 373], [454, 447], [438, 547]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the pot."}, {"from": "gpt", "value": "<point>[[537, 495], [341, 628], [244, 724]]</point>"}]}
{"id": 3626, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_774/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_780/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[521, 425], [496, 543], [532, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[476, 561], [349, 602], [255, 680]]</point>"}]}
{"id": 3628, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_787/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_793/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the item in the first container."}, {"from": "gpt", "value": "<point>[[543, 492], [445, 502], [366, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue cup."}, {"from": "gpt", "value": "<point>[[485, 465], [355, 465], [271, 513]]</point>"}]}
{"id": 3630, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_802/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_812/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cloth."}, {"from": "gpt", "value": "<point>[[645, 532], [526, 598], [526, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[535, 539], [427, 499], [416, 406]]</point>"}]}
{"id": 3632, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_823/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_826/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to lower the pot."}, {"from": "gpt", "value": "<point>[[537, 502], [349, 609], [263, 724]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the object over the second container."}, {"from": "gpt", "value": "<point>[[524, 403], [457, 532], [385, 539]]</point>"}]}
{"id": 3634, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_867/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_873/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[537, 458], [407, 476], [363, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to sweep the cloth to the left."}, {"from": "gpt", "value": "<point>[[490, 510], [355, 550], [294, 646]]</point>"}]}
{"id": 3636, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_879/frame_0.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_881/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[532, 473], [338, 565], [285, 650]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[593, 458], [490, 447], [413, 488]]</point>"}]}
{"id": 3638, "image": ["rtx_frames_success_20/43_berkeley_autolab_ur5#episode_885/frame_15.png", "rtx_frames_success_20/43_berkeley_autolab_ur5#episode_891/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the cloth to the left side of the table."}, {"from": "gpt", "value": "<point>[[429, 602], [316, 628], [291, 728]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower towards the bottle."}, {"from": "gpt", "value": "<point>[[535, 495], [382, 639], [349, 750]]</point>"}]}
{"id": 3640, "image": ["rtx_frames_success_21/55_dobbe#episode_105/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_115/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[396, 769], [471, 659], [494, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[509, 835], [497, 711], [462, 598], [425, 506], [393, 445]]</point>"}]}
{"id": 3642, "image": ["rtx_frames_success_21/55_dobbe#episode_122/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_130/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[445, 803], [474, 682], [483, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp the drawer handle."}, {"from": "gpt", "value": "<point>[[373, 775], [425, 662], [442, 546], [451, 428]]</point>"}]}
{"id": 3644, "image": ["rtx_frames_success_21/55_dobbe#episode_140/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_142/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[393, 740], [413, 598], [436, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[451, 723], [460, 590], [471, 439], [474, 312]]</point>"}]}
{"id": 3646, "image": ["rtx_frames_success_21/55_dobbe#episode_1000/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1000/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[445, 772], [465, 665], [451, 555], [436, 451], [413, 306]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the door closed."}, {"from": "gpt", "value": "<point>[[408, 786], [416, 685], [468, 491]]</point>"}]}
{"id": 3648, "image": ["rtx_frames_success_21/55_dobbe#episode_1003/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1015/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[439, 775], [471, 676], [503, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[460, 786], [419, 702], [373, 633]]</point>"}]}
{"id": 3650, "image": ["rtx_frames_success_21/55_dobbe#episode_1018/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1023/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet doors."}, {"from": "gpt", "value": "<point>[[454, 711], [540, 552], [604, 408], [642, 277]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[434, 795], [465, 671], [497, 566], [540, 457], [575, 309]]</point>"}]}
{"id": 3652, "image": ["rtx_frames_success_21/55_dobbe#episode_1038/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1040/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[387, 780], [419, 676], [462, 572], [506, 491], [529, 431]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[457, 847], [460, 685], [465, 592], [465, 413]]</point>"}]}
{"id": 3654, "image": ["rtx_frames_success_21/55_dobbe#episode_1054/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1072/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[494, 838], [561, 754], [595, 671], [595, 584], [587, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[488, 743], [517, 627], [517, 520], [500, 422]]</point>"}]}
{"id": 3656, "image": ["rtx_frames_success_21/55_dobbe#episode_1078/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1085/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[448, 751], [457, 668], [439, 578], [448, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[474, 760], [503, 668], [520, 610], [526, 520]]</point>"}]}
{"id": 3658, "image": ["rtx_frames_success_21/55_dobbe#episode_1087/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1090/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[341, 757], [364, 679], [355, 569], [350, 431]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[439, 772], [480, 653], [514, 555], [549, 491]]</point>"}]}
{"id": 3660, "image": ["rtx_frames_success_21/55_dobbe#episode_1097/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1099/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the battery."}, {"from": "gpt", "value": "<point>[[457, 749], [506, 584], [575, 486], [691, 408]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[410, 775], [416, 688], [416, 618], [402, 532]]</point>"}]}
{"id": 3662, "image": ["rtx_frames_success_21/55_dobbe#episode_1106/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1106/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[451, 766], [468, 676], [462, 604], [448, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer closed."}, {"from": "gpt", "value": "<point>[[451, 751], [465, 618], [465, 500], [465, 422]]</point>"}]}
{"id": 3664, "image": ["rtx_frames_success_21/55_dobbe#episode_1116/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1118/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[367, 737], [390, 578], [399, 439], [445, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[434, 829], [442, 734], [445, 616], [445, 457]]</point>"}]}
{"id": 3666, "image": ["rtx_frames_success_21/55_dobbe#episode_1144/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1147/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[425, 769], [471, 691], [486, 636], [488, 590]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[454, 760], [514, 714], [587, 668]]</point>"}]}
{"id": 3668, "image": ["rtx_frames_success_21/55_dobbe#episode_1152/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1162/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[460, 740], [445, 639], [405, 566]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[474, 777], [431, 665], [399, 538], [347, 422], [277, 185]]</point>"}]}
{"id": 3670, "image": ["rtx_frames_success_21/55_dobbe#episode_1162/frame_15.png", "rtx_frames_success_21/55_dobbe#episode_1164/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the cabinet door closed."}, {"from": "gpt", "value": "<point>[[465, 731], [552, 616], [671, 503], [795, 408]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[448, 723], [486, 642], [538, 607], [587, 566]]</point>"}]}
{"id": 3672, "image": ["rtx_frames_success_21/55_dobbe#episode_1164/frame_15.png", "rtx_frames_success_21/55_dobbe#episode_1169/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[436, 728], [451, 569], [445, 408], [460, 266]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handles."}, {"from": "gpt", "value": "<point>[[460, 725], [451, 604], [451, 483], [451, 275]]</point>"}]}
{"id": 3674, "image": ["rtx_frames_success_21/55_dobbe#episode_1169/frame_15.png", "rtx_frames_success_21/55_dobbe#episode_1174/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer inward."}, {"from": "gpt", "value": "<point>[[436, 757], [434, 616], [445, 480], [451, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[318, 821], [379, 740], [436, 653], [486, 488]]</point>"}]}
{"id": 3676, "image": ["rtx_frames_success_21/55_dobbe#episode_1179/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1187/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the water bottle."}, {"from": "gpt", "value": "<point>[[480, 824], [491, 699], [488, 610], [474, 488], [468, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[436, 754], [474, 607], [483, 474], [483, 312]]</point>"}]}
{"id": 3678, "image": ["rtx_frames_success_21/55_dobbe#episode_1194/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1194/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[512, 873], [566, 723], [590, 607], [601, 361]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[523, 931], [451, 711], [396, 595], [355, 535]]</point>"}]}
{"id": 3680, "image": ["rtx_frames_success_21/55_dobbe#episode_1195/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1203/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[436, 777], [460, 653], [457, 546], [486, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[428, 809], [477, 679], [514, 555]]</point>"}]}
{"id": 3682, "image": ["rtx_frames_success_21/55_dobbe#episode_1208/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1216/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[465, 783], [486, 705], [514, 590], [488, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[460, 867], [488, 786], [500, 749], [503, 699]]</point>"}]}
{"id": 3684, "image": ["rtx_frames_success_21/55_dobbe#episode_1216/frame_15.png", "rtx_frames_success_21/55_dobbe#episode_1227/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[445, 769], [454, 639], [477, 540], [503, 361], [538, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the block."}, {"from": "gpt", "value": "<point>[[396, 786], [399, 662], [405, 494]]</point>"}]}
{"id": 3686, "image": ["rtx_frames_success_21/55_dobbe#episode_1235/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1236/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[451, 792], [431, 734], [382, 688]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[468, 754], [514, 659], [561, 618], [624, 592]]</point>"}]}
{"id": 3688, "image": ["rtx_frames_success_21/55_dobbe#episode_1237/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1243/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the oven."}, {"from": "gpt", "value": "<point>[[486, 844], [512, 728], [494, 601], [465, 494], [416, 373]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the refrigerator."}, {"from": "gpt", "value": "<point>[[434, 815], [445, 671], [422, 549]]</point>"}]}
{"id": 3690, "image": ["rtx_frames_success_21/55_dobbe#episode_1249/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1250/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer open."}, {"from": "gpt", "value": "<point>[[491, 861], [486, 725], [503, 590], [584, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[451, 812], [445, 685], [419, 575]]</point>"}]}
{"id": 3692, "image": ["rtx_frames_success_21/55_dobbe#episode_1251/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1254/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the refrigerator door."}, {"from": "gpt", "value": "<point>[[474, 769], [460, 633], [460, 509], [460, 370]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[448, 772], [477, 653], [546, 598], [656, 558], [777, 549]]</point>"}]}
{"id": 3694, "image": ["rtx_frames_success_21/55_dobbe#episode_1263/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1282/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[460, 734], [436, 665], [402, 595], [355, 509]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[497, 873], [477, 766], [457, 659], [436, 488]]</point>"}]}
{"id": 3696, "image": ["rtx_frames_success_21/55_dobbe#episode_1286/frame_15.png", "rtx_frames_success_21/55_dobbe#episode_1292/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer inward."}, {"from": "gpt", "value": "<point>[[431, 751], [488, 555], [532, 431], [607, 312]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[462, 775], [454, 691], [416, 613], [370, 566]]</point>"}]}
{"id": 3698, "image": ["rtx_frames_success_21/55_dobbe#episode_1302/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1311/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[436, 803], [454, 743], [477, 639]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[405, 786], [434, 630], [460, 555], [500, 384]]</point>"}]}
{"id": 3700, "image": ["rtx_frames_success_21/55_dobbe#episode_1316/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1325/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[370, 760], [382, 650], [390, 587], [428, 477]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[460, 746], [514, 590], [566, 448], [601, 113]]</point>"}]}
{"id": 3702, "image": ["rtx_frames_success_21/55_dobbe#episode_1329/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1343/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[454, 780], [488, 723], [552, 679], [618, 682]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[471, 795], [488, 630], [549, 494], [636, 338], [702, 176]]</point>"}]}
{"id": 3704, "image": ["rtx_frames_success_21/55_dobbe#episode_1347/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1357/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[376, 772], [439, 665], [488, 552], [491, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[439, 780], [399, 673], [318, 566], [228, 503], [113, 468]]</point>"}]}
{"id": 3706, "image": ["rtx_frames_success_21/55_dobbe#episode_1360/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1361/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[465, 786], [480, 610], [497, 512], [517, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[428, 780], [465, 671], [506, 601], [613, 503]]</point>"}]}
{"id": 3708, "image": ["rtx_frames_success_21/55_dobbe#episode_1365/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1370/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[416, 824], [460, 708], [512, 561], [503, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the device."}, {"from": "gpt", "value": "<point>[[442, 798], [445, 679], [445, 575], [445, 416]]</point>"}]}
{"id": 3710, "image": ["rtx_frames_success_21/55_dobbe#episode_1383/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1389/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[486, 780], [520, 656], [532, 514], [494, 303]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handles."}, {"from": "gpt", "value": "<point>[[457, 754], [465, 676], [465, 578], [503, 486]]</point>"}]}
{"id": 3712, "image": ["rtx_frames_success_21/55_dobbe#episode_1391/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1393/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet doors."}, {"from": "gpt", "value": "<point>[[425, 775], [405, 653], [376, 529], [350, 457]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handles."}, {"from": "gpt", "value": "<point>[[471, 798], [471, 694], [483, 595], [532, 462]]</point>"}]}
{"id": 3714, "image": ["rtx_frames_success_21/55_dobbe#episode_1395/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1399/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[402, 780], [425, 728], [454, 659], [497, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the green bottle."}, {"from": "gpt", "value": "<point>[[480, 801], [471, 711], [460, 653], [471, 575]]</point>"}]}
{"id": 3716, "image": ["rtx_frames_success_21/55_dobbe#episode_1403/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1415/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[512, 757], [509, 682], [517, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[460, 746], [451, 682], [428, 618], [367, 532]]</point>"}]}
{"id": 3718, "image": ["rtx_frames_success_21/55_dobbe#episode_1416/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1416/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[460, 769], [436, 685], [422, 650], [399, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[460, 777], [474, 604], [486, 526], [488, 382]]</point>"}]}
{"id": 3720, "image": ["rtx_frames_success_21/55_dobbe#episode_1421/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1429/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[509, 815], [448, 679], [419, 549], [376, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[457, 806], [428, 697], [393, 633], [303, 480]]</point>"}]}
{"id": 3722, "image": ["rtx_frames_success_21/55_dobbe#episode_1438/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1445/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[474, 734], [457, 659], [428, 590]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp both door handles."}, {"from": "gpt", "value": "<point>[[471, 740], [540, 630], [613, 497], [662, 367]]</point>"}]}
{"id": 3724, "image": ["rtx_frames_success_21/55_dobbe#episode_1450/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1455/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[514, 841], [535, 702], [549, 569], [584, 387]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to turn the faucet handle to turn on the water."}, {"from": "gpt", "value": "<point>[[465, 757], [494, 659], [520, 566], [509, 439]]</point>"}]}
{"id": 3726, "image": ["rtx_frames_success_21/55_dobbe#episode_1457/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1458/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[460, 749], [448, 642], [428, 543], [399, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[448, 723], [457, 633], [457, 555], [477, 471]]</point>"}]}
{"id": 3728, "image": ["rtx_frames_success_21/55_dobbe#episode_1458/frame_15.png", "rtx_frames_success_21/55_dobbe#episode_1468/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[454, 746], [471, 575], [480, 462], [491, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open cabinet door."}, {"from": "gpt", "value": "<point>[[422, 754], [503, 642], [592, 540], [627, 408]]</point>"}]}
{"id": 3730, "image": ["rtx_frames_success_21/55_dobbe#episode_1480/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1483/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the bag."}, {"from": "gpt", "value": "<point>[[428, 723], [454, 578], [474, 480], [517, 309]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the doorknob."}, {"from": "gpt", "value": "<point>[[497, 792], [540, 685], [517, 581], [483, 520]]</point>"}]}
{"id": 3732, "image": ["rtx_frames_success_21/55_dobbe#episode_1492/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1496/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the face mask."}, {"from": "gpt", "value": "<point>[[471, 792], [549, 598], [564, 402], [526, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[503, 824], [509, 671], [517, 595], [526, 480], [506, 367], [468, 266], [408, 139], [376, 69]]</point>"}]}
{"id": 3734, "image": ["rtx_frames_success_21/55_dobbe#episode_1498/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1499/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[448, 783], [549, 665], [665, 572], [795, 526], [945, 483]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the drawer open."}, {"from": "gpt", "value": "<point>[[483, 734], [486, 642], [477, 538], [462, 448]]</point>"}]}
{"id": 3736, "image": ["rtx_frames_success_21/55_dobbe#episode_1501/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1508/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[480, 740], [439, 601], [408, 509], [396, 344]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[448, 746], [488, 601], [514, 454], [552, 301]]</point>"}]}
{"id": 3738, "image": ["rtx_frames_success_21/55_dobbe#episode_1516/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1533/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the water bottle."}, {"from": "gpt", "value": "<point>[[451, 763], [448, 676], [416, 592], [384, 422]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open cabinet door."}, {"from": "gpt", "value": "<point>[[448, 772], [399, 647], [367, 566], [350, 486]]</point>"}]}
{"id": 3740, "image": ["rtx_frames_success_21/55_dobbe#episode_1533/frame_15.png", "rtx_frames_success_21/55_dobbe#episode_1534/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the door to close it."}, {"from": "gpt", "value": "<point>[[480, 702], [581, 572], [659, 474], [705, 419]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[486, 743], [494, 618], [503, 543], [491, 474]]</point>"}]}
{"id": 3742, "image": ["rtx_frames_success_21/55_dobbe#episode_1536/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1538/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[462, 884], [477, 780], [451, 694]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[465, 751], [494, 642], [520, 526], [520, 465]]</point>"}]}
{"id": 3744, "image": ["rtx_frames_success_21/55_dobbe#episode_1538/frame_15.png", "rtx_frames_success_21/55_dobbe#episode_1571/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[488, 610], [540, 549], [561, 439], [564, 344]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the doorknob."}, {"from": "gpt", "value": "<point>[[451, 780], [535, 697], [575, 601], [532, 483]]</point>"}]}
{"id": 3746, "image": ["rtx_frames_success_21/55_dobbe#episode_1580/frame_0.png", "rtx_frames_success_21/55_dobbe#episode_1586/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet doors."}, {"from": "gpt", "value": "<point>[[436, 769], [462, 647], [512, 514], [561, 410], [621, 329]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[399, 786], [422, 671], [442, 595], [454, 532]]</point>"}]}
{"id": 3748, "image": ["rtx_frames_success_22/55_dobbe#episode_18/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_167/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[413, 989], [374, 651], [407, 358], [448, 190]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[429, 991], [416, 818], [386, 617], [460, 506]]</point>"}]}
{"id": 3750, "image": ["rtx_frames_success_22/55_dobbe#episode_190/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_194/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[444, 988], [434, 861], [443, 721], [485, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[420, 995], [411, 869], [409, 702], [462, 552]]</point>"}]}
{"id": 3752, "image": ["rtx_frames_success_22/55_dobbe#episode_207/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_211/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[314, 996], [300, 880], [347, 651], [453, 453]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[436, 993], [406, 899], [420, 774], [481, 718]]</point>"}]}
{"id": 3754, "image": ["rtx_frames_success_22/55_dobbe#episode_230/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1595/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[497, 995], [340, 887], [233, 799], [259, 743]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[419, 995], [369, 846], [386, 688], [575, 494]]</point>"}]}
{"id": 3756, "image": ["rtx_frames_success_22/55_dobbe#episode_1599/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1602/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[406, 995], [383, 842], [352, 553], [395, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[374, 986], [376, 811], [447, 650], [529, 619]]</point>"}]}
{"id": 3758, "image": ["rtx_frames_success_22/55_dobbe#episode_1607/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1607/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[425, 991], [400, 764], [473, 638], [520, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the door to close it."}, {"from": "gpt", "value": "<point>[[416, 995], [371, 827], [366, 652], [445, 577]]</point>"}]}
{"id": 3760, "image": ["rtx_frames_success_22/55_dobbe#episode_1622/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1642/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the pen."}, {"from": "gpt", "value": "<point>[[422, 993], [395, 794], [432, 647], [503, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the green rectangular object."}, {"from": "gpt", "value": "<point>[[309, 982], [321, 780], [354, 594], [407, 460]]</point>"}]}
{"id": 3762, "image": ["rtx_frames_success_22/55_dobbe#episode_1652/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1668/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[413, 988], [402, 757], [309, 534], [240, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[414, 986], [471, 827], [534, 640], [519, 437]]</point>"}]}
{"id": 3764, "image": ["rtx_frames_success_22/55_dobbe#episode_1673/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1693/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[439, 995], [441, 721], [526, 538], [716, 325]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the doorknob."}, {"from": "gpt", "value": "<point>[[427, 988], [492, 896], [601, 765], [638, 580]]</point>"}]}
{"id": 3766, "image": ["rtx_frames_success_22/55_dobbe#episode_1698/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1708/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[446, 996], [478, 866], [591, 698], [679, 566]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[402, 991], [416, 815], [455, 697], [506, 621]]</point>"}]}
{"id": 3768, "image": ["rtx_frames_success_22/55_dobbe#episode_1708/frame_15.png", "rtx_frames_success_22/55_dobbe#episode_1730/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to press the light switch."}, {"from": "gpt", "value": "<point>[[427, 988], [451, 730], [464, 568], [508, 515]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[439, 993], [434, 831], [395, 633], [349, 400]]</point>"}]}
{"id": 3770, "image": ["rtx_frames_success_22/55_dobbe#episode_1733/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1733/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[446, 998], [397, 822], [379, 638], [494, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the drawer open."}, {"from": "gpt", "value": "<point>[[427, 995], [386, 829], [346, 642], [420, 580]]</point>"}]}
{"id": 3772, "image": ["rtx_frames_success_22/55_dobbe#episode_1738/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1740/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[425, 989], [416, 852], [414, 665], [432, 519]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[430, 991], [464, 675], [573, 457], [705, 277]]</point>"}]}
{"id": 3774, "image": ["rtx_frames_success_22/55_dobbe#episode_1744/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1771/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[414, 993], [432, 801], [490, 619], [556, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp both door handles."}, {"from": "gpt", "value": "<point>[[429, 996], [404, 899], [386, 713], [418, 626]]</point>"}]}
{"id": 3776, "image": ["rtx_frames_success_22/55_dobbe#episode_1771/frame_15.png", "rtx_frames_success_22/55_dobbe#episode_1776/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull both doors open simultaneously."}, {"from": "gpt", "value": "<point>[[347, 996], [335, 866], [362, 686], [407, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[381, 984], [362, 859], [365, 672], [425, 547]]</point>"}]}
{"id": 3778, "image": ["rtx_frames_success_22/55_dobbe#episode_1786/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1817/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cube."}, {"from": "gpt", "value": "<point>[[478, 991], [427, 854], [388, 683], [399, 504]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[483, 995], [474, 781], [496, 605], [536, 492]]</point>"}]}
{"id": 3780, "image": ["rtx_frames_success_22/55_dobbe#episode_1820/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1836/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[432, 995], [388, 829], [374, 612], [400, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[427, 993], [393, 810], [356, 547], [418, 370]]</point>"}]}
{"id": 3782, "image": ["rtx_frames_success_22/55_dobbe#episode_1851/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1865/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[404, 993], [402, 686], [339, 369], [233, 152]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[407, 989], [370, 767], [298, 503], [303, 316]]</point>"}]}
{"id": 3784, "image": ["rtx_frames_success_22/55_dobbe#episode_1869/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1875/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[377, 995], [399, 818], [451, 624], [501, 524]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[422, 996], [376, 852], [335, 705], [362, 545]]</point>"}]}
{"id": 3786, "image": ["rtx_frames_success_22/55_dobbe#episode_1882/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1883/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[384, 989], [369, 840], [388, 683], [432, 466]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[441, 989], [432, 884], [434, 697], [519, 513]]</point>"}]}
{"id": 3788, "image": ["rtx_frames_success_22/55_dobbe#episode_1910/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1915/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the air fryer."}, {"from": "gpt", "value": "<point>[[413, 1000], [402, 871], [411, 751], [450, 690]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handles."}, {"from": "gpt", "value": "<point>[[444, 993], [390, 875], [344, 721], [367, 635]]</point>"}]}
{"id": 3790, "image": ["rtx_frames_success_22/55_dobbe#episode_1917/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1920/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[316, 984], [319, 765], [377, 545], [451, 346]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grab the drawer handle."}, {"from": "gpt", "value": "<point>[[430, 996], [413, 847], [400, 626], [441, 453]]</point>"}]}
{"id": 3792, "image": ["rtx_frames_success_22/55_dobbe#episode_1925/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1929/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[429, 991], [370, 834], [312, 591], [317, 293]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[448, 995], [407, 817], [388, 631], [418, 547]]</point>"}]}
{"id": 3794, "image": ["rtx_frames_success_22/55_dobbe#episode_1953/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1957/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[362, 993], [390, 755], [406, 587], [453, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the knitted object."}, {"from": "gpt", "value": "<point>[[420, 991], [429, 758], [483, 556], [559, 312]]</point>"}]}
{"id": 3796, "image": ["rtx_frames_success_22/55_dobbe#episode_1963/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_1968/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator door handle."}, {"from": "gpt", "value": "<point>[[326, 991], [370, 772], [444, 619], [563, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[409, 989], [439, 806], [503, 635], [667, 444]]</point>"}]}
{"id": 3798, "image": ["rtx_frames_success_22/55_dobbe#episode_1989/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2003/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[469, 995], [471, 725], [504, 464], [573, 286]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[381, 981], [360, 727], [399, 422], [467, 245]]</point>"}]}
{"id": 3800, "image": ["rtx_frames_success_22/55_dobbe#episode_2007/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2011/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the kettle."}, {"from": "gpt", "value": "<point>[[425, 995], [432, 757], [437, 614], [483, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the glue bottle."}, {"from": "gpt", "value": "<point>[[422, 991], [414, 898], [411, 753], [457, 683]]</point>"}]}
{"id": 3802, "image": ["rtx_frames_success_22/55_dobbe#episode_2015/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2040/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the slipper."}, {"from": "gpt", "value": "<point>[[444, 996], [418, 841], [409, 591], [494, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[436, 988], [402, 873], [407, 730], [487, 626]]</point>"}]}
{"id": 3804, "image": ["rtx_frames_success_22/55_dobbe#episode_2041/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2041/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[432, 981], [411, 825], [420, 700], [460, 651]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the door closed."}, {"from": "gpt", "value": "<point>[[430, 988], [407, 771], [453, 577], [540, 492]]</point>"}]}
{"id": 3806, "image": ["rtx_frames_success_22/55_dobbe#episode_2043/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2047/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab both door handles."}, {"from": "gpt", "value": "<point>[[377, 986], [347, 864], [344, 743], [374, 690]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[455, 989], [422, 899], [400, 734], [437, 623]]</point>"}]}
{"id": 3808, "image": ["rtx_frames_success_22/55_dobbe#episode_2052/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2057/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the paper towel roll."}, {"from": "gpt", "value": "<point>[[422, 988], [404, 804], [369, 610], [374, 418]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handles of the trash cabinet doors."}, {"from": "gpt", "value": "<point>[[443, 984], [416, 864], [430, 725], [448, 688]]</point>"}]}
{"id": 3810, "image": ["rtx_frames_success_22/55_dobbe#episode_2062/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2068/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet doors."}, {"from": "gpt", "value": "<point>[[443, 989], [411, 871], [399, 691], [409, 400]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the refrigerator."}, {"from": "gpt", "value": "<point>[[430, 993], [441, 801], [476, 661], [510, 534]]</point>"}]}
{"id": 3812, "image": ["rtx_frames_success_22/55_dobbe#episode_2078/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2085/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[376, 996], [358, 885], [362, 714], [404, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[434, 996], [429, 757], [520, 667], [582, 623]]</point>"}]}
{"id": 3814, "image": ["rtx_frames_success_22/55_dobbe#episode_2087/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2089/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[413, 981], [392, 769], [402, 675], [443, 536]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[430, 995], [404, 882], [422, 635], [457, 536]]</point>"}]}
{"id": 3816, "image": ["rtx_frames_success_22/55_dobbe#episode_2093/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2123/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[418, 988], [395, 855], [388, 679], [446, 540]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[434, 988], [418, 850], [427, 603], [473, 430]]</point>"}]}
{"id": 3818, "image": ["rtx_frames_success_22/55_dobbe#episode_2124/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2145/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[430, 993], [414, 915], [443, 847], [483, 802]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[432, 995], [392, 945], [367, 854], [395, 767]]</point>"}]}
{"id": 3820, "image": ["rtx_frames_success_22/55_dobbe#episode_2149/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2163/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[432, 993], [430, 810], [448, 580], [508, 340]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[386, 991], [414, 820], [489, 600], [573, 499]]</point>"}]}
{"id": 3822, "image": ["rtx_frames_success_22/55_dobbe#episode_2173/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2182/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[429, 979], [416, 783], [432, 605], [459, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[425, 989], [402, 827], [416, 691], [453, 533]]</point>"}]}
{"id": 3824, "image": ["rtx_frames_success_22/55_dobbe#episode_2195/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2196/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[409, 996], [388, 792], [416, 578], [510, 356]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[430, 988], [374, 799], [310, 511], [358, 392]]</point>"}]}
{"id": 3826, "image": ["rtx_frames_success_22/55_dobbe#episode_2199/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2202/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the power strip."}, {"from": "gpt", "value": "<point>[[414, 995], [425, 873], [487, 725], [543, 679]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[409, 991], [393, 873], [413, 709], [464, 644]]</point>"}]}
{"id": 3828, "image": ["rtx_frames_success_22/55_dobbe#episode_2203/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2204/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pen."}, {"from": "gpt", "value": "<point>[[432, 991], [411, 852], [409, 607], [434, 487]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[381, 986], [370, 776], [400, 693], [462, 589]]</point>"}]}
{"id": 3830, "image": ["rtx_frames_success_22/55_dobbe#episode_2212/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2218/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[434, 988], [416, 864], [430, 635], [478, 513]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the two handles of the double door."}, {"from": "gpt", "value": "<point>[[446, 998], [406, 935], [381, 806], [400, 760]]</point>"}]}
{"id": 3832, "image": ["rtx_frames_success_22/55_dobbe#episode_2229/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2231/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[425, 986], [406, 864], [397, 750], [455, 550]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[432, 979], [432, 885], [471, 693], [600, 443]]</point>"}]}
{"id": 3834, "image": ["rtx_frames_success_22/55_dobbe#episode_2238/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2253/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[291, 993], [312, 841], [418, 533], [494, 443]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grab both handles of the cabinet door."}, {"from": "gpt", "value": "<point>[[386, 995], [381, 845], [439, 654], [474, 593]]</point>"}]}
{"id": 3836, "image": ["rtx_frames_success_22/55_dobbe#episode_2261/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2273/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[425, 991], [413, 864], [411, 601], [499, 384]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cube."}, {"from": "gpt", "value": "<point>[[469, 993], [457, 792], [487, 638], [557, 483]]</point>"}]}
{"id": 3838, "image": ["rtx_frames_success_22/55_dobbe#episode_2284/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2290/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open cabinet door."}, {"from": "gpt", "value": "<point>[[443, 988], [416, 854], [418, 642], [473, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[443, 995], [439, 887], [443, 718], [460, 653]]</point>"}]}
{"id": 3840, "image": ["rtx_frames_success_22/55_dobbe#episode_2291/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2321/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[303, 975], [303, 751], [344, 541], [407, 397]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[397, 991], [466, 848], [520, 765], [605, 700]]</point>"}]}
{"id": 3842, "image": ["rtx_frames_success_22/55_dobbe#episode_2332/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2335/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[342, 991], [302, 827], [309, 651], [328, 547]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[416, 966], [383, 801], [388, 674], [397, 536]]</point>"}]}
{"id": 3844, "image": ["rtx_frames_success_22/55_dobbe#episode_2341/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2348/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[427, 991], [414, 859], [444, 668], [511, 511]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[432, 989], [414, 921], [448, 788], [531, 624]]</point>"}]}
{"id": 3846, "image": ["rtx_frames_success_22/55_dobbe#episode_2359/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2363/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[439, 995], [416, 806], [437, 596], [483, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sunscreen bottle."}, {"from": "gpt", "value": "<point>[[446, 989], [400, 852], [353, 623], [374, 481]]</point>"}]}
{"id": 3848, "image": ["rtx_frames_success_22/55_dobbe#episode_2365/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2371/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cube."}, {"from": "gpt", "value": "<point>[[466, 991], [444, 875], [466, 693], [533, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp the drawer handles."}, {"from": "gpt", "value": "<point>[[298, 993], [273, 829], [321, 681], [402, 559]]</point>"}]}
{"id": 3850, "image": ["rtx_frames_success_22/55_dobbe#episode_2373/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2392/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[427, 993], [416, 794], [443, 593], [487, 335]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the plush toy."}, {"from": "gpt", "value": "<point>[[422, 989], [392, 810], [342, 617], [284, 432]]</point>"}]}
{"id": 3852, "image": ["rtx_frames_success_22/55_dobbe#episode_2393/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2395/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[376, 991], [351, 840], [351, 610], [450, 374]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the spray bottle."}, {"from": "gpt", "value": "<point>[[413, 993], [414, 772], [446, 619], [519, 481]]</point>"}]}
{"id": 3854, "image": ["rtx_frames_success_22/55_dobbe#episode_2406/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2410/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[425, 989], [404, 908], [420, 783], [485, 668]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[418, 991], [383, 686], [395, 326], [460, 153]]</point>"}]}
{"id": 3856, "image": ["rtx_frames_success_22/55_dobbe#episode_2414/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2418/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[321, 991], [325, 841], [404, 621], [513, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[488, 632], [493, 529], [546, 499]]</point>"}]}
{"id": 3858, "image": ["rtx_frames_success_22/55_dobbe#episode_2421/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2427/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handles."}, {"from": "gpt", "value": "<point>[[391, 673], [396, 515], [446, 476]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[488, 615], [150, 543], [36, 465]]</point>"}]}
{"id": 3860, "image": ["rtx_frames_success_22/55_dobbe#episode_2435/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2445/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[468, 657], [277, 476], [305, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[460, 645], [335, 554], [368, 507]]</point>"}]}
{"id": 3862, "image": ["rtx_frames_success_22/55_dobbe#episode_2454/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2458/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[488, 637], [521, 712], [598, 723]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[485, 648], [418, 596], [474, 576]]</point>"}]}
{"id": 3864, "image": ["rtx_frames_success_22/55_dobbe#episode_2461/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2462/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[476, 657], [438, 548], [488, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the plush toy."}, {"from": "gpt", "value": "<point>[[438, 834], [321, 515], [418, 399]]</point>"}]}
{"id": 3866, "image": ["rtx_frames_success_22/55_dobbe#episode_2467/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2469/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[416, 837], [374, 623], [407, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[407, 801], [316, 690], [330, 596]]</point>"}]}
{"id": 3868, "image": ["rtx_frames_success_22/55_dobbe#episode_2471/frame_0.png", "rtx_frames_success_22/55_dobbe#episode_2473/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[488, 681], [391, 454], [460, 324]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp the drawer handle."}, {"from": "gpt", "value": "<point>[[460, 698], [363, 512], [499, 515]]</point>"}]}
{"id": 3870, "image": ["rtx_frames_success_22/55_dobbe#episode_2479/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_29/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[399, 842], [521, 518], [493, 307]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[390, 855], [422, 746], [442, 590], [477, 399], [500, 240]]</point>"}]}
{"id": 3872, "image": ["rtx_frames_success_23/55_dobbe#episode_257/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_259/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[448, 786], [422, 647], [422, 561], [428, 384]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp the drawer handle."}, {"from": "gpt", "value": "<point>[[321, 792], [344, 714], [367, 656], [390, 575]]</point>"}]}
{"id": 3874, "image": ["rtx_frames_success_23/55_dobbe#episode_265/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_270/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the refrigerator."}, {"from": "gpt", "value": "<point>[[465, 743], [460, 604], [454, 509], [422, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[460, 725], [486, 607], [506, 477], [540, 413]]</point>"}]}
{"id": 3876, "image": ["rtx_frames_success_23/55_dobbe#episode_291/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_310/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[353, 803], [416, 621], [436, 514], [454, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[309, 754], [358, 642], [413, 517], [514, 376]]</point>"}]}
{"id": 3878, "image": ["rtx_frames_success_23/55_dobbe#episode_310/frame_15.png", "rtx_frames_success_23/55_dobbe#episode_331/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[329, 754], [402, 581], [457, 460], [500, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[465, 829], [486, 720], [523, 616], [578, 503], [630, 439]]</point>"}]}
{"id": 3880, "image": ["rtx_frames_success_23/55_dobbe#episode_2487/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2495/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the partially opened drawer."}, {"from": "gpt", "value": "<point>[[445, 775], [442, 656], [425, 581], [402, 512]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[462, 777], [448, 668], [445, 572], [399, 428]]</point>"}]}
{"id": 3882, "image": ["rtx_frames_success_23/55_dobbe#episode_2523/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2530/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to open the fire extinguisher cabinet."}, {"from": "gpt", "value": "<point>[[454, 775], [503, 665], [543, 584], [627, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[442, 668], [474, 532], [480, 471], [497, 408]]</point>"}]}
{"id": 3884, "image": ["rtx_frames_success_23/55_dobbe#episode_2535/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2549/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grasp the drawer handle."}, {"from": "gpt", "value": "<point>[[324, 780], [373, 682], [439, 607], [512, 540]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[462, 743], [506, 630], [546, 520], [708, 416]]</point>"}]}
{"id": 3886, "image": ["rtx_frames_success_23/55_dobbe#episode_2557/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2564/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator door handle."}, {"from": "gpt", "value": "<point>[[465, 769], [488, 665], [491, 566]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[431, 772], [439, 656], [460, 578], [486, 517]]</point>"}]}
{"id": 3888, "image": ["rtx_frames_success_23/55_dobbe#episode_2569/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2576/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[445, 749], [445, 575], [460, 480], [491, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[399, 783], [399, 633], [402, 540], [379, 462]]</point>"}]}
{"id": 3890, "image": ["rtx_frames_success_23/55_dobbe#episode_2598/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2611/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[460, 777], [460, 650], [445, 514], [431, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[468, 757], [439, 653], [393, 590], [355, 575]]</point>"}]}
{"id": 3892, "image": ["rtx_frames_success_23/55_dobbe#episode_2613/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2616/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator door handle."}, {"from": "gpt", "value": "<point>[[439, 766], [422, 668], [396, 549], [353, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[445, 775], [451, 645], [474, 561], [491, 480]]</point>"}]}
{"id": 3894, "image": ["rtx_frames_success_23/55_dobbe#episode_2617/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2618/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the doorknob."}, {"from": "gpt", "value": "<point>[[442, 777], [457, 682], [457, 572], [462, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the object."}, {"from": "gpt", "value": "<point>[[462, 769], [434, 659], [370, 569], [277, 529]]</point>"}]}
{"id": 3896, "image": ["rtx_frames_success_23/55_dobbe#episode_2621/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2630/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[442, 798], [457, 636], [480, 529], [506, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door."}, {"from": "gpt", "value": "<point>[[448, 760], [387, 584], [341, 422], [266, 176]]</point>"}]}
{"id": 3898, "image": ["rtx_frames_success_23/55_dobbe#episode_2647/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2673/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[462, 749], [439, 624], [410, 491], [416, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the avocado plush toy."}, {"from": "gpt", "value": "<point>[[431, 734], [448, 558], [454, 436]]</point>"}]}
{"id": 3900, "image": ["rtx_frames_success_23/55_dobbe#episode_2687/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2698/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[439, 760], [462, 691], [494, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[451, 746], [434, 636], [413, 538], [399, 428]]</point>"}]}
{"id": 3902, "image": ["rtx_frames_success_23/55_dobbe#episode_2715/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2725/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[460, 746], [480, 572], [494, 402], [532, 254], [564, 150]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[445, 740], [413, 618], [379, 509], [332, 408]]</point>"}]}
{"id": 3904, "image": ["rtx_frames_success_23/55_dobbe#episode_2737/frame_15.png", "rtx_frames_success_23/55_dobbe#episode_2743/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[471, 720], [506, 546], [535, 428], [587, 292]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[457, 766], [460, 676], [460, 587], [454, 526]]</point>"}]}
{"id": 3906, "image": ["rtx_frames_success_23/55_dobbe#episode_2747/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2751/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[454, 754], [462, 647], [477, 584], [483, 543]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[486, 824], [483, 705], [480, 592], [480, 497], [477, 338]]</point>"}]}
{"id": 3908, "image": ["rtx_frames_success_23/55_dobbe#episode_2779/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2798/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[436, 818], [451, 749], [465, 679], [477, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[474, 763], [474, 668], [457, 616], [422, 538]]</point>"}]}
{"id": 3910, "image": ["rtx_frames_success_23/55_dobbe#episode_2808/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2818/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the partially opened drawer."}, {"from": "gpt", "value": "<point>[[486, 855], [460, 740], [442, 645], [445, 503]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[390, 783], [425, 705], [471, 653], [561, 587]]</point>"}]}
{"id": 3912, "image": ["rtx_frames_success_23/55_dobbe#episode_2851/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2874/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[434, 749], [451, 645], [460, 540], [460, 445]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[445, 777], [462, 691], [491, 564], [517, 428]]</point>"}]}
{"id": 3914, "image": ["rtx_frames_success_23/55_dobbe#episode_2874/frame_15.png", "rtx_frames_success_23/55_dobbe#episode_2882/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer inwards."}, {"from": "gpt", "value": "<point>[[457, 749], [465, 621], [474, 512], [480, 431]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[408, 775], [393, 630], [390, 494]]</point>"}]}
{"id": 3916, "image": ["rtx_frames_success_23/55_dobbe#episode_2895/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2916/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[402, 809], [454, 705], [462, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[445, 786], [442, 679], [448, 616], [471, 561]]</point>"}]}
{"id": 3918, "image": ["rtx_frames_success_23/55_dobbe#episode_2925/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2926/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grasp both door handles."}, {"from": "gpt", "value": "<point>[[431, 775], [462, 723], [512, 679], [578, 645]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[442, 780], [439, 587], [422, 431], [396, 338], [358, 225]]</point>"}]}
{"id": 3920, "image": ["rtx_frames_success_23/55_dobbe#episode_2931/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2935/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[405, 763], [425, 636], [451, 555], [480, 439]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pen."}, {"from": "gpt", "value": "<point>[[448, 827], [486, 714], [500, 621], [497, 494]]</point>"}]}
{"id": 3922, "image": ["rtx_frames_success_23/55_dobbe#episode_2939/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2942/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[445, 795], [448, 639], [439, 509], [445, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[439, 821], [457, 746], [462, 691], [483, 633]]</point>"}]}
{"id": 3924, "image": ["rtx_frames_success_23/55_dobbe#episode_2944/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2954/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[434, 847], [436, 763], [451, 691], [457, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[370, 754], [370, 624], [370, 526], [379, 442]]</point>"}]}
{"id": 3926, "image": ["rtx_frames_success_23/55_dobbe#episode_2955/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2958/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet doors."}, {"from": "gpt", "value": "<point>[[448, 870], [494, 789], [549, 699], [595, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[442, 838], [491, 668], [535, 595], [601, 503]]</point>"}]}
{"id": 3928, "image": ["rtx_frames_success_23/55_dobbe#episode_2980/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2985/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the tennis ball."}, {"from": "gpt", "value": "<point>[[503, 812], [514, 676], [526, 584], [514, 460]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[451, 821], [439, 610], [445, 413], [535, 168]]</point>"}]}
{"id": 3930, "image": ["rtx_frames_success_23/55_dobbe#episode_2987/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_2999/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[448, 792], [396, 647], [341, 538], [275, 422]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[474, 777], [434, 575], [419, 462], [387, 335]]</point>"}]}
{"id": 3932, "image": ["rtx_frames_success_23/55_dobbe#episode_3015/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3034/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[344, 853], [318, 682], [324, 529], [367, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[454, 812], [532, 705], [636, 572]]</point>"}]}
{"id": 3934, "image": ["rtx_frames_success_23/55_dobbe#episode_3036/frame_15.png", "rtx_frames_success_23/55_dobbe#episode_3087/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[471, 754], [488, 552], [512, 390], [523, 249]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[448, 757], [457, 621], [448, 517]]</point>"}]}
{"id": 3936, "image": ["rtx_frames_success_23/55_dobbe#episode_3088/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3089/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the tap."}, {"from": "gpt", "value": "<point>[[408, 775], [457, 610], [494, 506], [514, 396], [517, 220]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the tube."}, {"from": "gpt", "value": "<point>[[355, 786], [367, 618], [390, 529], [436, 390]]</point>"}]}
{"id": 3938, "image": ["rtx_frames_success_23/55_dobbe#episode_3106/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3123/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[436, 824], [460, 717], [486, 633], [514, 509]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the green bottle."}, {"from": "gpt", "value": "<point>[[425, 754], [439, 653], [509, 540]]</point>"}]}
{"id": 3940, "image": ["rtx_frames_success_23/55_dobbe#episode_3124/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3126/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the sponge."}, {"from": "gpt", "value": "<point>[[471, 798], [480, 676], [474, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[396, 824], [425, 694], [471, 581], [497, 523]]</point>"}]}
{"id": 3942, "image": ["rtx_frames_success_23/55_dobbe#episode_3138/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3152/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[405, 841], [451, 673], [474, 604], [514, 460]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[436, 769], [457, 659], [488, 514]]</point>"}]}
{"id": 3944, "image": ["rtx_frames_success_23/55_dobbe#episode_3161/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3161/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[439, 838], [451, 723], [477, 613], [494, 520], [523, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[451, 777], [457, 656], [448, 555], [410, 425]]</point>"}]}
{"id": 3946, "image": ["rtx_frames_success_23/55_dobbe#episode_3165/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3169/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handles of the cabinet door."}, {"from": "gpt", "value": "<point>[[445, 795], [451, 656], [457, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[439, 887], [497, 803], [517, 754], [540, 711]]</point>"}]}
{"id": 3948, "image": ["rtx_frames_success_23/55_dobbe#episode_3198/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3199/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[445, 806], [422, 636], [367, 517], [303, 442], [251, 408]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[439, 786], [454, 725], [503, 598]]</point>"}]}
{"id": 3950, "image": ["rtx_frames_success_23/55_dobbe#episode_3204/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3232/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[442, 818], [506, 685], [578, 578], [645, 451], [668, 292]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[442, 795], [471, 647], [477, 540], [488, 457]]</point>"}]}
{"id": 3952, "image": ["rtx_frames_success_23/55_dobbe#episode_3236/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3242/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[335, 798], [399, 616], [451, 506], [509, 439]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[465, 809], [471, 685], [468, 595], [460, 514], [445, 431]]</point>"}]}
{"id": 3954, "image": ["rtx_frames_success_23/55_dobbe#episode_3257/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3259/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[465, 803], [538, 618], [561, 526], [555, 387]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the power strip."}, {"from": "gpt", "value": "<point>[[442, 847], [465, 720], [465, 624], [442, 558]]</point>"}]}
{"id": 3956, "image": ["rtx_frames_success_23/55_dobbe#episode_3265/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3278/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[451, 780], [428, 642], [410, 535], [379, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open cabinet door."}, {"from": "gpt", "value": "<point>[[431, 803], [460, 723], [491, 662], [535, 621]]</point>"}]}
{"id": 3958, "image": ["rtx_frames_success_23/55_dobbe#episode_3281/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3282/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the cabinet door."}, {"from": "gpt", "value": "<point>[[460, 806], [483, 671], [462, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the oven."}, {"from": "gpt", "value": "<point>[[512, 890], [523, 760], [517, 676], [512, 549], [488, 454], [457, 358]]</point>"}]}
{"id": 3960, "image": ["rtx_frames_success_23/55_dobbe#episode_3283/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3292/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet handles."}, {"from": "gpt", "value": "<point>[[442, 838], [442, 760], [416, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[434, 798], [451, 708], [471, 569], [503, 416]]</point>"}]}
{"id": 3962, "image": ["rtx_frames_success_23/55_dobbe#episode_3298/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3302/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[497, 801], [494, 685], [460, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open cabinet door."}, {"from": "gpt", "value": "<point>[[471, 757], [477, 662], [457, 549]]</point>"}]}
{"id": 3964, "image": ["rtx_frames_success_23/55_dobbe#episode_3308/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3310/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the faucet."}, {"from": "gpt", "value": "<point>[[436, 803], [488, 699], [538, 621], [575, 558], [590, 494], [601, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[445, 844], [460, 682], [474, 607], [529, 445]]</point>"}]}
{"id": 3966, "image": ["rtx_frames_success_23/55_dobbe#episode_3317/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3323/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[471, 806], [422, 673], [370, 590], [301, 494], [260, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[442, 803], [413, 613], [373, 468], [347, 275]]</point>"}]}
{"id": 3968, "image": ["rtx_frames_success_23/55_dobbe#episode_3332/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3334/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[410, 870], [393, 760], [355, 659], [324, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[431, 832], [457, 645], [494, 546], [532, 390], [598, 225]]</point>"}]}
{"id": 3970, "image": ["rtx_frames_success_23/55_dobbe#episode_3345/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3350/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the oven door."}, {"from": "gpt", "value": "<point>[[427, 994], [425, 857], [445, 632], [496, 491]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[422, 827], [431, 737], [445, 613]]</point>"}]}
{"id": 3972, "image": ["rtx_frames_success_23/55_dobbe#episode_3360/frame_0.png", "rtx_frames_success_23/55_dobbe#episode_3365/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[434, 996], [395, 846], [383, 664], [402, 492]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the clothing item."}, {"from": "gpt", "value": "<point>[[440, 983], [378, 808], [335, 607], [301, 389]]</point>"}]}
{"id": 3974, "image": ["rtx_frames_success_23/55_dobbe#episode_3368/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_35/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[444, 992], [367, 739], [329, 316], [327, 21]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[443, 878], [623, 360], [657, 316]]</point>"}]}
{"id": 3976, "image": ["rtx_frames_success_24/55_dobbe#episode_36/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_38/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handles of the cabinet doors."}, {"from": "gpt", "value": "<point>[[454, 867], [427, 385], [471, 296]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handles."}, {"from": "gpt", "value": "<point>[[457, 925], [468, 737], [501, 687]]</point>"}]}
{"id": 3978, "image": ["rtx_frames_success_24/55_dobbe#episode_342/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_345/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the cabinet door."}, {"from": "gpt", "value": "<point>[[446, 898], [573, 562], [634, 532]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the drawer handles."}, {"from": "gpt", "value": "<point>[[393, 917], [429, 693], [482, 651]]</point>"}]}
{"id": 3980, "image": ["rtx_frames_success_24/55_dobbe#episode_349/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_351/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab both door handles."}, {"from": "gpt", "value": "<point>[[418, 870], [388, 607], [421, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[446, 895], [554, 382], [607, 332]]</point>"}]}
{"id": 3982, "image": ["rtx_frames_success_24/55_dobbe#episode_361/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_368/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the glue stick."}, {"from": "gpt", "value": "<point>[[490, 889], [457, 637], [474, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[457, 837], [524, 357], [560, 335]]</point>"}]}
{"id": 3984, "image": ["rtx_frames_success_24/55_dobbe#episode_371/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_375/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the microwave."}, {"from": "gpt", "value": "<point>[[465, 873], [413, 546], [463, 504], [485, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[452, 939], [463, 726], [532, 640]]</point>"}]}
{"id": 3986, "image": ["rtx_frames_success_24/55_dobbe#episode_391/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_397/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the oven door."}, {"from": "gpt", "value": "<point>[[416, 884], [377, 463], [443, 391]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[438, 889], [393, 645], [429, 579]]</point>"}]}
{"id": 3988, "image": ["rtx_frames_success_24/55_dobbe#episode_402/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_403/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the sponge."}, {"from": "gpt", "value": "<point>[[438, 812], [704, 429], [848, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[460, 875], [404, 571], [460, 518]]</point>"}]}
{"id": 3990, "image": ["rtx_frames_success_24/55_dobbe#episode_406/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_407/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[438, 856], [396, 391], [438, 283]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[435, 853], [418, 648], [463, 582]]</point>"}]}
{"id": 3992, "image": ["rtx_frames_success_24/55_dobbe#episode_409/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_416/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[457, 814], [490, 175], [463, 86]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer handle."}, {"from": "gpt", "value": "<point>[[454, 762], [490, 496], [579, 427]]</point>"}]}
{"id": 3994, "image": ["rtx_frames_success_24/55_dobbe#episode_421/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_424/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[418, 909], [396, 787], [460, 765]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the sock."}, {"from": "gpt", "value": "<point>[[443, 850], [452, 665], [496, 615]]</point>"}]}
{"id": 3996, "image": ["rtx_frames_success_24/55_dobbe#episode_425/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3384/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[335, 842], [488, 452], [440, 352]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the oven door."}, {"from": "gpt", "value": "<point>[[463, 892], [410, 476], [476, 391]]</point>"}]}
{"id": 3998, "image": ["rtx_frames_success_24/55_dobbe#episode_3399/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3406/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[440, 856], [360, 515], [382, 482]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[454, 801], [407, 557], [471, 510]]</point>"}]}
{"id": 4000, "image": ["rtx_frames_success_24/55_dobbe#episode_3432/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3438/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[446, 892], [413, 643], [371, 593]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[446, 861], [402, 593], [427, 543]]</point>"}]}
{"id": 4002, "image": ["rtx_frames_success_24/55_dobbe#episode_3447/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3448/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[452, 834], [460, 161], [510, 111]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[490, 817], [363, 432], [382, 382]]</point>"}]}
{"id": 4004, "image": ["rtx_frames_success_24/55_dobbe#episode_3449/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3453/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[465, 892], [485, 202], [512, 144]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the closed double door."}, {"from": "gpt", "value": "<point>[[454, 906], [402, 551], [102, 562], [28, 510], [105, 482]]</point>"}]}
{"id": 4006, "image": ["rtx_frames_success_24/55_dobbe#episode_3463/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3472/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the refrigerator door."}, {"from": "gpt", "value": "<point>[[468, 917], [476, 554], [410, 510]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[454, 825], [537, 587], [607, 560]]</point>"}]}
{"id": 4008, "image": ["rtx_frames_success_24/55_dobbe#episode_3475/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3478/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[474, 945], [418, 762], [388, 737]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the air fryer."}, {"from": "gpt", "value": "<point>[[452, 837], [479, 424], [535, 377]]</point>"}]}
{"id": 4010, "image": ["rtx_frames_success_24/55_dobbe#episode_3496/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3496/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[410, 820], [460, 249], [490, 199]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to press the light switch."}, {"from": "gpt", "value": "<point>[[463, 884], [421, 507], [449, 452]]</point>"}]}
{"id": 4012, "image": ["rtx_frames_success_24/55_dobbe#episode_3518/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3519/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cube."}, {"from": "gpt", "value": "<point>[[438, 798], [307, 521], [321, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[438, 853], [385, 443], [407, 391]]</point>"}]}
{"id": 4014, "image": ["rtx_frames_success_24/55_dobbe#episode_3534/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3552/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab both door handles."}, {"from": "gpt", "value": "<point>[[476, 864], [424, 388], [457, 263]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the refrigerator door."}, {"from": "gpt", "value": "<point>[[499, 925], [512, 615], [548, 560]]</point>"}]}
{"id": 4016, "image": ["rtx_frames_success_24/55_dobbe#episode_3573/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3589/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[471, 820], [568, 449], [537, 404]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[435, 878], [357, 551], [377, 488]]</point>"}]}
{"id": 4018, "image": ["rtx_frames_success_24/55_dobbe#episode_3596/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3606/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[413, 864], [438, 554], [471, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[438, 892], [363, 695], [380, 637]]</point>"}]}
{"id": 4020, "image": ["rtx_frames_success_24/55_dobbe#episode_3641/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3642/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[438, 911], [399, 654], [443, 604]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[471, 895], [355, 512], [380, 454]]</point>"}]}
{"id": 4022, "image": ["rtx_frames_success_24/55_dobbe#episode_3649/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3654/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[454, 892], [457, 706], [493, 651]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[488, 881], [349, 512], [316, 479]]</point>"}]}
{"id": 4024, "image": ["rtx_frames_success_24/55_dobbe#episode_3659/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3665/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the sliding door."}, {"from": "gpt", "value": "<point>[[493, 820], [413, 349], [446, 313]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[496, 925], [366, 463], [368, 407]]</point>"}]}
{"id": 4026, "image": ["rtx_frames_success_24/55_dobbe#episode_3668/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3672/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[454, 806], [471, 524], [493, 493]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[463, 848], [302, 332], [335, 269]]</point>"}]}
{"id": 4028, "image": ["rtx_frames_success_24/55_dobbe#episode_3702/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3710/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[446, 820], [427, 313], [377, 252]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[468, 817], [263, 299], [166, 249]]</point>"}]}
{"id": 4030, "image": ["rtx_frames_success_24/55_dobbe#episode_3711/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3712/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[515, 900], [616, 314], [605, 292]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[438, 848], [418, 380], [457, 313]]</point>"}]}
{"id": 4032, "image": ["rtx_frames_success_24/55_dobbe#episode_3714/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3717/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[474, 856], [421, 493], [460, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[471, 798], [601, 571], [825, 540], [875, 568], [842, 598]]</point>"}]}
{"id": 4034, "image": ["rtx_frames_success_24/55_dobbe#episode_3719/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3729/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[427, 831], [446, 402], [479, 352]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the red cup."}, {"from": "gpt", "value": "<point>[[454, 859], [443, 632], [499, 565]]</point>"}]}
{"id": 4036, "image": ["rtx_frames_success_24/55_dobbe#episode_3738/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3744/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer handle."}, {"from": "gpt", "value": "<point>[[452, 839], [471, 532], [524, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[501, 789], [457, 620], [471, 562]]</point>"}]}
{"id": 4038, "image": ["rtx_frames_success_24/55_dobbe#episode_3744/frame_15.png", "rtx_frames_success_24/55_dobbe#episode_3746/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to press the light switch."}, {"from": "gpt", "value": "<point>[[482, 864], [463, 609], [490, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the oven door."}, {"from": "gpt", "value": "<point>[[413, 850], [385, 540], [319, 468]]</point>"}]}
{"id": 4040, "image": ["rtx_frames_success_24/55_dobbe#episode_3751/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3752/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[452, 850], [404, 584], [421, 499]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the stuffed hedgehog."}, {"from": "gpt", "value": "<point>[[457, 881], [496, 546], [598, 418]]</point>"}]}
{"id": 4042, "image": ["rtx_frames_success_24/55_dobbe#episode_3754/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3767/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[435, 878], [510, 410], [476, 355]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[443, 886], [501, 501], [543, 454]]</point>"}]}
{"id": 4044, "image": ["rtx_frames_success_24/55_dobbe#episode_3776/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3777/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handles."}, {"from": "gpt", "value": "<point>[[382, 889], [598, 856], [651, 814]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the microwave."}, {"from": "gpt", "value": "<point>[[449, 878], [787, 784], [784, 715]]</point>"}]}
{"id": 4046, "image": ["rtx_frames_success_24/55_dobbe#episode_3786/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3794/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the air fryer drawer."}, {"from": "gpt", "value": "<point>[[404, 861], [357, 443], [399, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[385, 906], [402, 391], [468, 291]]</point>"}]}
{"id": 4048, "image": ["rtx_frames_success_24/55_dobbe#episode_3801/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3809/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[457, 903], [449, 720], [416, 687]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the white object."}, {"from": "gpt", "value": "<point>[[443, 903], [363, 726], [391, 620]]</point>"}]}
{"id": 4050, "image": ["rtx_frames_success_24/55_dobbe#episode_3812/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3813/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the paper bag."}, {"from": "gpt", "value": "<point>[[449, 842], [349, 435], [457, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the object."}, {"from": "gpt", "value": "<point>[[468, 906], [529, 898], [512, 623]]</point>"}]}
{"id": 4052, "image": ["rtx_frames_success_24/55_dobbe#episode_3831/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3835/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[338, 895], [360, 584], [410, 540]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[468, 900], [407, 391], [446, 346]]</point>"}]}
{"id": 4054, "image": ["rtx_frames_success_24/55_dobbe#episode_3841/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3842/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the handle."}, {"from": "gpt", "value": "<point>[[465, 964], [410, 806], [363, 789]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[443, 842], [338, 263], [396, 183]]</point>"}]}
{"id": 4056, "image": ["rtx_frames_success_24/55_dobbe#episode_3851/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3856/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the eye drop bottle."}, {"from": "gpt", "value": "<point>[[452, 873], [421, 609], [452, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven door handle."}, {"from": "gpt", "value": "<point>[[413, 889], [429, 651], [485, 571]]</point>"}]}
{"id": 4058, "image": ["rtx_frames_success_24/55_dobbe#episode_3865/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3887/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[463, 958], [385, 817], [399, 759]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handles."}, {"from": "gpt", "value": "<point>[[446, 856], [438, 471], [493, 399]]</point>"}]}
{"id": 4060, "image": ["rtx_frames_success_24/55_dobbe#episode_3890/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3898/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[429, 953], [321, 759], [371, 637], [543, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the faucet handle."}, {"from": "gpt", "value": "<point>[[443, 889], [522, 461], [544, 431]]</point>"}]}
{"id": 4062, "image": ["rtx_frames_success_24/55_dobbe#episode_3911/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3918/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[476, 906], [343, 540], [374, 438]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the tennis ball."}, {"from": "gpt", "value": "<point>[[452, 870], [349, 620], [393, 546]]</point>"}]}
{"id": 4064, "image": ["rtx_frames_success_24/55_dobbe#episode_3923/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3925/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[440, 892], [526, 496], [623, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[496, 895], [479, 457], [529, 327]]</point>"}]}
{"id": 4066, "image": ["rtx_frames_success_24/55_dobbe#episode_3926/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3929/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[449, 936], [421, 792], [471, 723]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[457, 853], [310, 357], [346, 238]]</point>"}]}
{"id": 4068, "image": ["rtx_frames_success_24/55_dobbe#episode_3940/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3945/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[366, 842], [388, 501], [416, 360]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[463, 889], [551, 388], [596, 302]]</point>"}]}
{"id": 4070, "image": ["rtx_frames_success_24/55_dobbe#episode_3952/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3955/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[452, 875], [449, 706], [490, 665]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[485, 900], [327, 562], [355, 490]]</point>"}]}
{"id": 4072, "image": ["rtx_frames_success_24/55_dobbe#episode_3960/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3960/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[471, 889], [366, 687], [346, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the drawer open."}, {"from": "gpt", "value": "<point>[[421, 920], [524, 693], [565, 535]]</point>"}]}
{"id": 4074, "image": ["rtx_frames_success_24/55_dobbe#episode_3961/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3968/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[499, 911], [474, 529], [526, 418]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the refrigerator door."}, {"from": "gpt", "value": "<point>[[463, 881], [410, 546], [488, 410]]</point>"}]}
{"id": 4076, "image": ["rtx_frames_success_24/55_dobbe#episode_3971/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3974/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[277, 920], [393, 609], [471, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the double door."}, {"from": "gpt", "value": "<point>[[410, 886], [637, 510], [759, 512]]</point>"}]}
{"id": 4078, "image": ["rtx_frames_success_24/55_dobbe#episode_3994/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_3998/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[427, 850], [583, 148], [664, 94]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle of the cabinet door."}, {"from": "gpt", "value": "<point>[[446, 873], [421, 560], [482, 482]]</point>"}]}
{"id": 4080, "image": ["rtx_frames_success_24/55_dobbe#episode_4003/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4013/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handles."}, {"from": "gpt", "value": "<point>[[380, 917], [313, 640], [352, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[471, 834], [615, 504], [756, 474]]</point>"}]}
{"id": 4082, "image": ["rtx_frames_success_24/55_dobbe#episode_4014/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4018/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[440, 867], [413, 598], [452, 543]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[465, 789], [573, 355], [673, 283]]</point>"}]}
{"id": 4084, "image": ["rtx_frames_success_24/55_dobbe#episode_4019/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4019/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[465, 859], [355, 568], [343, 485]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to press the light switch."}, {"from": "gpt", "value": "<point>[[454, 828], [476, 548], [399, 510]]</point>"}]}
{"id": 4086, "image": ["rtx_frames_success_24/55_dobbe#episode_4022/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4034/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[454, 875], [476, 690], [604, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the dishwasher door."}, {"from": "gpt", "value": "<point>[[446, 864], [432, 471], [540, 330]]</point>"}]}
{"id": 4088, "image": ["rtx_frames_success_24/55_dobbe#episode_4037/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4040/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[438, 845], [427, 640], [493, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[512, 906], [485, 593], [507, 535]]</point>"}]}
{"id": 4090, "image": ["rtx_frames_success_24/55_dobbe#episode_4046/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4053/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[449, 820], [429, 413], [471, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[452, 884], [388, 479], [418, 363]]</point>"}]}
{"id": 4092, "image": ["rtx_frames_success_24/55_dobbe#episode_4056/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4068/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the tube."}, {"from": "gpt", "value": "<point>[[343, 898], [371, 584], [446, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[452, 834], [416, 607], [446, 476]]</point>"}]}
{"id": 4094, "image": ["rtx_frames_success_24/55_dobbe#episode_4075/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4085/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the closet door."}, {"from": "gpt", "value": "<point>[[440, 903], [402, 742], [435, 693]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[413, 834], [457, 576], [521, 521]]</point>"}]}
{"id": 4096, "image": ["rtx_frames_success_24/55_dobbe#episode_4095/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4095/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[526, 781], [490, 468], [446, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to press the light switch."}, {"from": "gpt", "value": "<point>[[476, 814], [501, 632], [471, 590], [476, 645]]</point>"}]}
{"id": 4098, "image": ["rtx_frames_success_24/55_dobbe#episode_4102/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4110/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handles."}, {"from": "gpt", "value": "<point>[[440, 853], [371, 670], [463, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[352, 817], [349, 446], [399, 305]]</point>"}]}
{"id": 4100, "image": ["rtx_frames_success_24/55_dobbe#episode_4121/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4123/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[452, 848], [343, 601], [360, 554]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the tape roll."}, {"from": "gpt", "value": "<point>[[332, 828], [296, 565], [349, 507]]</point>"}]}
{"id": 4102, "image": ["rtx_frames_success_24/55_dobbe#episode_4137/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4142/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handles."}, {"from": "gpt", "value": "<point>[[452, 895], [551, 584], [618, 532]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[501, 898], [374, 568], [407, 432]]</point>"}]}
{"id": 4104, "image": ["rtx_frames_success_24/55_dobbe#episode_4143/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4144/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[460, 751], [501, 291], [432, 197]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[424, 809], [413, 407], [452, 357]]</point>"}]}
{"id": 4106, "image": ["rtx_frames_success_24/55_dobbe#episode_4144/frame_15.png", "rtx_frames_success_24/55_dobbe#episode_4151/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to press the light switch."}, {"from": "gpt", "value": "<point>[[471, 864], [407, 493], [438, 402]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer handle."}, {"from": "gpt", "value": "<point>[[482, 861], [604, 440], [643, 368]]</point>"}]}
{"id": 4108, "image": ["rtx_frames_success_24/55_dobbe#episode_4152/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4152/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[440, 842], [717, 299], [698, 244]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the door to close it."}, {"from": "gpt", "value": "<point>[[446, 900], [399, 659], [438, 562]]</point>"}]}
{"id": 4110, "image": ["rtx_frames_success_24/55_dobbe#episode_4153/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4154/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the faucet."}, {"from": "gpt", "value": "<point>[[454, 867], [501, 349], [460, 269]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[454, 884], [510, 601], [565, 557]]</point>"}]}
{"id": 4112, "image": ["rtx_frames_success_24/55_dobbe#episode_4156/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4156/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the microwave."}, {"from": "gpt", "value": "<point>[[429, 864], [598, 543], [646, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to press the stop button."}, {"from": "gpt", "value": "<point>[[457, 839], [435, 695], [518, 601]]</point>"}]}
{"id": 4114, "image": ["rtx_frames_success_24/55_dobbe#episode_4161/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4161/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[449, 814], [382, 460], [440, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to flip the light switch up."}, {"from": "gpt", "value": "<point>[[443, 886], [343, 753], [366, 654]]</point>"}]}
{"id": 4116, "image": ["rtx_frames_success_24/55_dobbe#episode_4166/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4168/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[449, 867], [485, 573], [524, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[440, 853], [402, 543], [457, 460]]</point>"}]}
{"id": 4118, "image": ["rtx_frames_success_24/55_dobbe#episode_4171/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4178/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[465, 848], [490, 407], [557, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the yellow cube."}, {"from": "gpt", "value": "<point>[[482, 817], [399, 573], [432, 482]]</point>"}]}
{"id": 4120, "image": ["rtx_frames_success_24/55_dobbe#episode_4179/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4182/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[446, 881], [449, 695], [507, 643]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open cabinet door."}, {"from": "gpt", "value": "<point>[[440, 825], [482, 565], [438, 493]]</point>"}]}
{"id": 4122, "image": ["rtx_frames_success_24/55_dobbe#episode_4197/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4200/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the sunscreen bottle."}, {"from": "gpt", "value": "<point>[[443, 881], [440, 537], [385, 452]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[424, 850], [435, 543], [377, 382]]</point>"}]}
{"id": 4124, "image": ["rtx_frames_success_24/55_dobbe#episode_4200/frame_15.png", "rtx_frames_success_24/55_dobbe#episode_4206/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the cabinet door closed."}, {"from": "gpt", "value": "<point>[[457, 837], [490, 587], [346, 515]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[452, 837], [388, 493], [416, 416]]</point>"}]}
{"id": 4126, "image": ["rtx_frames_success_24/55_dobbe#episode_4219/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4226/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[493, 895], [715, 501], [704, 368]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle of the cabinet door."}, {"from": "gpt", "value": "<point>[[346, 842], [427, 560], [382, 440]]</point>"}]}
{"id": 4128, "image": ["rtx_frames_success_24/55_dobbe#episode_4227/frame_0.png", "rtx_frames_success_24/55_dobbe#episode_4238/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[377, 842], [529, 615], [499, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the dishwasher."}, {"from": "gpt", "value": "<point>[[454, 825], [452, 429], [521, 355]]</point>"}]}
{"id": 4130, "image": ["rtx_frames_success_24/55_dobbe#episode_4249/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_49/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[438, 895], [535, 690], [440, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[468, 829], [465, 725], [451, 673], [445, 555]]</point>"}]}
{"id": 4132, "image": ["rtx_frames_success_25/55_dobbe#episode_427/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_433/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the refrigerator."}, {"from": "gpt", "value": "<point>[[441, 958], [445, 845], [445, 735], [450, 580]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[433, 954], [433, 786], [429, 651], [412, 340]]</point>"}]}
{"id": 4134, "image": ["rtx_frames_success_25/55_dobbe#episode_438/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_447/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the trash bins."}, {"from": "gpt", "value": "<point>[[437, 975], [466, 878], [525, 786], [592, 655]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[429, 987], [454, 891], [479, 769], [450, 618], [454, 954]]</point>"}]}
{"id": 4136, "image": ["rtx_frames_success_25/55_dobbe#episode_454/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_456/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[441, 962], [429, 807], [395, 592], [261, 231]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handle."}, {"from": "gpt", "value": "<point>[[448, 815], [503, 734], [488, 598]]</point>"}]}
{"id": 4138, "image": ["rtx_frames_success_25/55_dobbe#episode_482/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_483/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handles."}, {"from": "gpt", "value": "<point>[[471, 775], [474, 601], [410, 431], [338, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[448, 832], [460, 708], [512, 636], [581, 592], [673, 569]]</point>"}]}
{"id": 4140, "image": ["rtx_frames_success_25/55_dobbe#episode_485/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_493/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab both door handles."}, {"from": "gpt", "value": "<point>[[431, 832], [494, 783], [552, 702]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[480, 737], [460, 592], [445, 500], [428, 376]]</point>"}]}
{"id": 4142, "image": ["rtx_frames_success_25/55_dobbe#episode_494/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_495/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[442, 740], [442, 662], [410, 584], [353, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[462, 887], [488, 835], [491, 757], [448, 697]]</point>"}]}
{"id": 4144, "image": ["rtx_frames_success_25/55_dobbe#episode_496/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_505/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[436, 818], [494, 691], [526, 581], [566, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the doorknob."}, {"from": "gpt", "value": "<point>[[462, 815], [491, 711], [509, 561]]</point>"}]}
{"id": 4146, "image": ["rtx_frames_success_25/55_dobbe#episode_507/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_511/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[436, 786], [486, 601], [503, 434], [497, 217]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[387, 809], [436, 685], [488, 601]]</point>"}]}
{"id": 4148, "image": ["rtx_frames_success_25/55_dobbe#episode_4277/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4281/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the tissue box."}, {"from": "gpt", "value": "<point>[[420, 958], [416, 840], [424, 723], [437, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the oven door."}, {"from": "gpt", "value": "<point>[[429, 941], [437, 811], [445, 685], [454, 529]]</point>"}]}
{"id": 4150, "image": ["rtx_frames_success_25/55_dobbe#episode_4300/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4310/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the orange."}, {"from": "gpt", "value": "<point>[[462, 941], [450, 832], [445, 685], [441, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cube."}, {"from": "gpt", "value": "<point>[[420, 962], [412, 832], [412, 676], [387, 517]]</point>"}]}
{"id": 4152, "image": ["rtx_frames_success_25/55_dobbe#episode_4315/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4317/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[424, 979], [445, 866], [471, 744], [500, 559]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[492, 950], [492, 790], [483, 643], [462, 424]]</point>"}]}
{"id": 4154, "image": ["rtx_frames_success_25/55_dobbe#episode_4334/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4337/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[475, 950], [521, 807], [546, 647], [571, 437]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[479, 954], [487, 756], [500, 584], [529, 336]]</point>"}]}
{"id": 4156, "image": ["rtx_frames_success_25/55_dobbe#episode_4341/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4347/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[496, 941], [462, 761], [429, 567], [399, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[429, 950], [462, 794], [517, 630], [609, 433]]</point>"}]}
{"id": 4158, "image": ["rtx_frames_success_25/55_dobbe#episode_4348/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4353/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the partially opened drawer."}, {"from": "gpt", "value": "<point>[[433, 954], [429, 752], [441, 588], [571, 349], [445, 908]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[391, 937], [416, 807], [441, 693], [462, 521]]</point>"}]}
{"id": 4160, "image": ["rtx_frames_success_25/55_dobbe#episode_4365/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4377/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab the handles of the door."}, {"from": "gpt", "value": "<point>[[445, 962], [454, 887], [479, 836], [487, 786]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[382, 979], [382, 895], [378, 807], [378, 601]]</point>"}]}
{"id": 4162, "image": ["rtx_frames_success_25/55_dobbe#episode_4394/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4406/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[424, 979], [475, 903], [504, 794], [546, 702]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the bottle."}, {"from": "gpt", "value": "<point>[[433, 929], [420, 718], [378, 517], [319, 286]]</point>"}]}
{"id": 4164, "image": ["rtx_frames_success_25/55_dobbe#episode_4409/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4413/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the rubik's cube."}, {"from": "gpt", "value": "<point>[[454, 971], [458, 845], [487, 689], [487, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[420, 971], [429, 832], [437, 693], [441, 576]]</point>"}]}
{"id": 4166, "image": ["rtx_frames_success_25/55_dobbe#episode_4416/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4418/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[471, 971], [471, 882], [471, 794], [496, 651]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[420, 971], [462, 878], [504, 773], [542, 643]]</point>"}]}
{"id": 4168, "image": ["rtx_frames_success_25/55_dobbe#episode_4421/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4432/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[403, 962], [416, 807], [420, 647], [445, 403]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[445, 954], [437, 748], [429, 563], [424, 332]]</point>"}]}
{"id": 4170, "image": ["rtx_frames_success_25/55_dobbe#episode_4436/frame_15.png", "rtx_frames_success_25/55_dobbe#episode_4440/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to press the light switch."}, {"from": "gpt", "value": "<point>[[487, 664], [529, 496], [550, 357], [563, 206]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the microwave oven."}, {"from": "gpt", "value": "<point>[[450, 962], [521, 811], [559, 681], [639, 538]]</point>"}]}
{"id": 4172, "image": ["rtx_frames_success_25/55_dobbe#episode_4442/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4446/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[412, 966], [370, 861], [336, 782], [345, 660]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handles."}, {"from": "gpt", "value": "<point>[[424, 941], [466, 761], [538, 592], [655, 298]]</point>"}]}
{"id": 4174, "image": ["rtx_frames_success_25/55_dobbe#episode_4448/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4457/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[403, 937], [416, 786], [416, 639], [416, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp the drawer handle."}, {"from": "gpt", "value": "<point>[[403, 975], [454, 866], [479, 786], [546, 706]]</point>"}]}
{"id": 4176, "image": ["rtx_frames_success_25/55_dobbe#episode_4467/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4476/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator door handle."}, {"from": "gpt", "value": "<point>[[454, 945], [534, 786], [546, 676], [592, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[420, 954], [395, 794], [370, 647], [328, 437]]</point>"}]}
{"id": 4178, "image": ["rtx_frames_success_25/55_dobbe#episode_4485/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4489/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[429, 958], [466, 828], [525, 735], [605, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[429, 971], [441, 811], [445, 693], [462, 567]]</point>"}]}
{"id": 4180, "image": ["rtx_frames_success_25/55_dobbe#episode_4490/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4499/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the sunscreen bottle."}, {"from": "gpt", "value": "<point>[[462, 958], [496, 815], [534, 676], [555, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[349, 975], [412, 845], [479, 685], [529, 458]]</point>"}]}
{"id": 4182, "image": ["rtx_frames_success_25/55_dobbe#episode_4501/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4529/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[445, 937], [454, 773], [466, 571], [471, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[445, 945], [462, 794], [412, 634], [412, 513]]</point>"}]}
{"id": 4184, "image": ["rtx_frames_success_25/55_dobbe#episode_4530/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4545/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[437, 987], [479, 845], [496, 702], [504, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[445, 979], [445, 870], [458, 765], [513, 647]]</point>"}]}
{"id": 4186, "image": ["rtx_frames_success_25/55_dobbe#episode_4547/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4550/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the rubik's cube."}, {"from": "gpt", "value": "<point>[[468, 783], [514, 702], [535, 590], [543, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handles."}, {"from": "gpt", "value": "<point>[[379, 873], [367, 780], [408, 662], [474, 535]]</point>"}]}
{"id": 4188, "image": ["rtx_frames_success_25/55_dobbe#episode_4556/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4559/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[434, 908], [436, 803], [379, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[428, 801], [402, 639], [434, 497], [494, 312]]</point>"}]}
{"id": 4190, "image": ["rtx_frames_success_25/55_dobbe#episode_4560/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4562/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the electric kettle."}, {"from": "gpt", "value": "<point>[[454, 861], [483, 751], [509, 656], [506, 575]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[422, 806], [514, 659], [566, 552], [561, 301]]</point>"}]}
{"id": 4192, "image": ["rtx_frames_success_25/55_dobbe#episode_4562/frame_15.png", "rtx_frames_success_25/55_dobbe#episode_4572/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to flip the light switch down."}, {"from": "gpt", "value": "<point>[[451, 792], [500, 564], [468, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[465, 827], [436, 737], [436, 642], [509, 434]]</point>"}]}
{"id": 4194, "image": ["rtx_frames_success_25/55_dobbe#episode_4576/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4580/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handles."}, {"from": "gpt", "value": "<point>[[402, 905], [465, 806], [451, 711]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[439, 795], [402, 685], [405, 535], [471, 402]]</point>"}]}
{"id": 4196, "image": ["rtx_frames_success_25/55_dobbe#episode_4583/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4589/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[422, 809], [428, 601], [474, 405]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the microwave oven."}, {"from": "gpt", "value": "<point>[[419, 841], [552, 647], [682, 384]]</point>"}]}
{"id": 4198, "image": ["rtx_frames_success_25/55_dobbe#episode_4605/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4610/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[436, 798], [431, 702], [451, 607], [512, 462]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[309, 798], [295, 682], [318, 578], [416, 520]]</point>"}]}
{"id": 4200, "image": ["rtx_frames_success_25/55_dobbe#episode_4619/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4634/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the cabinet door."}, {"from": "gpt", "value": "<point>[[462, 841], [460, 662], [434, 578], [358, 451]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[419, 766], [425, 613], [480, 460], [561, 405], [682, 329]]</point>"}]}
{"id": 4202, "image": ["rtx_frames_success_25/55_dobbe#episode_4644/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4645/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[315, 803], [318, 694], [387, 621], [535, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[462, 772], [509, 676], [506, 584], [477, 462]]</point>"}]}
{"id": 4204, "image": ["rtx_frames_success_25/55_dobbe#episode_4655/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4656/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[344, 780], [329, 645], [367, 520], [460, 373]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door handle."}, {"from": "gpt", "value": "<point>[[457, 884], [500, 749], [497, 618], [462, 483]]</point>"}]}
{"id": 4206, "image": ["rtx_frames_success_25/55_dobbe#episode_4662/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4674/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[402, 827], [382, 685], [399, 569], [442, 434], [517, 315]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[303, 777], [408, 714], [506, 601], [503, 509]]</point>"}]}
{"id": 4208, "image": ["rtx_frames_success_25/55_dobbe#episode_4679/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4688/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[422, 783], [384, 659], [416, 512], [535, 373], [601, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[442, 803], [445, 671], [434, 590], [471, 491]]</point>"}]}
{"id": 4210, "image": ["rtx_frames_success_25/55_dobbe#episode_4694/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4695/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet doors."}, {"from": "gpt", "value": "<point>[[393, 795], [399, 598], [465, 497], [590, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[416, 777], [422, 676], [419, 592], [413, 500]]</point>"}]}
{"id": 4212, "image": ["rtx_frames_success_25/55_dobbe#episode_4700/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4703/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[373, 792], [373, 676], [373, 572], [419, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[379, 798], [462, 682], [514, 566], [552, 364]]</point>"}]}
{"id": 4214, "image": ["rtx_frames_success_25/55_dobbe#episode_4724/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4727/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the fire extinguisher cabinet."}, {"from": "gpt", "value": "<point>[[454, 835], [509, 760], [512, 682]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp the refrigerator door handle."}, {"from": "gpt", "value": "<point>[[491, 835], [555, 725], [578, 529]]</point>"}]}
{"id": 4216, "image": ["rtx_frames_success_25/55_dobbe#episode_4728/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4734/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door."}, {"from": "gpt", "value": "<point>[[431, 777], [416, 659], [457, 526], [558, 431]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grip the orange."}, {"from": "gpt", "value": "<point>[[465, 795], [497, 691], [439, 514]]</point>"}]}
{"id": 4218, "image": ["rtx_frames_success_25/55_dobbe#episode_4742/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4751/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the faucet."}, {"from": "gpt", "value": "<point>[[445, 772], [500, 633], [618, 474], [769, 318], [887, 269]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the cabinet door closed."}, {"from": "gpt", "value": "<point>[[468, 809], [500, 705], [569, 575], [616, 361]]</point>"}]}
{"id": 4220, "image": ["rtx_frames_success_25/55_dobbe#episode_4766/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4767/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[480, 786], [477, 717], [486, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[442, 751], [506, 645], [543, 613], [711, 564]]</point>"}]}
{"id": 4222, "image": ["rtx_frames_success_25/55_dobbe#episode_4783/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4794/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the tennis ball."}, {"from": "gpt", "value": "<point>[[488, 853], [486, 783], [486, 731], [483, 627]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[436, 832], [477, 630], [514, 529], [613, 410]]</point>"}]}
{"id": 4224, "image": ["rtx_frames_success_25/55_dobbe#episode_4798/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4802/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[425, 803], [436, 665], [480, 578], [520, 497], [642, 393]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the tube."}, {"from": "gpt", "value": "<point>[[341, 855], [384, 783], [413, 708], [387, 538]]</point>"}]}
{"id": 4226, "image": ["rtx_frames_success_25/55_dobbe#episode_4809/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4811/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the microwave."}, {"from": "gpt", "value": "<point>[[428, 809], [428, 728], [448, 656], [581, 520]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the hand sanitizer bottle."}, {"from": "gpt", "value": "<point>[[445, 798], [497, 647], [434, 483], [303, 350]]</point>"}]}
{"id": 4228, "image": ["rtx_frames_success_25/55_dobbe#episode_4812/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4818/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[436, 760], [460, 671], [480, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the block."}, {"from": "gpt", "value": "<point>[[474, 801], [442, 705], [405, 679], [358, 633]]</point>"}]}
{"id": 4230, "image": ["rtx_frames_success_25/55_dobbe#episode_4819/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4825/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[442, 795], [445, 720], [448, 656], [486, 590]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[486, 801], [486, 723], [445, 645], [396, 595], [309, 517]]</point>"}]}
{"id": 4232, "image": ["rtx_frames_success_25/55_dobbe#episode_4827/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4830/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[514, 829], [503, 725], [497, 575], [509, 376]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[321, 847], [364, 763], [413, 679], [434, 618], [448, 509]]</point>"}]}
{"id": 4234, "image": ["rtx_frames_success_25/55_dobbe#episode_4832/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4860/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[451, 844], [506, 769], [555, 682], [555, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[367, 792], [367, 717], [399, 618], [488, 555]]</point>"}]}
{"id": 4236, "image": ["rtx_frames_success_25/55_dobbe#episode_4865/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4873/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grasp the cabinet door handle."}, {"from": "gpt", "value": "<point>[[402, 789], [405, 676], [445, 572], [552, 457]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[457, 818], [488, 728], [451, 656], [402, 590]]</point>"}]}
{"id": 4238, "image": ["rtx_frames_success_25/55_dobbe#episode_4873/frame_15.png", "rtx_frames_success_25/55_dobbe#episode_4879/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer inwards."}, {"from": "gpt", "value": "<point>[[416, 786], [468, 616], [520, 526], [613, 353]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[439, 798], [491, 685], [460, 575]]</point>"}]}
{"id": 4240, "image": ["rtx_frames_success_25/55_dobbe#episode_4882/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4885/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[460, 798], [457, 699], [451, 627], [442, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[329, 812], [315, 688], [344, 592], [439, 497]]</point>"}]}
{"id": 4242, "image": ["rtx_frames_success_25/55_dobbe#episode_4887/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4892/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[416, 835], [408, 734], [425, 653], [488, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[434, 832], [448, 754], [468, 668], [532, 587]]</point>"}]}
{"id": 4244, "image": ["rtx_frames_success_25/55_dobbe#episode_4898/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4906/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[439, 850], [439, 763], [494, 639], [595, 543], [720, 494]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[367, 786], [373, 668], [370, 598], [393, 500]]</point>"}]}
{"id": 4246, "image": ["rtx_frames_success_25/55_dobbe#episode_4916/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4918/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[454, 792], [434, 702], [382, 595], [350, 486]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[451, 838], [442, 737], [483, 618], [584, 578]]</point>"}]}
{"id": 4248, "image": ["rtx_frames_success_25/55_dobbe#episode_4924/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4942/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[428, 803], [445, 691], [480, 604], [590, 529], [656, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[419, 806], [529, 639], [564, 561], [581, 434], [540, 321]]</point>"}]}
{"id": 4250, "image": ["rtx_frames_success_25/55_dobbe#episode_4942/frame_15.png", "rtx_frames_success_25/55_dobbe#episode_4946/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[460, 798], [477, 630], [488, 535], [488, 410]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[419, 876], [448, 789], [483, 679], [465, 549]]</point>"}]}
{"id": 4252, "image": ["rtx_frames_success_25/55_dobbe#episode_4956/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4960/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handles of the cabinet door."}, {"from": "gpt", "value": "<point>[[486, 769], [480, 656], [419, 546], [309, 431]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[526, 861], [497, 749], [520, 639], [538, 512], [561, 405], [624, 298]]</point>"}]}
{"id": 4254, "image": ["rtx_frames_success_25/55_dobbe#episode_4961/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4962/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[393, 786], [379, 685], [393, 561], [428, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[439, 754], [416, 645], [416, 529], [448, 387], [491, 338], [610, 231]]</point>"}]}
{"id": 4256, "image": ["rtx_frames_success_25/55_dobbe#episode_4963/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4965/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open cabinet door."}, {"from": "gpt", "value": "<point>[[448, 812], [462, 659], [471, 564], [457, 442]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the faucet handle."}, {"from": "gpt", "value": "<point>[[460, 795], [526, 671], [538, 561], [543, 347]]</point>"}]}
{"id": 4258, "image": ["rtx_frames_success_25/55_dobbe#episode_4967/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4979/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the water bottle."}, {"from": "gpt", "value": "<point>[[477, 838], [477, 746], [442, 668], [402, 572], [382, 491]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the plush toy."}, {"from": "gpt", "value": "<point>[[439, 783], [448, 720], [460, 665]]</point>"}]}
{"id": 4260, "image": ["rtx_frames_success_25/55_dobbe#episode_4981/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4982/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[376, 821], [405, 746], [442, 653], [486, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grab the handle of the air fryer drawer."}, {"from": "gpt", "value": "<point>[[402, 763], [436, 685], [451, 639], [483, 564]]</point>"}]}
{"id": 4262, "image": ["rtx_frames_success_25/55_dobbe#episode_4995/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_4997/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[454, 772], [462, 691], [468, 645], [465, 618]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[405, 798], [451, 711], [488, 618]]</point>"}]}
{"id": 4264, "image": ["rtx_frames_success_25/55_dobbe#episode_4998/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5000/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[451, 725], [462, 590], [529, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the bottle."}, {"from": "gpt", "value": "<point>[[454, 780], [474, 668], [462, 578], [439, 497]]</point>"}]}
{"id": 4266, "image": ["rtx_frames_success_25/55_dobbe#episode_5002/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5004/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[457, 806], [462, 679], [480, 561], [506, 428]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[462, 699], [512, 595], [610, 497]]</point>"}]}
{"id": 4268, "image": ["rtx_frames_success_25/55_dobbe#episode_5012/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5023/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the button."}, {"from": "gpt", "value": "<point>[[462, 821], [448, 731], [439, 630]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the object."}, {"from": "gpt", "value": "<point>[[436, 763], [457, 682], [494, 543]]</point>"}]}
{"id": 4270, "image": ["rtx_frames_success_25/55_dobbe#episode_5024/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5029/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door handles."}, {"from": "gpt", "value": "<point>[[471, 792], [457, 702], [442, 601], [462, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[408, 760], [422, 665], [434, 538], [474, 379]]</point>"}]}
{"id": 4272, "image": ["rtx_frames_success_25/55_dobbe#episode_5032/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5033/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[431, 777], [457, 630], [480, 540], [506, 439]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[306, 792], [338, 633], [361, 532], [425, 358]]</point>"}]}
{"id": 4274, "image": ["rtx_frames_success_25/55_dobbe#episode_5040/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5041/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[442, 803], [483, 656], [543, 520], [679, 434]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[434, 780], [468, 639], [494, 509], [532, 341]]</point>"}]}
{"id": 4276, "image": ["rtx_frames_success_25/55_dobbe#episode_5041/frame_15.png", "rtx_frames_success_25/55_dobbe#episode_5048/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[413, 818], [457, 662], [462, 543], [471, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[448, 783], [529, 682], [595, 517], [653, 301]]</point>"}]}
{"id": 4278, "image": ["rtx_frames_success_25/55_dobbe#episode_5053/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5055/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[474, 829], [465, 734], [454, 607], [408, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[460, 824], [460, 714], [436, 621], [353, 491]]</point>"}]}
{"id": 4280, "image": ["rtx_frames_success_25/55_dobbe#episode_5059/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5066/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the button."}, {"from": "gpt", "value": "<point>[[419, 772], [439, 618], [454, 540], [477, 436]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[422, 841], [468, 679], [465, 503], [465, 416]]</point>"}]}
{"id": 4282, "image": ["rtx_frames_success_25/55_dobbe#episode_5091/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5092/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the hand sanitizer bottle."}, {"from": "gpt", "value": "<point>[[460, 801], [497, 671], [494, 546], [434, 379]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[431, 887], [488, 763], [540, 624], [561, 465], [540, 283]]</point>"}]}
{"id": 4284, "image": ["rtx_frames_success_25/55_dobbe#episode_5095/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5104/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[382, 835], [387, 671], [428, 546], [468, 457]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the tennis ball."}, {"from": "gpt", "value": "<point>[[448, 815], [457, 688], [468, 604]]</point>"}]}
{"id": 4286, "image": ["rtx_frames_success_25/55_dobbe#episode_5124/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5135/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the oven door."}, {"from": "gpt", "value": "<point>[[448, 844], [462, 708], [462, 595], [457, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[439, 792], [442, 717], [451, 561]]</point>"}]}
{"id": 4288, "image": ["rtx_frames_success_25/55_dobbe#episode_5138/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5139/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the power strip."}, {"from": "gpt", "value": "<point>[[399, 890], [355, 757], [329, 613], [350, 422]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the block."}, {"from": "gpt", "value": "<point>[[419, 948], [367, 792], [344, 659], [358, 460]]</point>"}]}
{"id": 4290, "image": ["rtx_frames_success_25/55_dobbe#episode_5144/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5145/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab the handles of the double door cabinet."}, {"from": "gpt", "value": "<point>[[422, 922], [370, 829], [286, 633], [240, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp the cabinet door handles."}, {"from": "gpt", "value": "<point>[[416, 965], [442, 902], [448, 841], [431, 697]]</point>"}]}
{"id": 4292, "image": ["rtx_frames_success_25/55_dobbe#episode_5150/frame_0.png", "rtx_frames_success_25/55_dobbe#episode_5151/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pen."}, {"from": "gpt", "value": "<point>[[408, 948], [384, 855], [376, 728], [422, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[379, 942], [373, 841], [399, 740], [448, 642], [558, 581]]</point>"}]}
{"id": 4294, "image": ["rtx_frames_success_25/55_dobbe#episode_5153/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_52/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the wallet."}, {"from": "gpt", "value": "<point>[[344, 934], [329, 838], [382, 734], [526, 717]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[446, 889], [421, 717], [504, 645]]</point>"}]}
{"id": 4296, "image": ["rtx_frames_success_26/55_dobbe#episode_61/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_65/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grasp the handle of the refrigerator."}, {"from": "gpt", "value": "<point>[[429, 850], [479, 460], [380, 346]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the object."}, {"from": "gpt", "value": "<point>[[429, 837], [413, 679], [471, 596]]</point>"}]}
{"id": 4298, "image": ["rtx_frames_success_26/55_dobbe#episode_82/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_86/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[465, 773], [249, 438], [139, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the faucet handle."}, {"from": "gpt", "value": "<point>[[454, 806], [446, 374], [571, 211]]</point>"}]}
{"id": 4300, "image": ["rtx_frames_success_26/55_dobbe#episode_99/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_518/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[521, 892], [421, 501], [452, 391]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the sponge."}, {"from": "gpt", "value": "<point>[[460, 809], [377, 565], [443, 440]]</point>"}]}
{"id": 4302, "image": ["rtx_frames_success_26/55_dobbe#episode_521/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_526/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[479, 820], [402, 687], [510, 582], [532, 670]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[457, 859], [352, 767], [385, 626]]</point>"}]}
{"id": 4304, "image": ["rtx_frames_success_26/55_dobbe#episode_536/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_539/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the refrigerator door."}, {"from": "gpt", "value": "<point>[[454, 839], [404, 576], [463, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[404, 853], [355, 596], [443, 501]]</point>"}]}
{"id": 4306, "image": ["rtx_frames_success_26/55_dobbe#episode_540/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_544/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[454, 864], [432, 612], [504, 496]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the plushie."}, {"from": "gpt", "value": "<point>[[446, 839], [307, 385], [382, 205]]</point>"}]}
{"id": 4308, "image": ["rtx_frames_success_26/55_dobbe#episode_552/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_552/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handles."}, {"from": "gpt", "value": "<point>[[452, 814], [584, 349], [704, 258]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the handles to open the door."}, {"from": "gpt", "value": "<point>[[435, 925], [429, 634], [532, 590]]</point>"}]}
{"id": 4310, "image": ["rtx_frames_success_26/55_dobbe#episode_555/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_557/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[435, 861], [366, 654], [402, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grasp the drawer handle."}, {"from": "gpt", "value": "<point>[[391, 859], [349, 737], [482, 632]]</point>"}]}
{"id": 4312, "image": ["rtx_frames_success_26/55_dobbe#episode_558/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_564/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab the drawer handles."}, {"from": "gpt", "value": "<point>[[482, 814], [452, 609], [537, 493]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[490, 823], [440, 598], [496, 537]]</point>"}]}
{"id": 4314, "image": ["rtx_frames_success_26/55_dobbe#episode_564/frame_15.png", "rtx_frames_success_26/55_dobbe#episode_574/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to press the light switch."}, {"from": "gpt", "value": "<point>[[490, 837], [438, 607], [510, 532]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handles."}, {"from": "gpt", "value": "<point>[[454, 892], [482, 734], [557, 704]]</point>"}]}
{"id": 4316, "image": ["rtx_frames_success_26/55_dobbe#episode_584/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_585/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the bottle on the table."}, {"from": "gpt", "value": "<point>[[460, 817], [759, 463], [898, 404]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[440, 812], [388, 421], [421, 360]]</point>"}]}
{"id": 4318, "image": ["rtx_frames_success_26/55_dobbe#episode_586/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_587/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[396, 853], [410, 634], [440, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle."}, {"from": "gpt", "value": "<point>[[391, 856], [366, 560], [418, 515]]</point>"}]}
{"id": 4320, "image": ["rtx_frames_success_26/55_dobbe#episode_590/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_594/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[515, 850], [432, 490], [501, 416]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handles."}, {"from": "gpt", "value": "<point>[[438, 848], [410, 294], [449, 177]]</point>"}]}
{"id": 4322, "image": ["rtx_frames_success_26/55_dobbe#episode_606/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_612/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the partially opened drawer."}, {"from": "gpt", "value": "<point>[[449, 856], [493, 596], [399, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[429, 720], [465, 449], [562, 380]]</point>"}]}
{"id": 4324, "image": ["rtx_frames_success_26/55_dobbe#episode_614/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_621/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[468, 717], [396, 548], [438, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[429, 831], [366, 571], [404, 496]]</point>"}]}
{"id": 4326, "image": ["rtx_frames_success_26/55_dobbe#episode_623/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_635/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[404, 853], [349, 690], [368, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[474, 825], [435, 601], [499, 568]]</point>"}]}
{"id": 4328, "image": ["rtx_frames_success_26/55_dobbe#episode_639/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_641/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[510, 795], [355, 532], [230, 537]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cube."}, {"from": "gpt", "value": "<point>[[432, 878], [407, 665], [321, 620]]</point>"}]}
{"id": 4330, "image": ["rtx_frames_success_26/55_dobbe#episode_649/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_654/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[485, 848], [377, 715], [380, 620]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[479, 795], [424, 537], [471, 452]]</point>"}]}
{"id": 4332, "image": ["rtx_frames_success_26/55_dobbe#episode_656/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_657/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the kettle."}, {"from": "gpt", "value": "<point>[[438, 828], [388, 634], [416, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[421, 809], [604, 476], [562, 418]]</point>"}]}
{"id": 4334, "image": ["rtx_frames_success_26/55_dobbe#episode_661/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_663/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[501, 695], [319, 524], [330, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the partially opened drawer."}, {"from": "gpt", "value": "<point>[[457, 731], [521, 468], [476, 388]]</point>"}]}
{"id": 4336, "image": ["rtx_frames_success_26/55_dobbe#episode_668/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_678/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the glue stick."}, {"from": "gpt", "value": "<point>[[440, 950], [385, 731], [416, 657]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[471, 825], [366, 233], [205, 169]]</point>"}]}
{"id": 4338, "image": ["rtx_frames_success_26/55_dobbe#episode_693/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_703/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the refrigerator door handle."}, {"from": "gpt", "value": "<point>[[463, 848], [510, 554], [579, 521]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[427, 850], [404, 629], [446, 488]]</point>"}]}
{"id": 4340, "image": ["rtx_frames_success_26/55_dobbe#episode_717/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_718/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[388, 900], [363, 753], [396, 709]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[457, 831], [382, 410], [468, 277]]</point>"}]}
{"id": 4342, "image": ["rtx_frames_success_26/55_dobbe#episode_720/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_728/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[490, 909], [252, 637], [163, 684]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[507, 834], [465, 632], [540, 507]]</point>"}]}
{"id": 4344, "image": ["rtx_frames_success_26/55_dobbe#episode_737/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_738/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the doorknob."}, {"from": "gpt", "value": "<point>[[421, 848], [471, 310], [421, 211]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[449, 911], [476, 715], [554, 681]]</point>"}]}
{"id": 4346, "image": ["rtx_frames_success_26/55_dobbe#episode_739/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_743/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[438, 917], [396, 765], [443, 684]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet handle."}, {"from": "gpt", "value": "<point>[[435, 834], [679, 454], [759, 460]]</point>"}]}
{"id": 4348, "image": ["rtx_frames_success_26/55_dobbe#episode_749/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_763/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[526, 898], [518, 493], [576, 388]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to grab the drawer handle."}, {"from": "gpt", "value": "<point>[[296, 834], [576, 535], [662, 584]]</point>"}]}
{"id": 4350, "image": ["rtx_frames_success_26/55_dobbe#episode_766/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_773/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cabinet door handle."}, {"from": "gpt", "value": "<point>[[435, 898], [421, 654], [501, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green object."}, {"from": "gpt", "value": "<point>[[330, 903], [283, 449], [346, 238]]</point>"}]}
{"id": 4352, "image": ["rtx_frames_success_26/55_dobbe#episode_786/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_791/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[452, 845], [407, 573], [454, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[454, 873], [609, 684], [759, 693]]</point>"}]}
{"id": 4354, "image": ["rtx_frames_success_26/55_dobbe#episode_805/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_810/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[429, 878], [391, 681], [407, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[454, 861], [515, 731], [576, 687]]</point>"}]}
{"id": 4356, "image": ["rtx_frames_success_26/55_dobbe#episode_815/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_820/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[440, 850], [399, 576], [485, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[510, 906], [474, 601], [510, 499]]</point>"}]}
{"id": 4358, "image": ["rtx_frames_success_26/55_dobbe#episode_827/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_830/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red cup."}, {"from": "gpt", "value": "<point>[[438, 864], [377, 681], [449, 604]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the blue cube."}, {"from": "gpt", "value": "<point>[[449, 792], [332, 568], [224, 568]]</point>"}]}
{"id": 4360, "image": ["rtx_frames_success_26/55_dobbe#episode_831/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_840/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[465, 914], [551, 557], [488, 393]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[479, 812], [482, 593], [562, 501]]</point>"}]}
{"id": 4362, "image": ["rtx_frames_success_26/55_dobbe#episode_842/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_849/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the oven door."}, {"from": "gpt", "value": "<point>[[446, 792], [548, 537], [507, 443]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[391, 909], [393, 776], [460, 715]]</point>"}]}
{"id": 4364, "image": ["rtx_frames_success_26/55_dobbe#episode_850/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_853/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[457, 809], [443, 524], [596, 360]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the oven door."}, {"from": "gpt", "value": "<point>[[507, 875], [418, 573], [463, 416]]</point>"}]}
{"id": 4366, "image": ["rtx_frames_success_26/55_dobbe#episode_854/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_860/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the glue stick."}, {"from": "gpt", "value": "<point>[[429, 861], [371, 601], [421, 504]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[440, 848], [582, 626], [565, 501]]</point>"}]}
{"id": 4368, "image": ["rtx_frames_success_26/55_dobbe#episode_866/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_867/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet handles."}, {"from": "gpt", "value": "<point>[[399, 809], [332, 637], [360, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the cabinet door."}, {"from": "gpt", "value": "<point>[[382, 947], [512, 737], [607, 720]]</point>"}]}
{"id": 4370, "image": ["rtx_frames_success_26/55_dobbe#episode_872/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_889/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handle of the top drawer."}, {"from": "gpt", "value": "<point>[[427, 839], [432, 526], [474, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[540, 881], [454, 524], [335, 393]]</point>"}]}
{"id": 4372, "image": ["rtx_frames_success_26/55_dobbe#episode_894/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_896/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[460, 870], [385, 557], [449, 399]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[562, 934], [543, 825], [576, 801]]</point>"}]}
{"id": 4374, "image": ["rtx_frames_success_26/55_dobbe#episode_896/frame_15.png", "rtx_frames_success_26/55_dobbe#episode_909/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to flip the light switch down."}, {"from": "gpt", "value": "<point>[[476, 867], [457, 532], [507, 454]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[471, 834], [421, 582], [499, 471]]</point>"}]}
{"id": 4376, "image": ["rtx_frames_success_26/55_dobbe#episode_910/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_920/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[443, 795], [432, 565], [518, 488]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the microwave."}, {"from": "gpt", "value": "<point>[[474, 920], [485, 770], [546, 706]]</point>"}]}
{"id": 4378, "image": ["rtx_frames_success_26/55_dobbe#episode_927/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_943/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[504, 837], [418, 490], [457, 407]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[449, 823], [427, 659], [474, 548]]</point>"}]}
{"id": 4380, "image": ["rtx_frames_success_26/55_dobbe#episode_952/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_953/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the spray bottle."}, {"from": "gpt", "value": "<point>[[452, 817], [382, 139], [269, 17]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[490, 906], [499, 413], [510, 346]]</point>"}]}
{"id": 4382, "image": ["rtx_frames_success_26/55_dobbe#episode_959/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_960/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the cabinet."}, {"from": "gpt", "value": "<point>[[482, 837], [537, 601], [598, 543]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[404, 831], [374, 299], [471, 144]]</point>"}]}
{"id": 4384, "image": ["rtx_frames_success_26/55_dobbe#episode_961/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_962/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the gray grid object."}, {"from": "gpt", "value": "<point>[[432, 803], [402, 565], [452, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[446, 909], [382, 737], [427, 654]]</point>"}]}
{"id": 4386, "image": ["rtx_frames_success_26/55_dobbe#episode_964/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_970/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the handles of the cabinet door."}, {"from": "gpt", "value": "<point>[[471, 737], [532, 346], [632, 219]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the door handle."}, {"from": "gpt", "value": "<point>[[454, 776], [371, 460], [424, 388]]</point>"}]}
{"id": 4388, "image": ["rtx_frames_success_26/55_dobbe#episode_981/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_981/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[521, 895], [454, 584], [521, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to turn the door handle."}, {"from": "gpt", "value": "<point>[[499, 861], [512, 590], [402, 512]]</point>"}]}
{"id": 4390, "image": ["rtx_frames_success_26/55_dobbe#episode_983/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_984/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[438, 875], [382, 584], [391, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[440, 848], [368, 524], [404, 452]]</point>"}]}
{"id": 4392, "image": ["rtx_frames_success_26/55_dobbe#episode_992/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_997/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door."}, {"from": "gpt", "value": "<point>[[457, 850], [607, 515], [571, 435]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the stove burner."}, {"from": "gpt", "value": "<point>[[474, 848], [363, 476], [269, 443]]</point>"}]}
{"id": 4394, "image": ["rtx_frames_success_26/55_dobbe#episode_5154/frame_0.png", "rtx_frames_success_26/55_dobbe#episode_5172/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the door handle."}, {"from": "gpt", "value": "<point>[[432, 801], [407, 662], [446, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the light switch."}, {"from": "gpt", "value": "<point>[[476, 798], [457, 554], [526, 452]]</point>"}]}
{"id": 4396, "image": ["rtx_frames_success_26/55_dobbe#episode_5172/frame_15.png", "rtx_frames_success_26/55_dobbe#episode_5174/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to flip the light switch."}, {"from": "gpt", "value": "<point>[[460, 837], [413, 654], [460, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the stuffed toy."}, {"from": "gpt", "value": "<point>[[460, 831], [410, 532], [501, 438]]</point>"}]}
{"id": 4398, "image": ["rtx_frames_success_26/55_dobbe#episode_5195/frame_0.png", "rtx_frames_success_28/57_fmb#episode_2175/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[452, 875], [393, 643], [438, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the double-square object."}, {"from": "gpt", "value": "<point>[[497, 4], [481, 85], [444, 138], [422, 196]]</point>"}]}
{"id": 4400, "image": ["rtx_frames_success_29/57_fmb#episode_390/frame_0.png", "rtx_frames_success_29/57_fmb#episode_3893/frame_0.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the cylinder."}, {"from": "gpt", "value": "<point>[[432, 80], [419, 88], [412, 99], [418, 114]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the rectangular prism."}, {"from": "gpt", "value": "<point>[[435, 80], [423, 116], [416, 161], [426, 222]]</point>"}]}
{"id": 4402, "image": ["rtx_frames_success_31/57_fmb#episode_5201/frame_0.png", "rtx_frames_success_31/57_fmb#episode_5201/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the rectangular object."}, {"from": "gpt", "value": "<point>[[543, 55], [438, 114], [380, 69]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to insert the rectangular object in its corresponding slot."}, {"from": "gpt", "value": "<point>[[316, 30], [537, 562], [687, 634]]</point>"}]}
{"id": 4404, "image": ["rtx_frames_success_33/57_fmb#episode_687/frame_15.png", "rtx_frames_success_33/57_fmb#episode_707/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the star-shaped object into the slot."}, {"from": "gpt", "value": "<point>[[422, 210], [430, 259], [428, 301], [419, 348]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cylinder over the circular hole."}, {"from": "gpt", "value": "<point>[[771, 4], [739, 49], [712, 95], [695, 152]]</point>"}]}
{"id": 4406, "image": ["rtx_frames_success_33/57_fmb#episode_712/frame_15.png", "rtx_frames_success_33/57_fmb#episode_753/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the rectangle into the slot."}, {"from": "gpt", "value": "<point>[[648, 250], [642, 299], [629, 337], [614, 371]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the rectangular prism."}, {"from": "gpt", "value": "<point>[[593, 68], [661, 138], [695, 244], [695, 354]]</point>"}]}
{"id": 4408, "image": ["rtx_frames_success_33/57_fmb#episode_6887/frame_15.png", "rtx_frames_success_33/57_fmb#episode_6907/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the star object towards its slot."}, {"from": "gpt", "value": "<point>[[415, 55], [407, 108], [409, 157], [419, 216]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the rectangular prism towards the slot."}, {"from": "gpt", "value": "<point>[[684, 129], [680, 186], [667, 218], [634, 261]]</point>"}]}
{"id": 4410, "image": ["rtx_frames_success_33/57_fmb#episode_6916/frame_15.png", "rtx_frames_success_33/57_fmb#episode_6925/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the green block towards the pink base."}, {"from": "gpt", "value": "<point>[[706, 153], [648, 176], [587, 220], [542, 265]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the blue object towards the slot."}, {"from": "gpt", "value": "<point>[[411, 134], [479, 197], [509, 280], [528, 411]]</point>"}]}
{"id": 4412, "image": ["rtx_frames_success_33/57_fmb#episode_6943/frame_15.png", "rtx_frames_success_33/57_fmb#episode_6998/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the arch into the designated slot."}, {"from": "gpt", "value": "<point>[[631, 0], [616, 30], [610, 53], [608, 81]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the blue shape sorter."}, {"from": "gpt", "value": "<point>[[532, 144], [536, 174], [536, 212], [530, 261]]</point>"}]}
{"id": 4414, "image": ["rtx_frames_success_33/57_fmb#episode_7025/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7033/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the arch-shaped object towards the blue board."}, {"from": "gpt", "value": "<point>[[792, 117], [763, 174], [737, 229], [712, 299]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the star object."}, {"from": "gpt", "value": "<point>[[379, 6], [373, 83], [373, 159], [383, 229]]</point>"}]}
{"id": 4416, "image": ["rtx_frames_success_33/57_fmb#episode_7056/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7136/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the arch-shaped object."}, {"from": "gpt", "value": "<point>[[684, 131], [667, 161], [663, 186], [669, 218]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the hexagon towards the slot."}, {"from": "gpt", "value": "<point>[[402, 85], [407, 114], [407, 133], [403, 161]]</point>"}]}
{"id": 4418, "image": ["rtx_frames_success_33/57_fmb#episode_7173/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7187/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position the arch over the matching slot."}, {"from": "gpt", "value": "<point>[[669, 178], [670, 201], [667, 225], [661, 256]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the double-square object towards the slot."}, {"from": "gpt", "value": "<point>[[434, 53], [532, 76], [619, 125], [691, 218]]</point>"}]}
{"id": 4420, "image": ["rtx_frames_success_33/57_fmb#episode_7213/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7216/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position the cylinder above the circular hole."}, {"from": "gpt", "value": "<point>[[513, 295], [509, 341], [509, 386], [521, 420]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to release the star."}, {"from": "gpt", "value": "<point>[[354, 208], [364, 254], [371, 288], [371, 320]]</point>"}]}
{"id": 4422, "image": ["rtx_frames_success_33/57_fmb#episode_7218/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7259/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the purple block towards the designated slot."}, {"from": "gpt", "value": "<point>[[674, 169], [595, 222], [521, 318], [472, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to insert the three-pronged object into its corresponding slot."}, {"from": "gpt", "value": "<point>[[562, 89], [555, 127], [551, 165], [553, 210]]</point>"}]}
{"id": 4424, "image": ["rtx_frames_success_33/57_fmb#episode_7264/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7283/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the double-square shape into its corresponding slot."}, {"from": "gpt", "value": "<point>[[627, 163], [623, 218], [621, 248], [619, 288]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to insert the double-square object into its corresponding slot."}, {"from": "gpt", "value": "<point>[[805, 119], [799, 172], [786, 214], [775, 246]]</point>"}]}
{"id": 4426, "image": ["rtx_frames_success_33/57_fmb#episode_7313/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7325/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the rectangular object into its corresponding slot."}, {"from": "gpt", "value": "<point>[[597, 188], [598, 256], [593, 330], [583, 407]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position the three-pronged object over its designated slot."}, {"from": "gpt", "value": "<point>[[555, 104], [544, 140], [542, 172], [542, 208]]</point>"}]}
{"id": 4428, "image": ["rtx_frames_success_33/57_fmb#episode_7327/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7337/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the blue block towards the puzzle board."}, {"from": "gpt", "value": "<point>[[532, 163], [517, 246], [515, 330], [519, 417]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position the arch over the matching slot."}, {"from": "gpt", "value": "<point>[[705, 9], [703, 38], [697, 70], [688, 98]]</point>"}]}
{"id": 4430, "image": ["rtx_frames_success_33/57_fmb#episode_7340/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7349/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position the oval object over the oval slot."}, {"from": "gpt", "value": "<point>[[545, 297], [547, 360], [547, 394], [538, 449]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the cylinder towards the matching hole."}, {"from": "gpt", "value": "<point>[[640, 153], [676, 87], [598, 38], [498, 11]]</point>"}]}
{"id": 4432, "image": ["rtx_frames_success_33/57_fmb#episode_7380/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7400/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the cylinder into the hole."}, {"from": "gpt", "value": "<point>[[540, 123], [534, 182], [538, 241], [549, 290]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the arch-shaped object towards the green board."}, {"from": "gpt", "value": "<point>[[515, 27], [521, 68], [519, 97], [511, 133]]</point>"}]}
{"id": 4434, "image": ["rtx_frames_success_33/57_fmb#episode_7409/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7410/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position the rectangular object over the matching slot."}, {"from": "gpt", "value": "<point>[[598, 121], [591, 155], [589, 197], [585, 241]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to insert the double-square object into its corresponding slot."}, {"from": "gpt", "value": "<point>[[688, 36], [688, 95], [682, 161], [674, 222]]</point>"}]}
{"id": 4436, "image": ["rtx_frames_success_33/57_fmb#episode_7446/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7457/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position the rectangular object over the matching slot."}, {"from": "gpt", "value": "<point>[[627, 256], [616, 311], [610, 367], [610, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to position the object above the correct slot."}, {"from": "gpt", "value": "<point>[[729, 110], [714, 153], [701, 206], [695, 250]]</point>"}]}
{"id": 4438, "image": ["rtx_frames_success_33/57_fmb#episode_7461/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7477/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the arch-shaped object into its corresponding slot."}, {"from": "gpt", "value": "<point>[[598, 91], [616, 134], [617, 178], [604, 225]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lower the rectangle towards the slot."}, {"from": "gpt", "value": "<point>[[612, 186], [616, 231], [610, 269], [600, 318]]</point>"}]}
{"id": 4440, "image": ["rtx_frames_success_33/57_fmb#episode_7501/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7536/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the star object."}, {"from": "gpt", "value": "<point>[[407, 4], [400, 110], [398, 165], [403, 284]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the arch-shaped object."}, {"from": "gpt", "value": "<point>[[608, 95], [587, 133], [570, 172], [562, 212]]</point>"}]}
{"id": 4442, "image": ["rtx_frames_success_33/57_fmb#episode_7539/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7561/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the star-shaped object into its corresponding slot."}, {"from": "gpt", "value": "<point>[[419, 30], [411, 97], [411, 150], [417, 208]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to align the object with the star-shaped slot."}, {"from": "gpt", "value": "<point>[[297, 244], [311, 288], [318, 331], [324, 390]]</point>"}]}
{"id": 4444, "image": ["rtx_frames_success_33/57_fmb#episode_7605/frame_15.png", "rtx_frames_success_33/57_fmb#episode_7609/frame_15.png"], "height": [256, 256], "width": [256, 256], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position the arch-shaped object over the matching slot."}, {"from": "gpt", "value": "<point>[[566, 153], [568, 189], [568, 220], [561, 254]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the round object."}, {"from": "gpt", "value": "<point>[[485, 13], [477, 64], [468, 131], [468, 189]]</point>"}]}
{"id": 4446, "image": ["rtx_frames_success_33/57_fmb#episode_7629/frame_15.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_4/frame_0.png"], "height": [256, 480], "width": [256, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to insert the double-square object into its corresponding slot."}, {"from": "gpt", "value": "<point>[[627, 189], [627, 233], [617, 277], [600, 326]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[460, 336], [510, 363], [533, 405], [550, 440]]</point>"}]}
{"id": 4448, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_4/frame_15.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_64/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the blue cup towards the red cup."}, {"from": "gpt", "value": "<point>[[484, 355], [475, 398], [469, 440], [469, 483], [501, 440], [518, 394]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up a green cup."}, {"from": "gpt", "value": "<point>[[463, 351], [492, 409], [420, 390], [365, 444]]</point>"}]}
{"id": 4450, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_84/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_84/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[486, 336], [536, 320], [556, 351], [562, 390]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the pan."}, {"from": "gpt", "value": "<point>[[602, 320], [542, 363], [527, 448], [539, 510]]</point>"}]}
{"id": 4452, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_106/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_124/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the crumpled paper."}, {"from": "gpt", "value": "<point>[[452, 328], [434, 386], [423, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the highlighter."}, {"from": "gpt", "value": "<point>[[463, 324], [437, 382], [408, 444], [385, 490]]</point>"}]}
{"id": 4454, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_130/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_135/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to grab the watering can."}, {"from": "gpt", "value": "<point>[[469, 332], [492, 382], [513, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the watering can."}, {"from": "gpt", "value": "<point>[[469, 336], [504, 367], [539, 409]]</point>"}]}
{"id": 4456, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_142/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_142/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the vegetables."}, {"from": "gpt", "value": "<point>[[463, 336], [492, 390], [513, 440], [524, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the vegetable towards the plate."}, {"from": "gpt", "value": "<point>[[501, 402], [501, 448], [513, 521]]</point>"}]}
{"id": 4458, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_144/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_158/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the blue bowl."}, {"from": "gpt", "value": "<point>[[466, 351], [510, 375], [550, 402], [579, 429]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place a tissue on the green plate."}, {"from": "gpt", "value": "<point>[[466, 286], [455, 371], [446, 436], [458, 502]]</point>"}]}
{"id": 4460, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_159/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_159/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[463, 347], [429, 382], [385, 429], [362, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pan."}, {"from": "gpt", "value": "<point>[[362, 398], [403, 409], [449, 444]]</point>"}]}
{"id": 4462, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_162/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_162/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for a green bowl."}, {"from": "gpt", "value": "<point>[[469, 347], [440, 390], [417, 432], [417, 479]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the yellow bowl."}, {"from": "gpt", "value": "<point>[[443, 382], [463, 432], [475, 475]]</point>"}]}
{"id": 4464, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_164/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_164/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the frying pan."}, {"from": "gpt", "value": "<point>[[466, 328], [492, 378], [539, 432], [573, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to place the capsicum in the pan."}, {"from": "gpt", "value": "<point>[[475, 363], [495, 398], [510, 417]]</point>"}]}
{"id": 4466, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_169/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_169/frame_15.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for a red cup."}, {"from": "gpt", "value": "<point>[[466, 320], [486, 382], [504, 506]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to release the red cup."}, {"from": "gpt", "value": "<point>[[472, 332], [515, 351], [544, 394]]</point>"}]}
{"id": 4468, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_173/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_177/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the red cup."}, {"from": "gpt", "value": "<point>[[458, 340], [385, 378], [353, 452], [345, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up a sponge."}, {"from": "gpt", "value": "<point>[[463, 332], [440, 378], [411, 463]]</point>"}]}
{"id": 4470, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_177/frame_15.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_179/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place a tissue in the pot."}, {"from": "gpt", "value": "<point>[[463, 363], [533, 378], [573, 421], [594, 471]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the pink plate."}, {"from": "gpt", "value": "<point>[[472, 328], [423, 382], [385, 425], [391, 471]]</point>"}]}
{"id": 4472, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_185/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_188/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up a food item from the pan."}, {"from": "gpt", "value": "<point>[[466, 340], [489, 382], [513, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the vegetable with a fork."}, {"from": "gpt", "value": "<point>[[463, 340], [437, 398], [443, 459], [495, 456], [568, 432], [599, 421]]</point>"}]}
{"id": 4474, "image": ["rtx_frames_success_35/61_qut_dexterous_manpulation#episode_193/frame_0.png", "rtx_frames_success_35/61_qut_dexterous_manpulation#episode_199/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the dice."}, {"from": "gpt", "value": "<point>[[469, 351], [495, 398], [518, 425], [550, 517]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the watering can."}, {"from": "gpt", "value": "<point>[[460, 328], [440, 355], [408, 398], [394, 479]]</point>"}]}
{"id": 4476, "image": ["rtx_frames_success_35/62_robo_set#episode_0/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_100/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[309, 454], [382, 541], [491, 567], [590, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the oven."}, {"from": "gpt", "value": "<point>[[298, 444], [387, 490], [494, 526], [633, 541], [867, 531]]</point>"}]}
{"id": 4478, "image": ["rtx_frames_success_35/62_robo_set#episode_1016/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_1025/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[312, 475], [382, 444], [439, 408], [520, 327]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[324, 485], [405, 551], [506, 587], [604, 587]]</point>"}]}
{"id": 4480, "image": ["rtx_frames_success_35/62_robo_set#episode_1027/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_1034/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[321, 475], [410, 434], [486, 403], [535, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[329, 460], [413, 546], [503, 567], [607, 577]]</point>"}]}
{"id": 4482, "image": ["rtx_frames_success_35/62_robo_set#episode_10004/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10016/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 439], [382, 511], [480, 567], [561, 618]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[329, 485], [428, 541], [486, 551], [543, 567], [604, 567]]</point>"}]}
{"id": 4484, "image": ["rtx_frames_success_35/62_robo_set#episode_10020/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10024/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[347, 439], [465, 505], [581, 511], [685, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[318, 495], [410, 546], [483, 572], [584, 582]]</point>"}]}
{"id": 4486, "image": ["rtx_frames_success_35/62_robo_set#episode_10039/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10042/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red object on the plate."}, {"from": "gpt", "value": "<point>[[332, 460], [410, 505], [491, 531], [561, 551], [627, 551], [688, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[321, 439], [353, 587], [393, 633], [439, 659]]</point>"}]}
{"id": 4488, "image": ["rtx_frames_success_35/62_robo_set#episode_10077/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10078/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[306, 470], [361, 439], [439, 408], [538, 317]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[390, 710], [439, 557], [488, 511], [543, 485]]</point>"}]}
{"id": 4490, "image": ["rtx_frames_success_35/62_robo_set#episode_10080/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10093/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[301, 454], [376, 505], [468, 546], [566, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[309, 460], [408, 587], [512, 669], [627, 684]]</point>"}]}
{"id": 4492, "image": ["rtx_frames_success_35/62_robo_set#episode_10098/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10110/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[312, 480], [347, 562], [370, 592], [410, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana in the container."}, {"from": "gpt", "value": "<point>[[306, 480], [382, 551], [465, 592], [558, 618], [691, 628], [806, 628]]</point>"}]}
{"id": 4494, "image": ["rtx_frames_success_35/62_robo_set#episode_10125/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10132/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[312, 460], [399, 419], [474, 388], [514, 352]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[451, 633], [558, 623], [618, 628]]</point>"}]}
{"id": 4496, "image": ["rtx_frames_success_35/62_robo_set#episode_10136/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10142/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[318, 465], [402, 557], [483, 587], [564, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 449], [422, 449], [517, 449], [682, 429]]</point>"}]}
{"id": 4498, "image": ["rtx_frames_success_35/62_robo_set#episode_10143/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10150/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[315, 475], [382, 531], [457, 562], [607, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[324, 470], [419, 526], [494, 546], [621, 572]]</point>"}]}
{"id": 4500, "image": ["rtx_frames_success_35/62_robo_set#episode_10156/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10162/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[315, 454], [373, 536], [442, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[329, 475], [364, 368], [439, 281], [566, 230]]</point>"}]}
{"id": 4502, "image": ["rtx_frames_success_35/62_robo_set#episode_10166/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10167/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[318, 454], [382, 531], [454, 557], [581, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[292, 465], [361, 551], [436, 592], [509, 608], [598, 608]]</point>"}]}
{"id": 4504, "image": ["rtx_frames_success_35/62_robo_set#episode_10199/frame_15.png", "rtx_frames_success_35/62_robo_set#episode_10200/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[439, 311], [402, 383], [358, 449], [312, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 460], [399, 562], [483, 597], [592, 608]]</point>"}]}
{"id": 4506, "image": ["rtx_frames_success_35/62_robo_set#episode_10203/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10217/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[335, 480], [419, 567], [514, 582], [598, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 470], [399, 531], [520, 541], [616, 546], [780, 551]]</point>"}]}
{"id": 4508, "image": ["rtx_frames_success_35/62_robo_set#episode_10225/frame_15.png", "rtx_frames_success_35/62_robo_set#episode_10227/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[379, 398], [335, 475], [301, 551], [254, 623]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 470], [448, 500], [566, 500], [812, 500]]</point>"}]}
{"id": 4510, "image": ["rtx_frames_success_35/62_robo_set#episode_10294/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10299/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[295, 434], [373, 516], [442, 567], [564, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[318, 449], [396, 429], [448, 398], [520, 347]]</point>"}]}
{"id": 4512, "image": ["rtx_frames_success_35/62_robo_set#episode_10320/frame_0.png", "rtx_frames_success_35/62_robo_set#episode_10320/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[324, 465], [422, 378], [529, 306], [653, 184]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[592, 281], [607, 475], [627, 592], [653, 669]]</point>"}]}
{"id": 4514, "image": ["rtx_frames_success_35/62_robo_set#episode_10332/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_111/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 449], [413, 536], [512, 577], [604, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the drawer."}, {"from": "gpt", "value": "<point>[[399, 301], [335, 373], [283, 444]]</point>"}]}
{"id": 4516, "image": ["rtx_frames_success_36/62_robo_set#episode_1042/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_1065/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[327, 460], [393, 546], [460, 592], [552, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[312, 465], [396, 526], [483, 562], [595, 587]]</point>"}]}
{"id": 4518, "image": ["rtx_frames_success_36/62_robo_set#episode_1067/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_1073/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[329, 465], [442, 531], [538, 572], [656, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[321, 460], [387, 521], [465, 572], [581, 638]]</point>"}]}
{"id": 4520, "image": ["rtx_frames_success_36/62_robo_set#episode_1091/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_1097/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[332, 470], [425, 526], [512, 541], [592, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 470], [382, 373], [436, 322], [512, 271]]</point>"}]}
{"id": 4522, "image": ["rtx_frames_success_36/62_robo_set#episode_1122/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10350/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[321, 449], [439, 551], [509, 587], [575, 618]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[344, 444], [410, 511], [483, 557], [587, 567]]</point>"}]}
{"id": 4524, "image": ["rtx_frames_success_36/62_robo_set#episode_10354/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10368/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[303, 465], [422, 526], [491, 567], [578, 603]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[306, 470], [399, 516], [486, 562], [610, 623]]</point>"}]}
{"id": 4526, "image": ["rtx_frames_success_36/62_robo_set#episode_10373/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10390/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[303, 470], [364, 516], [402, 551], [439, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[321, 470], [416, 505], [483, 546], [607, 587]]</point>"}]}
{"id": 4528, "image": ["rtx_frames_success_36/62_robo_set#episode_10426/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10438/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[344, 449], [434, 363], [509, 317]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[321, 424], [436, 357], [532, 306], [627, 245]]</point>"}]}
{"id": 4530, "image": ["rtx_frames_success_36/62_robo_set#episode_10445/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10460/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[327, 485], [413, 526], [483, 567], [549, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[321, 480], [393, 521], [465, 577], [555, 603]]</point>"}]}
{"id": 4532, "image": ["rtx_frames_success_36/62_robo_set#episode_10462/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10471/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[324, 454], [413, 511], [520, 541], [642, 551], [694, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the container lid."}, {"from": "gpt", "value": "<point>[[321, 470], [413, 500], [500, 511], [587, 521], [679, 536]]</point>"}]}
{"id": 4534, "image": ["rtx_frames_success_36/62_robo_set#episode_10474/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10477/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[318, 465], [402, 521], [468, 562], [538, 597], [595, 603]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[338, 454], [422, 505], [491, 531], [558, 572]]</point>"}]}
{"id": 4536, "image": ["rtx_frames_success_36/62_robo_set#episode_10480/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10488/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[312, 495], [399, 434], [480, 388], [535, 317]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[321, 475], [382, 516], [454, 562], [517, 587], [610, 603]]</point>"}]}
{"id": 4538, "image": ["rtx_frames_success_36/62_robo_set#episode_10490/frame_15.png", "rtx_frames_success_36/62_robo_set#episode_10497/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[439, 332], [408, 424], [367, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[318, 460], [410, 490], [506, 516], [566, 541], [630, 567]]</point>"}]}
{"id": 4540, "image": ["rtx_frames_success_36/62_robo_set#episode_10541/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10558/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[335, 465], [451, 368], [532, 281], [647, 143]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[318, 449], [364, 388], [454, 301], [584, 194]]</point>"}]}
{"id": 4542, "image": ["rtx_frames_success_36/62_robo_set#episode_10558/frame_15.png", "rtx_frames_success_36/62_robo_set#episode_10568/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[616, 306], [601, 449], [604, 577], [616, 674]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[321, 470], [387, 541], [471, 572], [532, 592], [584, 597]]</point>"}]}
{"id": 4544, "image": ["rtx_frames_success_36/62_robo_set#episode_10571/frame_15.png", "rtx_frames_success_36/62_robo_set#episode_10573/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[361, 408], [422, 342], [477, 260], [500, 235]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 439], [376, 500], [451, 572], [520, 603], [575, 623]]</point>"}]}
{"id": 4546, "image": ["rtx_frames_success_36/62_robo_set#episode_10574/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10581/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[332, 490], [410, 434], [474, 393], [523, 337]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[324, 460], [387, 516], [448, 536], [500, 546], [564, 557], [610, 567]]</point>"}]}
{"id": 4548, "image": ["rtx_frames_success_36/62_robo_set#episode_10590/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10594/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[321, 475], [413, 505], [491, 536], [561, 546], [604, 562]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[350, 439], [428, 531], [488, 597], [561, 638], [618, 674]]</point>"}]}
{"id": 4550, "image": ["rtx_frames_success_36/62_robo_set#episode_10602/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10608/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[312, 470], [382, 526], [451, 551], [517, 557], [601, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[309, 460], [376, 511], [462, 551], [572, 597]]</point>"}]}
{"id": 4552, "image": ["rtx_frames_success_36/62_robo_set#episode_10612/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10616/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[335, 475], [410, 536], [488, 572], [578, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[332, 460], [390, 398], [448, 347], [512, 311]]</point>"}]}
{"id": 4554, "image": ["rtx_frames_success_36/62_robo_set#episode_10631/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10665/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[312, 475], [393, 511], [483, 521], [601, 562]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[396, 434], [439, 357], [514, 245]]</point>"}]}
{"id": 4556, "image": ["rtx_frames_success_36/62_robo_set#episode_10672/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10676/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[324, 454], [422, 511], [509, 541], [590, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[318, 460], [396, 490], [471, 526], [555, 587]]</point>"}]}
{"id": 4558, "image": ["rtx_frames_success_36/62_robo_set#episode_10689/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10713/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[321, 465], [384, 388], [448, 342], [506, 327]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the container lid."}, {"from": "gpt", "value": "<point>[[306, 449], [402, 475], [506, 500], [613, 521], [671, 531]]</point>"}]}
{"id": 4560, "image": ["rtx_frames_success_36/62_robo_set#episode_10723/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10725/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[324, 485], [422, 505], [497, 521], [604, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[315, 480], [399, 526], [474, 567], [595, 597]]</point>"}]}
{"id": 4562, "image": ["rtx_frames_success_36/62_robo_set#episode_10753/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10771/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[324, 460], [367, 562], [422, 648]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[303, 444], [390, 490], [477, 536], [581, 592]]</point>"}]}
{"id": 4564, "image": ["rtx_frames_success_36/62_robo_set#episode_10778/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10789/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 470], [396, 480], [503, 500], [598, 505], [688, 516], [827, 500]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[312, 439], [428, 536], [503, 562], [575, 587]]</point>"}]}
{"id": 4566, "image": ["rtx_frames_success_36/62_robo_set#episode_10801/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10802/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[321, 485], [408, 495], [494, 516], [558, 541], [601, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[315, 444], [416, 541], [509, 562], [581, 572]]</point>"}]}
{"id": 4568, "image": ["rtx_frames_success_36/62_robo_set#episode_10807/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10815/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[306, 470], [387, 511], [457, 551], [590, 643]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the banana."}, {"from": "gpt", "value": "<point>[[312, 470], [387, 511], [477, 546], [601, 557]]</point>"}]}
{"id": 4570, "image": ["rtx_frames_success_36/62_robo_set#episode_10825/frame_15.png", "rtx_frames_success_36/62_robo_set#episode_10847/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer open."}, {"from": "gpt", "value": "<point>[[376, 541], [431, 449], [494, 368]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[306, 465], [379, 419], [471, 368], [529, 332]]</point>"}]}
{"id": 4572, "image": ["rtx_frames_success_36/62_robo_set#episode_10852/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10853/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[312, 454], [408, 541], [494, 567], [566, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[344, 475], [434, 500], [506, 500], [607, 500], [676, 495]]</point>"}]}
{"id": 4574, "image": ["rtx_frames_success_36/62_robo_set#episode_10854/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10861/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[309, 480], [376, 424], [442, 368], [514, 342]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[321, 475], [396, 383], [471, 337], [520, 317]]</point>"}]}
{"id": 4576, "image": ["rtx_frames_success_36/62_robo_set#episode_10872/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10875/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[318, 485], [396, 546], [465, 582], [535, 608], [604, 613]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 465], [382, 465], [471, 475], [575, 485], [665, 505], [795, 505]]</point>"}]}
{"id": 4578, "image": ["rtx_frames_success_36/62_robo_set#episode_10891/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10893/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[344, 449], [503, 562], [584, 603], [679, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[329, 500], [382, 531], [480, 557], [543, 562], [613, 567], [688, 562]]</point>"}]}
{"id": 4580, "image": ["rtx_frames_success_36/62_robo_set#episode_10904/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10909/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 475], [402, 526], [474, 541], [566, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[324, 454], [422, 516], [517, 546], [607, 572]]</point>"}]}
{"id": 4582, "image": ["rtx_frames_success_36/62_robo_set#episode_10914/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10915/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[327, 470], [399, 516], [486, 551], [549, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[332, 470], [460, 526], [569, 546], [705, 572], [821, 587]]</point>"}]}
{"id": 4584, "image": ["rtx_frames_success_36/62_robo_set#episode_10922/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10923/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 470], [405, 388], [460, 347], [529, 322]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[329, 460], [413, 526], [503, 546], [598, 567]]</point>"}]}
{"id": 4586, "image": ["rtx_frames_success_36/62_robo_set#episode_10925/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10948/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 449], [382, 511], [460, 541], [543, 562], [642, 577], [734, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[315, 460], [367, 546], [451, 618]]</point>"}]}
{"id": 4588, "image": ["rtx_frames_success_36/62_robo_set#episode_10967/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10980/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[321, 460], [402, 526], [486, 557], [581, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[393, 332], [361, 403], [338, 454], [318, 526]]</point>"}]}
{"id": 4590, "image": ["rtx_frames_success_36/62_robo_set#episode_10997/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_10997/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[309, 449], [393, 505], [465, 531], [543, 541], [639, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the microwave."}, {"from": "gpt", "value": "<point>[[488, 551], [558, 531], [656, 500], [760, 516], [844, 546]]</point>"}]}
{"id": 4592, "image": ["rtx_frames_success_36/62_robo_set#episode_11001/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11024/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[306, 475], [434, 531], [532, 562], [621, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[321, 470], [419, 511], [506, 531], [613, 546]]</point>"}]}
{"id": 4594, "image": ["rtx_frames_success_36/62_robo_set#episode_11028/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11030/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[309, 460], [379, 393], [462, 317], [509, 301]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[350, 230], [324, 306], [298, 444]]</point>"}]}
{"id": 4596, "image": ["rtx_frames_success_36/62_robo_set#episode_11039/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11043/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[329, 454], [355, 546], [384, 618], [410, 654]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[318, 460], [390, 521], [448, 567], [578, 597]]</point>"}]}
{"id": 4598, "image": ["rtx_frames_success_36/62_robo_set#episode_11051/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11062/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[306, 480], [379, 546], [425, 562], [566, 603]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 460], [460, 485], [584, 500], [775, 562]]</point>"}]}
{"id": 4600, "image": ["rtx_frames_success_36/62_robo_set#episode_11071/frame_15.png", "rtx_frames_success_36/62_robo_set#episode_11075/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[321, 276], [301, 357], [257, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[468, 296], [410, 383], [379, 444], [355, 500]]</point>"}]}
{"id": 4602, "image": ["rtx_frames_success_36/62_robo_set#episode_11092/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11109/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 439], [425, 521], [497, 572], [564, 608]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[329, 475], [416, 429], [483, 393], [555, 322]]</point>"}]}
{"id": 4604, "image": ["rtx_frames_success_36/62_robo_set#episode_11123/frame_15.png", "rtx_frames_success_36/62_robo_set#episode_11128/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[454, 301], [396, 403], [347, 475], [306, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[327, 485], [405, 562], [483, 567], [604, 567]]</point>"}]}
{"id": 4606, "image": ["rtx_frames_success_36/62_robo_set#episode_11143/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11147/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[312, 454], [451, 526], [558, 587], [676, 608], [786, 608]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[312, 485], [405, 526], [488, 557], [604, 562]]</point>"}]}
{"id": 4608, "image": ["rtx_frames_success_36/62_robo_set#episode_11160/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11167/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[298, 465], [399, 516], [483, 546], [564, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[309, 460], [439, 536], [514, 562], [581, 603]]</point>"}]}
{"id": 4610, "image": ["rtx_frames_success_36/62_robo_set#episode_11172/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11173/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[312, 470], [410, 541], [512, 572], [587, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[327, 475], [376, 424], [422, 388], [514, 327]]</point>"}]}
{"id": 4612, "image": ["rtx_frames_success_36/62_robo_set#episode_11187/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11202/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[318, 505], [408, 505], [514, 516], [633, 505], [789, 485]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[335, 460], [390, 511], [471, 541], [566, 557], [618, 567]]</point>"}]}
{"id": 4614, "image": ["rtx_frames_success_36/62_robo_set#episode_11221/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11227/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 460], [396, 500], [457, 526], [584, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[312, 480], [428, 541], [532, 562]]</point>"}]}
{"id": 4616, "image": ["rtx_frames_success_36/62_robo_set#episode_11231/frame_0.png", "rtx_frames_success_36/62_robo_set#episode_11232/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the strainer."}, {"from": "gpt", "value": "<point>[[315, 480], [390, 526], [497, 551], [587, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[306, 475], [428, 500], [529, 521], [598, 577]]</point>"}]}
{"id": 4618, "image": ["rtx_frames_success_37/62_robo_set#episode_113/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_117/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[351, 425], [382, 358], [444, 306], [520, 306]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[353, 440], [410, 328], [509, 246], [632, 216]]</point>"}]}
{"id": 4620, "image": ["rtx_frames_success_37/62_robo_set#episode_117/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_1125/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[627, 291], [598, 425], [598, 534], [629, 604]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[338, 463], [325, 545], [336, 604], [393, 649]]</point>"}]}
{"id": 4622, "image": ["rtx_frames_success_37/62_robo_set#episode_1170/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_1170/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[351, 463], [359, 549], [384, 590], [425, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[475, 582], [539, 549], [598, 575], [644, 623]]</point>"}]}
{"id": 4624, "image": ["rtx_frames_success_37/62_robo_set#episode_1202/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11238/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[349, 473], [391, 342], [444, 286], [532, 299]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[353, 470], [433, 437], [553, 444], [621, 545]]</point>"}]}
{"id": 4626, "image": ["rtx_frames_success_37/62_robo_set#episode_11246/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11264/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[334, 470], [323, 549], [346, 634], [391, 675]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[355, 463], [427, 470], [503, 511], [577, 601]]</point>"}]}
{"id": 4628, "image": ["rtx_frames_success_37/62_robo_set#episode_11272/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11279/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[342, 470], [420, 496], [482, 560], [589, 642]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[368, 463], [452, 470], [536, 534], [600, 672]]</point>"}]}
{"id": 4630, "image": ["rtx_frames_success_37/62_robo_set#episode_11279/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_11287/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[638, 489], [623, 384], [619, 295], [653, 239]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[353, 463], [359, 511], [359, 567], [391, 608]]</point>"}]}
{"id": 4632, "image": ["rtx_frames_success_37/62_robo_set#episode_11290/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11290/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[359, 440], [427, 306], [511, 243], [627, 187]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[629, 276], [587, 403], [577, 500], [610, 597]]</point>"}]}
{"id": 4634, "image": ["rtx_frames_success_37/62_robo_set#episode_11294/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11320/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[340, 444], [387, 522], [444, 575], [541, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[357, 466], [346, 537], [357, 604], [387, 627]]</point>"}]}
{"id": 4636, "image": ["rtx_frames_success_37/62_robo_set#episode_11326/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11348/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[351, 489], [391, 612], [465, 649], [560, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 485], [454, 519], [562, 612], [608, 698]]</point>"}]}
{"id": 4638, "image": ["rtx_frames_success_37/62_robo_set#episode_11348/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_11350/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[640, 425], [608, 295], [613, 231], [636, 146]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[361, 470], [431, 455], [498, 470], [593, 556]]</point>"}]}
{"id": 4640, "image": ["rtx_frames_success_37/62_robo_set#episode_11355/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11360/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[338, 466], [384, 567], [469, 593], [560, 556]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 470], [418, 463], [503, 522], [560, 597]]</point>"}]}
{"id": 4642, "image": ["rtx_frames_success_37/62_robo_set#episode_11365/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11368/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[348, 463], [397, 556], [458, 608], [532, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana on the plate."}, {"from": "gpt", "value": "<point>[[359, 459], [389, 384], [433, 336], [507, 336]]</point>"}]}
{"id": 4644, "image": ["rtx_frames_success_37/62_robo_set#episode_11370/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11370/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 466], [410, 354], [507, 280], [583, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[619, 511], [615, 590], [615, 646], [644, 679]]</point>"}]}
{"id": 4646, "image": ["rtx_frames_success_37/62_robo_set#episode_11384/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11387/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 511], [368, 578], [399, 627], [446, 642]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[374, 455], [431, 306], [498, 224], [634, 190]]</point>"}]}
{"id": 4648, "image": ["rtx_frames_success_37/62_robo_set#episode_11387/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_11393/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to release the mug."}, {"from": "gpt", "value": "<point>[[610, 295], [570, 437], [562, 567], [625, 660]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[344, 466], [416, 444], [458, 474], [536, 560]]</point>"}]}
{"id": 4650, "image": ["rtx_frames_success_37/62_robo_set#episode_11399/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11407/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 478], [348, 552], [357, 612], [391, 649]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[359, 455], [391, 373], [448, 325], [507, 325]]</point>"}]}
{"id": 4652, "image": ["rtx_frames_success_37/62_robo_set#episode_11420/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11428/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[365, 485], [397, 384], [465, 313], [520, 317]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[361, 481], [363, 534], [368, 597], [399, 619]]</point>"}]}
{"id": 4654, "image": ["rtx_frames_success_37/62_robo_set#episode_11461/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11472/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[351, 463], [395, 563], [467, 597], [587, 604]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[344, 474], [393, 571], [456, 604], [551, 575]]</point>"}]}
{"id": 4656, "image": ["rtx_frames_success_37/62_robo_set#episode_11479/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11499/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[359, 485], [435, 463], [503, 485], [579, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 474], [422, 463], [484, 507], [547, 627]]</point>"}]}
{"id": 4658, "image": ["rtx_frames_success_37/62_robo_set#episode_11509/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11511/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[359, 455], [425, 444], [473, 463], [547, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[351, 478], [435, 478], [507, 515], [577, 616]]</point>"}]}
{"id": 4660, "image": ["rtx_frames_success_37/62_robo_set#episode_11529/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11532/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[365, 474], [425, 601], [513, 657], [574, 619]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[376, 463], [503, 440], [583, 448], [803, 485]]</point>"}]}
{"id": 4662, "image": ["rtx_frames_success_37/62_robo_set#episode_11542/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11548/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[340, 466], [334, 563], [346, 657], [393, 675]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[370, 474], [425, 440], [490, 481], [562, 604]]</point>"}]}
{"id": 4664, "image": ["rtx_frames_success_37/62_robo_set#episode_11607/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11608/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[348, 451], [448, 418], [507, 440], [598, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 560], [414, 619], [507, 616], [604, 571]]</point>"}]}
{"id": 4666, "image": ["rtx_frames_success_37/62_robo_set#episode_11614/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11631/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 463], [429, 463], [486, 496], [587, 593]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[359, 463], [357, 530], [357, 593], [384, 642]]</point>"}]}
{"id": 4668, "image": ["rtx_frames_success_37/62_robo_set#episode_11636/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11643/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[353, 478], [422, 463], [479, 493], [562, 590]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[353, 448], [433, 448], [494, 493], [572, 578]]</point>"}]}
{"id": 4670, "image": ["rtx_frames_success_37/62_robo_set#episode_11649/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11660/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[348, 493], [372, 403], [429, 347], [520, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[359, 459], [427, 463], [477, 493], [551, 612]]</point>"}]}
{"id": 4672, "image": ["rtx_frames_success_37/62_robo_set#episode_11673/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11673/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[370, 459], [359, 545], [365, 627], [406, 657]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[391, 627], [458, 582], [530, 560], [606, 608]]</point>"}]}
{"id": 4674, "image": ["rtx_frames_success_37/62_robo_set#episode_11682/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11698/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[348, 470], [380, 381], [418, 340], [524, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[351, 455], [391, 563], [454, 582], [570, 571]]</point>"}]}
{"id": 4676, "image": ["rtx_frames_success_37/62_robo_set#episode_11699/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11702/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[351, 463], [401, 440], [492, 451], [566, 519]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[359, 466], [439, 440], [494, 489], [577, 563]]</point>"}]}
{"id": 4678, "image": ["rtx_frames_success_37/62_robo_set#episode_11712/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11712/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[334, 470], [382, 463], [422, 515], [425, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug to the right."}, {"from": "gpt", "value": "<point>[[471, 601], [524, 552], [577, 545], [623, 590]]</point>"}]}
{"id": 4680, "image": ["rtx_frames_success_37/62_robo_set#episode_11722/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11727/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[376, 466], [465, 459], [515, 463], [593, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[359, 478], [422, 478], [494, 526], [587, 612]]</point>"}]}
{"id": 4682, "image": ["rtx_frames_success_37/62_robo_set#episode_11736/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11753/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle placed on a oven."}, {"from": "gpt", "value": "<point>[[346, 470], [460, 470], [566, 493], [777, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[368, 485], [420, 586], [501, 631], [593, 590]]</point>"}]}
{"id": 4684, "image": ["rtx_frames_success_37/62_robo_set#episode_11760/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11772/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 451], [418, 504], [475, 534], [566, 507]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[376, 466], [372, 556], [365, 601], [395, 657]]</point>"}]}
{"id": 4686, "image": ["rtx_frames_success_37/62_robo_set#episode_11775/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11775/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[368, 470], [448, 474], [562, 399], [646, 235]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[627, 347], [602, 451], [596, 560], [625, 634]]</point>"}]}
{"id": 4688, "image": ["rtx_frames_success_37/62_robo_set#episode_11778/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11784/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 455], [454, 448], [503, 481], [566, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[365, 463], [422, 552], [494, 552], [562, 522]]</point>"}]}
{"id": 4690, "image": ["rtx_frames_success_37/62_robo_set#episode_11787/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11788/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[356, 489], [426, 583], [511, 583], [612, 554]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[356, 470], [441, 573], [521, 601], [585, 573]]</point>"}]}
{"id": 4692, "image": ["rtx_frames_success_37/62_robo_set#episode_11797/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11803/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[363, 448], [472, 622], [570, 635], [792, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[351, 460], [388, 573], [452, 592], [537, 554]]</point>"}]}
{"id": 4694, "image": ["rtx_frames_success_37/62_robo_set#episode_11812/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11816/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana in the container."}, {"from": "gpt", "value": "<point>[[356, 498], [423, 597], [539, 603], [754, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 454], [356, 554], [363, 610], [394, 641]]</point>"}]}
{"id": 4696, "image": ["rtx_frames_success_37/62_robo_set#episode_11828/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11847/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[349, 460], [349, 554], [370, 616], [405, 635]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[345, 485], [377, 579], [454, 610], [556, 554]]</point>"}]}
{"id": 4698, "image": ["rtx_frames_success_37/62_robo_set#episode_11856/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11857/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[352, 479], [405, 579], [482, 597], [599, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[331, 460], [377, 398], [419, 373], [504, 379]]</point>"}]}
{"id": 4700, "image": ["rtx_frames_success_37/62_robo_set#episode_11864/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11874/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[338, 460], [363, 535], [426, 572], [528, 547]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[366, 485], [359, 560], [363, 616], [401, 659]]</point>"}]}
{"id": 4702, "image": ["rtx_frames_success_37/62_robo_set#episode_11874/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_11881/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[401, 635], [444, 535], [528, 504], [592, 523]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[363, 460], [394, 361], [468, 286], [532, 305]]</point>"}]}
{"id": 4704, "image": ["rtx_frames_success_37/62_robo_set#episode_11909/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11913/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[356, 473], [415, 572], [486, 585], [592, 554]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 460], [412, 554], [496, 585], [535, 547]]</point>"}]}
{"id": 4706, "image": ["rtx_frames_success_37/62_robo_set#episode_11916/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11916/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[335, 479], [349, 560], [394, 591], [440, 603]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[500, 616], [546, 585], [592, 603], [620, 659]]</point>"}]}
{"id": 4708, "image": ["rtx_frames_success_37/62_robo_set#episode_11931/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11931/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 485], [468, 498], [549, 547], [595, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[623, 448], [592, 386], [599, 286], [634, 218]]</point>"}]}
{"id": 4710, "image": ["rtx_frames_success_37/62_robo_set#episode_11932/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11935/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[366, 585], [423, 622], [504, 635], [581, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[391, 460], [458, 448], [532, 485], [595, 572]]</point>"}]}
{"id": 4712, "image": ["rtx_frames_success_37/62_robo_set#episode_11954/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11961/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[359, 479], [394, 579], [468, 579], [585, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[370, 473], [398, 261], [475, 180], [585, 168]]</point>"}]}
{"id": 4714, "image": ["rtx_frames_success_37/62_robo_set#episode_11961/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_11967/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[613, 299], [592, 442], [599, 547], [634, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[324, 460], [461, 622], [620, 647], [799, 566]]</point>"}]}
{"id": 4716, "image": ["rtx_frames_success_37/62_robo_set#episode_11969/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11975/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[363, 460], [391, 560], [461, 591], [595, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[345, 448], [352, 566], [377, 635], [408, 672]]</point>"}]}
{"id": 4718, "image": ["rtx_frames_success_37/62_robo_set#episode_11975/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_11977/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[359, 529], [345, 435], [356, 361], [373, 311]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[342, 460], [408, 603], [521, 647], [736, 597]]</point>"}]}
{"id": 4720, "image": ["rtx_frames_success_37/62_robo_set#episode_11988/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_11995/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[363, 485], [415, 610], [468, 610], [546, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[363, 491], [405, 585], [482, 610], [616, 572]]</point>"}]}
{"id": 4722, "image": ["rtx_frames_success_37/62_robo_set#episode_11996/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_12016/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana peel on the table."}, {"from": "gpt", "value": "<point>[[363, 485], [423, 603], [489, 622], [574, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 467], [461, 610], [574, 635], [715, 566]]</point>"}]}
{"id": 4724, "image": ["rtx_frames_success_37/62_robo_set#episode_12019/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_12035/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[349, 473], [412, 498], [472, 473], [511, 379]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[370, 460], [433, 585], [511, 690], [616, 666]]</point>"}]}
{"id": 4726, "image": ["rtx_frames_success_37/62_robo_set#episode_12038/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_12045/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[356, 491], [440, 610], [507, 647], [581, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[356, 454], [349, 547], [356, 622], [398, 659]]</point>"}]}
{"id": 4728, "image": ["rtx_frames_success_37/62_robo_set#episode_12045/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_12051/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[380, 498], [359, 429], [363, 342], [398, 323]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 485], [380, 585], [458, 647], [563, 622]]</point>"}]}
{"id": 4730, "image": ["rtx_frames_success_37/62_robo_set#episode_12060/frame_0.png", "rtx_frames_success_37/62_robo_set#episode_12106/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 435], [419, 541], [504, 616], [658, 653]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[370, 460], [415, 348], [475, 280], [623, 212]]</point>"}]}
{"id": 4732, "image": ["rtx_frames_success_37/62_robo_set#episode_12106/frame_15.png", "rtx_frames_success_37/62_robo_set#episode_12117/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the ketchup bottle backwards."}, {"from": "gpt", "value": "<point>[[623, 311], [599, 454], [599, 554], [620, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[342, 467], [468, 622], [616, 641], [789, 585]]</point>"}]}
{"id": 4734, "image": ["rtx_frames_success_37/62_robo_set#episode_12118/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_126/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 473], [412, 597], [486, 635], [588, 635]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[290, 534], [437, 527], [571, 512], [697, 534]]</point>"}]}
{"id": 4736, "image": ["rtx_frames_success_38/62_robo_set#episode_1217/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_1226/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[269, 505], [462, 512], [643, 534], [807, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[282, 534], [403, 512], [496, 557], [592, 631]]</point>"}]}
{"id": 4738, "image": ["rtx_frames_success_38/62_robo_set#episode_1234/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_1236/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[366, 497], [378, 631], [382, 727], [408, 817]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[311, 542], [462, 564], [571, 616], [647, 720]]</point>"}]}
{"id": 4740, "image": ["rtx_frames_success_38/62_robo_set#episode_1236/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_1240/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[601, 527], [613, 393], [605, 267], [613, 171]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[277, 482], [399, 482], [483, 497], [567, 557]]</point>"}]}
{"id": 4742, "image": ["rtx_frames_success_38/62_robo_set#episode_1248/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_1250/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[315, 497], [416, 497], [496, 505], [613, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[328, 505], [462, 505], [546, 512], [630, 601]]</point>"}]}
{"id": 4744, "image": ["rtx_frames_success_38/62_robo_set#episode_1259/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_1259/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[286, 542], [437, 431], [542, 297], [639, 104]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[613, 334], [622, 468], [622, 557], [626, 675]]</point>"}]}
{"id": 4746, "image": ["rtx_frames_success_38/62_robo_set#episode_1274/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_1297/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[340, 497], [420, 490], [521, 490], [597, 505]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[261, 512], [420, 505], [559, 505], [706, 534]]</point>"}]}
{"id": 4748, "image": ["rtx_frames_success_38/62_robo_set#episode_1299/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12134/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 497], [471, 490], [601, 505], [773, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[252, 490], [324, 557], [374, 572], [433, 646]]</point>"}]}
{"id": 4750, "image": ["rtx_frames_success_38/62_robo_set#episode_12134/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12165/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[466, 586], [529, 586], [580, 572], [651, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[303, 549], [429, 520], [571, 520], [798, 557]]</point>"}]}
{"id": 4752, "image": ["rtx_frames_success_38/62_robo_set#episode_12180/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12187/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[307, 542], [361, 564], [395, 594], [437, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 534], [391, 534], [487, 557], [584, 616]]</point>"}]}
{"id": 4754, "image": ["rtx_frames_success_38/62_robo_set#episode_12207/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12217/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[277, 505], [412, 512], [517, 512], [647, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[294, 475], [403, 497], [504, 520], [580, 586]]</point>"}]}
{"id": 4756, "image": ["rtx_frames_success_38/62_robo_set#episode_12218/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12218/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 490], [429, 386], [521, 304], [618, 156]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the ketchup bottle backward."}, {"from": "gpt", "value": "<point>[[605, 275], [605, 401], [622, 542], [630, 683]]</point>"}]}
{"id": 4758, "image": ["rtx_frames_success_38/62_robo_set#episode_12225/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12227/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 490], [437, 482], [588, 512], [786, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[311, 512], [357, 557], [387, 579], [403, 646]]</point>"}]}
{"id": 4760, "image": ["rtx_frames_success_38/62_robo_set#episode_12228/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12236/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[319, 490], [366, 542], [395, 564], [429, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[248, 520], [319, 482], [370, 438], [399, 371]]</point>"}]}
{"id": 4762, "image": ["rtx_frames_success_38/62_robo_set#episode_12236/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12243/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[353, 505], [370, 594], [403, 661], [429, 765]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[328, 534], [466, 505], [597, 512], [731, 512]]</point>"}]}
{"id": 4764, "image": ["rtx_frames_success_38/62_robo_set#episode_12244/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12248/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[336, 527], [387, 557], [429, 586], [487, 653]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[303, 468], [450, 505], [563, 505], [697, 475]]</point>"}]}
{"id": 4766, "image": ["rtx_frames_success_38/62_robo_set#episode_12251/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12254/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[311, 475], [437, 490], [529, 512], [630, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[286, 512], [387, 520], [479, 520], [622, 572]]</point>"}]}
{"id": 4768, "image": ["rtx_frames_success_38/62_robo_set#episode_12279/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12285/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[277, 527], [408, 520], [496, 520], [592, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[261, 497], [429, 505], [597, 527], [794, 594]]</point>"}]}
{"id": 4770, "image": ["rtx_frames_success_38/62_robo_set#episode_12296/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12296/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[303, 482], [403, 416], [517, 349], [634, 200]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[601, 379], [613, 512], [622, 609], [630, 735]]</point>"}]}
{"id": 4772, "image": ["rtx_frames_success_38/62_robo_set#episode_12300/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12304/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[315, 438], [416, 468], [521, 505], [613, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[277, 475], [395, 497], [479, 557], [588, 572]]</point>"}]}
{"id": 4774, "image": ["rtx_frames_success_38/62_robo_set#episode_12309/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12317/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[294, 520], [378, 445], [466, 379], [521, 297]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 557], [403, 527], [483, 527], [580, 527]]</point>"}]}
{"id": 4776, "image": ["rtx_frames_success_38/62_robo_set#episode_12321/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12322/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[303, 527], [420, 512], [538, 512], [630, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[294, 527], [399, 445], [513, 341], [622, 186]]</point>"}]}
{"id": 4778, "image": ["rtx_frames_success_38/62_robo_set#episode_12322/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12333/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[605, 327], [630, 460], [664, 572], [668, 720]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[273, 527], [445, 534], [567, 557], [647, 675]]</point>"}]}
{"id": 4780, "image": ["rtx_frames_success_38/62_robo_set#episode_12359/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12359/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[294, 482], [345, 557], [382, 616], [420, 683]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[361, 653], [361, 534], [353, 379], [391, 282]]</point>"}]}
{"id": 4782, "image": ["rtx_frames_success_38/62_robo_set#episode_12360/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12368/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[319, 534], [437, 482], [559, 520], [718, 527]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[307, 520], [378, 527], [437, 534], [437, 631]]</point>"}]}
{"id": 4784, "image": ["rtx_frames_success_38/62_robo_set#episode_12381/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12445/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[328, 482], [424, 497], [479, 497], [580, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[319, 497], [366, 564], [403, 594], [408, 690]]</point>"}]}
{"id": 4786, "image": ["rtx_frames_success_38/62_robo_set#episode_12448/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12450/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[303, 534], [416, 542], [513, 557], [618, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[307, 482], [408, 490], [487, 497], [609, 542]]</point>"}]}
{"id": 4788, "image": ["rtx_frames_success_38/62_robo_set#episode_12458/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12471/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[311, 564], [416, 505], [517, 505], [622, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[391, 282], [336, 408], [286, 482], [202, 549]]</point>"}]}
{"id": 4790, "image": ["rtx_frames_success_38/62_robo_set#episode_12481/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12495/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[307, 512], [454, 497], [584, 497], [739, 527]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 534], [420, 534], [504, 534], [567, 579]]</point>"}]}
{"id": 4792, "image": ["rtx_frames_success_38/62_robo_set#episode_12498/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12500/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[294, 490], [403, 490], [496, 512], [580, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[273, 557], [462, 431], [571, 341], [643, 260]]</point>"}]}
{"id": 4794, "image": ["rtx_frames_success_38/62_robo_set#episode_12518/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12521/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the strainer."}, {"from": "gpt", "value": "<point>[[311, 512], [391, 520], [471, 549], [609, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[252, 475], [315, 445], [349, 423], [382, 379]]</point>"}]}
{"id": 4796, "image": ["rtx_frames_success_38/62_robo_set#episode_12521/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12528/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[361, 527], [382, 631], [395, 713], [403, 802]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[277, 512], [378, 505], [483, 527], [639, 690]]</point>"}]}
{"id": 4798, "image": ["rtx_frames_success_38/62_robo_set#episode_12528/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12529/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[605, 512], [622, 386], [639, 260], [639, 141]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[311, 497], [370, 557], [399, 601], [420, 698]]</point>"}]}
{"id": 4800, "image": ["rtx_frames_success_38/62_robo_set#episode_12542/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12542/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[298, 482], [441, 520], [550, 616], [647, 705]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[597, 497], [622, 393], [622, 304], [630, 178]]</point>"}]}
{"id": 4802, "image": ["rtx_frames_success_38/62_robo_set#episode_12544/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12544/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[303, 549], [403, 468], [525, 349], [643, 163]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the ketchup bottle backward."}, {"from": "gpt", "value": "<point>[[605, 238], [613, 393], [613, 512], [622, 638]]</point>"}]}
{"id": 4804, "image": ["rtx_frames_success_38/62_robo_set#episode_12552/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12554/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[277, 490], [374, 490], [508, 505], [647, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[294, 497], [361, 557], [408, 594], [433, 646]]</point>"}]}
{"id": 4806, "image": ["rtx_frames_success_38/62_robo_set#episode_12559/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12563/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[307, 460], [437, 512], [529, 572], [634, 653]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[315, 490], [403, 468], [496, 520], [592, 609]]</point>"}]}
{"id": 4808, "image": ["rtx_frames_success_38/62_robo_set#episode_12578/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12585/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer inward."}, {"from": "gpt", "value": "<point>[[450, 386], [387, 475], [340, 542], [303, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[454, 282], [412, 393], [357, 482], [294, 594]]</point>"}]}
{"id": 4810, "image": ["rtx_frames_success_38/62_robo_set#episode_12599/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12602/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[307, 490], [395, 482], [504, 490], [571, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[294, 527], [424, 527], [580, 527], [782, 579]]</point>"}]}
{"id": 4812, "image": ["rtx_frames_success_38/62_robo_set#episode_12606/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12616/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 520], [483, 505], [622, 505], [807, 505]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[315, 520], [412, 542], [513, 572], [626, 616]]</point>"}]}
{"id": 4814, "image": ["rtx_frames_success_38/62_robo_set#episode_12631/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12632/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[311, 527], [403, 468], [471, 408], [534, 312]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[336, 505], [378, 572], [412, 631], [420, 698]]</point>"}]}
{"id": 4816, "image": ["rtx_frames_success_38/62_robo_set#episode_12635/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12638/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[307, 542], [429, 527], [529, 527], [613, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 520], [399, 497], [496, 557], [571, 609]]</point>"}]}
{"id": 4818, "image": ["rtx_frames_success_38/62_robo_set#episode_12658/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12658/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[319, 542], [437, 542], [538, 601], [647, 735]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[622, 475], [639, 349], [613, 245], [622, 141]]</point>"}]}
{"id": 4820, "image": ["rtx_frames_success_38/62_robo_set#episode_12673/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12684/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[311, 549], [403, 497], [504, 438], [634, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 512], [408, 520], [487, 520], [592, 542]]</point>"}]}
{"id": 4822, "image": ["rtx_frames_success_38/62_robo_set#episode_12701/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12707/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[332, 505], [412, 505], [475, 557], [555, 594]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[277, 453], [408, 468], [513, 512], [622, 616]]</point>"}]}
{"id": 4824, "image": ["rtx_frames_success_38/62_robo_set#episode_12708/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12712/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[315, 542], [412, 512], [500, 512], [634, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[319, 512], [420, 505], [504, 527], [588, 594]]</point>"}]}
{"id": 4826, "image": ["rtx_frames_success_38/62_robo_set#episode_12730/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12733/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[546, 334], [513, 408], [466, 490], [416, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the drawer."}, {"from": "gpt", "value": "<point>[[235, 171], [235, 267], [244, 341], [261, 497]]</point>"}]}
{"id": 4828, "image": ["rtx_frames_success_38/62_robo_set#episode_12742/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12743/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[256, 468], [361, 468], [471, 512], [571, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 497], [420, 482], [525, 512], [576, 579]]</point>"}]}
{"id": 4830, "image": ["rtx_frames_success_38/62_robo_set#episode_12753/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12754/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[273, 482], [378, 482], [496, 490], [618, 594]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[261, 497], [395, 431], [517, 304], [630, 111]]</point>"}]}
{"id": 4832, "image": ["rtx_frames_success_38/62_robo_set#episode_12754/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12760/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[613, 297], [647, 468], [655, 609], [664, 750]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[286, 490], [345, 468], [378, 438], [395, 371]]</point>"}]}
{"id": 4834, "image": ["rtx_frames_success_38/62_robo_set#episode_12760/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12771/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[378, 557], [395, 661], [395, 735], [403, 824]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[307, 542], [399, 475], [475, 408], [542, 341]]</point>"}]}
{"id": 4836, "image": ["rtx_frames_success_38/62_robo_set#episode_12775/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12777/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[303, 482], [416, 490], [500, 557], [609, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[298, 512], [403, 512], [479, 520], [559, 557]]</point>"}]}
{"id": 4838, "image": ["rtx_frames_success_38/62_robo_set#episode_12796/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12808/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[294, 505], [403, 505], [504, 505], [613, 505]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[294, 460], [387, 401], [471, 379], [546, 289]]</point>"}]}
{"id": 4840, "image": ["rtx_frames_success_38/62_robo_set#episode_12813/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12814/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[294, 475], [345, 527], [370, 579], [387, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 497], [429, 490], [580, 505], [765, 601]]</point>"}]}
{"id": 4842, "image": ["rtx_frames_success_38/62_robo_set#episode_12825/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12830/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 482], [382, 482], [479, 482], [580, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[239, 482], [294, 482], [345, 475], [391, 438]]</point>"}]}
{"id": 4844, "image": ["rtx_frames_success_38/62_robo_set#episode_12833/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12833/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[303, 497], [437, 438], [550, 341], [630, 148]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[618, 238], [622, 379], [626, 505], [626, 675]]</point>"}]}
{"id": 4846, "image": ["rtx_frames_success_38/62_robo_set#episode_12846/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12854/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[445, 334], [420, 438], [387, 512], [328, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[277, 527], [378, 534], [471, 557], [571, 586]]</point>"}]}
{"id": 4848, "image": ["rtx_frames_success_38/62_robo_set#episode_12855/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12857/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[298, 557], [399, 527], [483, 549], [571, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[315, 468], [412, 512], [529, 624], [655, 765]]</point>"}]}
{"id": 4850, "image": ["rtx_frames_success_38/62_robo_set#episode_12862/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12868/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[298, 497], [370, 542], [416, 564], [445, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[315, 520], [378, 542], [429, 542], [441, 631]]</point>"}]}
{"id": 4852, "image": ["rtx_frames_success_38/62_robo_set#episode_12881/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12881/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[303, 505], [416, 482], [538, 393], [618, 186]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[613, 379], [622, 482], [639, 616], [647, 727]]</point>"}]}
{"id": 4854, "image": ["rtx_frames_success_38/62_robo_set#episode_12890/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12895/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer closed."}, {"from": "gpt", "value": "<point>[[559, 379], [479, 482], [378, 564], [282, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[269, 505], [374, 527], [471, 520], [613, 557]]</point>"}]}
{"id": 4856, "image": ["rtx_frames_success_38/62_robo_set#episode_12907/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12914/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[231, 505], [319, 460], [382, 408], [441, 356]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[261, 497], [336, 520], [382, 527], [437, 616]]</point>"}]}
{"id": 4858, "image": ["rtx_frames_success_38/62_robo_set#episode_12927/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12927/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[294, 512], [366, 542], [412, 557], [445, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[450, 616], [538, 594], [601, 594], [672, 594]]</point>"}]}
{"id": 4860, "image": ["rtx_frames_success_38/62_robo_set#episode_12936/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12939/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[290, 512], [395, 468], [487, 423], [546, 312]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[307, 497], [403, 431], [479, 393], [534, 304]]</point>"}]}
{"id": 4862, "image": ["rtx_frames_success_38/62_robo_set#episode_12941/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12946/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[282, 527], [387, 468], [496, 401], [546, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[315, 497], [412, 497], [487, 527], [597, 586]]</point>"}]}
{"id": 4864, "image": ["rtx_frames_success_38/62_robo_set#episode_12949/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12952/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana peel on the table."}, {"from": "gpt", "value": "<point>[[315, 527], [416, 527], [504, 527], [601, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[311, 475], [353, 527], [387, 586], [429, 646]]</point>"}]}
{"id": 4866, "image": ["rtx_frames_success_38/62_robo_set#episode_12952/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12959/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the strainer from left to right."}, {"from": "gpt", "value": "<point>[[445, 616], [504, 616], [555, 601], [626, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[311, 520], [445, 505], [529, 512], [634, 572]]</point>"}]}
{"id": 4868, "image": ["rtx_frames_success_38/62_robo_set#episode_12962/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12962/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[252, 520], [311, 572], [353, 601], [416, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[420, 572], [496, 594], [555, 594], [630, 586]]</point>"}]}
{"id": 4870, "image": ["rtx_frames_success_38/62_robo_set#episode_12963/frame_0.png", "rtx_frames_success_38/62_robo_set#episode_12964/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[315, 542], [357, 557], [387, 594], [445, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[269, 497], [387, 438], [546, 364], [634, 96]]</point>"}]}
{"id": 4872, "image": ["rtx_frames_success_38/62_robo_set#episode_12964/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12989/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[609, 334], [634, 468], [643, 564], [655, 698]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[265, 512], [420, 438], [559, 364], [613, 200]]</point>"}]}
{"id": 4874, "image": ["rtx_frames_success_38/62_robo_set#episode_12989/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_12991/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[630, 438], [630, 579], [655, 646], [676, 750]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 490], [395, 505], [487, 534], [588, 631]]</point>"}]}
{"id": 4876, "image": ["rtx_frames_success_38/62_robo_set#episode_12993/frame_15.png", "rtx_frames_success_38/62_robo_set#episode_13018/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[311, 505], [378, 438], [420, 364], [445, 289]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[286, 527], [395, 423], [471, 379], [521, 327]]</point>"}]}
{"id": 4878, "image": ["rtx_frames_success_39/62_robo_set#episode_1304/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_1305/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[357, 499], [507, 465], [521, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[368, 509], [657, 734], [789, 636]]</point>"}]}
{"id": 4880, "image": ["rtx_frames_success_39/62_robo_set#episode_1317/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_1323/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[377, 602], [371, 734], [410, 837]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 509], [510, 612], [576, 573]]</point>"}]}
{"id": 4882, "image": ["rtx_frames_success_39/62_robo_set#episode_1333/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_1345/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[399, 568], [371, 705], [396, 822]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the air fryer drawer."}, {"from": "gpt", "value": "<point>[[396, 284], [291, 362], [299, 470]]</point>"}]}
{"id": 4884, "image": ["rtx_frames_success_39/62_robo_set#episode_1350/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_1366/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the microwave oven door handle."}, {"from": "gpt", "value": "<point>[[366, 499], [618, 455], [679, 392]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[355, 504], [521, 538], [571, 587]]</point>"}]}
{"id": 4886, "image": ["rtx_frames_success_39/62_robo_set#episode_1370/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_1372/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 494], [676, 392], [828, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[385, 489], [368, 612], [413, 641]]</point>"}]}
{"id": 4888, "image": ["rtx_frames_success_39/62_robo_set#episode_1389/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_1389/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 499], [407, 626], [457, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[471, 626], [571, 577], [629, 646]]</point>"}]}
{"id": 4890, "image": ["rtx_frames_success_39/62_robo_set#episode_13026/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13032/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle in the strainer."}, {"from": "gpt", "value": "<point>[[360, 480], [360, 670], [418, 700]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 509], [396, 651], [460, 626]]</point>"}]}
{"id": 4892, "image": ["rtx_frames_success_39/62_robo_set#episode_13061/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13071/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 499], [524, 641], [582, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 504], [512, 592], [579, 568]]</point>"}]}
{"id": 4894, "image": ["rtx_frames_success_39/62_robo_set#episode_13084/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13089/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 514], [551, 666], [615, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[366, 558], [537, 661], [607, 592]]</point>"}]}
{"id": 4896, "image": ["rtx_frames_success_39/62_robo_set#episode_13091/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13092/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 499], [515, 705], [582, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[410, 416], [313, 450], [280, 519]]</point>"}]}
{"id": 4898, "image": ["rtx_frames_success_39/62_robo_set#episode_13101/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13107/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 489], [626, 607], [770, 504]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[346, 499], [488, 460], [515, 347]]</point>"}]}
{"id": 4900, "image": ["rtx_frames_success_39/62_robo_set#episode_13108/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13108/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[360, 504], [454, 504], [449, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[463, 617], [560, 524], [623, 607]]</point>"}]}
{"id": 4902, "image": ["rtx_frames_success_39/62_robo_set#episode_13109/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13128/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 480], [529, 666], [571, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 509], [557, 631], [651, 661]]</point>"}]}
{"id": 4904, "image": ["rtx_frames_success_39/62_robo_set#episode_13130/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13138/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 499], [698, 382], [823, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 504], [476, 651], [587, 607]]</point>"}]}
{"id": 4906, "image": ["rtx_frames_success_39/62_robo_set#episode_13140/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13164/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[343, 504], [496, 455], [524, 294]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[352, 514], [587, 568], [657, 685]]</point>"}]}
{"id": 4908, "image": ["rtx_frames_success_39/62_robo_set#episode_13167/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13174/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 484], [526, 558], [571, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 494], [543, 279], [637, 225]]</point>"}]}
{"id": 4910, "image": ["rtx_frames_success_39/62_robo_set#episode_13174/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13189/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[620, 215], [590, 607], [634, 734], [640, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 524], [679, 675], [823, 626]]</point>"}]}
{"id": 4912, "image": ["rtx_frames_success_39/62_robo_set#episode_13193/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13204/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[435, 352], [313, 416], [316, 553]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[338, 504], [657, 705], [834, 617]]</point>"}]}
{"id": 4914, "image": ["rtx_frames_success_39/62_robo_set#episode_13210/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13217/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 494], [499, 499], [587, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 529], [479, 548], [579, 622]]</point>"}]}
{"id": 4916, "image": ["rtx_frames_success_39/62_robo_set#episode_13226/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13226/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[371, 509], [615, 303], [637, 196]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[615, 269], [571, 573], [632, 685]]</point>"}]}
{"id": 4918, "image": ["rtx_frames_success_39/62_robo_set#episode_13229/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13233/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[338, 499], [659, 489], [839, 504]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[332, 504], [465, 338], [537, 347]]</point>"}]}
{"id": 4920, "image": ["rtx_frames_success_39/62_robo_set#episode_13235/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13239/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[349, 509], [618, 465], [695, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[366, 514], [474, 465], [554, 568]]</point>"}]}
{"id": 4922, "image": ["rtx_frames_success_39/62_robo_set#episode_13244/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13251/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 499], [529, 675], [573, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[363, 494], [338, 666], [407, 670]]</point>"}]}
{"id": 4924, "image": ["rtx_frames_success_39/62_robo_set#episode_13254/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13263/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 504], [488, 617], [576, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 494], [391, 661], [463, 641]]</point>"}]}
{"id": 4926, "image": ["rtx_frames_success_39/62_robo_set#episode_13268/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13273/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[360, 494], [609, 626], [704, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 509], [490, 519], [576, 607]]</point>"}]}
{"id": 4928, "image": ["rtx_frames_success_39/62_robo_set#episode_13279/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13288/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 504], [391, 636], [438, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[363, 494], [499, 666], [568, 631]]</point>"}]}
{"id": 4930, "image": ["rtx_frames_success_39/62_robo_set#episode_13295/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13298/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[343, 494], [476, 465], [521, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[360, 489], [485, 666], [562, 636]]</point>"}]}
{"id": 4932, "image": ["rtx_frames_success_39/62_robo_set#episode_13310/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13312/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[357, 514], [463, 626], [560, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana peel on the table."}, {"from": "gpt", "value": "<point>[[366, 499], [515, 646], [576, 602]]</point>"}]}
{"id": 4934, "image": ["rtx_frames_success_39/62_robo_set#episode_13320/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13326/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[355, 519], [607, 666], [693, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[363, 514], [380, 685], [432, 710]]</point>"}]}
{"id": 4936, "image": ["rtx_frames_success_39/62_robo_set#episode_13326/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13327/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[393, 587], [421, 421], [377, 289]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[366, 514], [382, 636], [440, 631]]</point>"}]}
{"id": 4938, "image": ["rtx_frames_success_39/62_robo_set#episode_13328/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13330/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[446, 333], [360, 392], [346, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[404, 573], [360, 763], [416, 852]]</point>"}]}
{"id": 4940, "image": ["rtx_frames_success_39/62_robo_set#episode_13334/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13340/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[355, 499], [496, 475], [521, 372]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 480], [488, 612], [584, 582]]</point>"}]}
{"id": 4942, "image": ["rtx_frames_success_39/62_robo_set#episode_13356/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13356/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[352, 519], [540, 656], [601, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the oven."}, {"from": "gpt", "value": "<point>[[346, 514], [643, 426], [801, 533]]</point>"}]}
{"id": 4944, "image": ["rtx_frames_success_39/62_robo_set#episode_13360/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13366/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[368, 499], [391, 661], [465, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 514], [512, 661], [582, 587]]</point>"}]}
{"id": 4946, "image": ["rtx_frames_success_39/62_robo_set#episode_13392/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13393/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 519], [551, 670], [596, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[360, 504], [526, 710], [584, 612]]</point>"}]}
{"id": 4948, "image": ["rtx_frames_success_39/62_robo_set#episode_13397/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13401/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[357, 514], [512, 480], [532, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[360, 519], [524, 670], [596, 563]]</point>"}]}
{"id": 4950, "image": ["rtx_frames_success_39/62_robo_set#episode_13411/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13420/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[440, 372], [327, 465], [291, 651]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[338, 494], [654, 582], [693, 548]]</point>"}]}
{"id": 4952, "image": ["rtx_frames_success_39/62_robo_set#episode_13423/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13430/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[368, 480], [620, 597], [668, 700]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 514], [526, 646], [568, 597]]</point>"}]}
{"id": 4954, "image": ["rtx_frames_success_39/62_robo_set#episode_13436/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13440/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[357, 504], [521, 680], [568, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 504], [609, 303], [632, 206]]</point>"}]}
{"id": 4956, "image": ["rtx_frames_success_39/62_robo_set#episode_13440/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13466/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[618, 313], [601, 577], [637, 675]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[368, 494], [548, 685], [584, 607]]</point>"}]}
{"id": 4958, "image": ["rtx_frames_success_39/62_robo_set#episode_13477/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13489/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 489], [499, 675], [562, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[374, 499], [352, 700], [418, 734]]</point>"}]}
{"id": 4960, "image": ["rtx_frames_success_39/62_robo_set#episode_13500/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13502/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[363, 514], [629, 607], [698, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 509], [537, 670], [579, 582]]</point>"}]}
{"id": 4962, "image": ["rtx_frames_success_39/62_robo_set#episode_13512/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13541/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[357, 504], [543, 646], [598, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[330, 499], [623, 406], [740, 514]]</point>"}]}
{"id": 4964, "image": ["rtx_frames_success_39/62_robo_set#episode_13548/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13556/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[349, 504], [460, 328], [521, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 499], [515, 465], [593, 538]]</point>"}]}
{"id": 4966, "image": ["rtx_frames_success_39/62_robo_set#episode_13557/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13563/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[366, 499], [512, 690], [584, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 480], [493, 543], [587, 631]]</point>"}]}
{"id": 4968, "image": ["rtx_frames_success_39/62_robo_set#episode_13595/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13602/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 504], [540, 484], [601, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 494], [468, 533], [560, 587]]</point>"}]}
{"id": 4970, "image": ["rtx_frames_success_39/62_robo_set#episode_13609/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13610/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[349, 504], [454, 289], [529, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[357, 519], [537, 622], [582, 563]]</point>"}]}
{"id": 4972, "image": ["rtx_frames_success_39/62_robo_set#episode_13621/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13633/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[349, 524], [435, 289], [512, 308]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[360, 509], [629, 617], [701, 582]]</point>"}]}
{"id": 4974, "image": ["rtx_frames_success_39/62_robo_set#episode_13649/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13660/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[338, 494], [704, 690], [765, 553]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[457, 347], [349, 431], [343, 548]]</point>"}]}
{"id": 4976, "image": ["rtx_frames_success_39/62_robo_set#episode_13667/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13702/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 499], [521, 524], [579, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[366, 509], [402, 612], [446, 631]]</point>"}]}
{"id": 4978, "image": ["rtx_frames_success_39/62_robo_set#episode_13702/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13709/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[457, 558], [560, 529], [632, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[371, 494], [524, 631], [584, 592]]</point>"}]}
{"id": 4980, "image": ["rtx_frames_success_39/62_robo_set#episode_13719/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13721/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[368, 484], [471, 294], [535, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[357, 499], [507, 440], [540, 328]]</point>"}]}
{"id": 4982, "image": ["rtx_frames_success_39/62_robo_set#episode_13726/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13726/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[357, 494], [393, 651], [465, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer from left to right."}, {"from": "gpt", "value": "<point>[[460, 543], [576, 533], [643, 631]]</point>"}]}
{"id": 4984, "image": ["rtx_frames_success_39/62_robo_set#episode_13735/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13741/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 509], [499, 436], [535, 323]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[357, 494], [634, 470], [679, 396]]</point>"}]}
{"id": 4986, "image": ["rtx_frames_success_39/62_robo_set#episode_13742/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13750/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 509], [529, 695], [587, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 519], [526, 670], [609, 626]]</point>"}]}
{"id": 4988, "image": ["rtx_frames_success_39/62_robo_set#episode_13774/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13777/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[341, 480], [540, 622], [593, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[355, 519], [612, 372], [629, 274]]</point>"}]}
{"id": 4990, "image": ["rtx_frames_success_39/62_robo_set#episode_13777/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13783/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[643, 480], [615, 622], [654, 719]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 509], [568, 333], [568, 240]]</point>"}]}
{"id": 4992, "image": ["rtx_frames_success_39/62_robo_set#episode_13783/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13786/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[615, 396], [596, 602], [637, 680]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[357, 514], [551, 685], [587, 602]]</point>"}]}
{"id": 4994, "image": ["rtx_frames_success_39/62_robo_set#episode_13787/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13788/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 499], [565, 626], [626, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[346, 499], [537, 651], [590, 612]]</point>"}]}
{"id": 4996, "image": ["rtx_frames_success_39/62_robo_set#episode_13789/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13792/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 509], [582, 695], [770, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[474, 299], [374, 406], [366, 524]]</point>"}]}
{"id": 4998, "image": ["rtx_frames_success_39/62_robo_set#episode_13794/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13800/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[338, 499], [629, 489], [693, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 475], [609, 416], [745, 484]]</point>"}]}
{"id": 5000, "image": ["rtx_frames_success_39/62_robo_set#episode_13819/frame_15.png", "rtx_frames_success_39/62_robo_set#episode_13837/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[429, 372], [299, 460], [305, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[349, 509], [590, 387], [676, 396]]</point>"}]}
{"id": 5002, "image": ["rtx_frames_success_39/62_robo_set#episode_13850/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13856/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 519], [584, 607], [634, 670]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[343, 480], [490, 553], [571, 631]]</point>"}]}
{"id": 5004, "image": ["rtx_frames_success_39/62_robo_set#episode_13857/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13857/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[377, 494], [355, 607], [388, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[391, 524], [407, 406], [380, 323]]</point>"}]}
{"id": 5006, "image": ["rtx_frames_success_39/62_robo_set#episode_13859/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13864/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[349, 494], [515, 533], [604, 553]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[366, 499], [366, 641], [407, 641]]</point>"}]}
{"id": 5008, "image": ["rtx_frames_success_39/62_robo_set#episode_13865/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13867/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[366, 514], [532, 524], [607, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[346, 524], [535, 666], [598, 597]]</point>"}]}
{"id": 5010, "image": ["rtx_frames_success_39/62_robo_set#episode_13870/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13870/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[352, 529], [410, 612], [457, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[449, 519], [565, 494], [604, 587]]</point>"}]}
{"id": 5012, "image": ["rtx_frames_success_39/62_robo_set#episode_13871/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13882/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[357, 519], [540, 509], [596, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[355, 514], [568, 636], [643, 656]]</point>"}]}
{"id": 5014, "image": ["rtx_frames_success_39/62_robo_set#episode_13886/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13899/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 484], [554, 700], [593, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 514], [598, 357], [670, 387]]</point>"}]}
{"id": 5016, "image": ["rtx_frames_success_39/62_robo_set#episode_13902/frame_0.png", "rtx_frames_success_39/62_robo_set#episode_13902/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 519], [629, 308], [626, 225]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[618, 352], [593, 602], [645, 700]]</point>"}]}
{"id": 5018, "image": ["rtx_frames_success_39/62_robo_set#episode_13906/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_1396/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 494], [512, 602], [571, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[271, 498], [329, 385], [408, 319], [523, 296]]</point>"}]}
{"id": 5020, "image": ["rtx_frames_success_40/62_robo_set#episode_1405/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_1408/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[288, 498], [340, 412], [417, 346], [513, 306]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[508, 20], [455, 103], [429, 243], [444, 392]]</point>"}]}
{"id": 5022, "image": ["rtx_frames_success_40/62_robo_set#episode_1422/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_1430/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[258, 478], [365, 518], [457, 528], [583, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[275, 492], [374, 558], [463, 571], [553, 551]]</point>"}]}
{"id": 5024, "image": ["rtx_frames_success_40/62_robo_set#episode_1434/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_1449/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[265, 485], [361, 545], [476, 585], [575, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[303, 492], [399, 578], [493, 591], [577, 591]]</point>"}]}
{"id": 5026, "image": ["rtx_frames_success_40/62_robo_set#episode_1462/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_1462/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[285, 476], [369, 543], [471, 573], [594, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[471, 653], [466, 468], [471, 304], [529, 230]]</point>"}]}
{"id": 5028, "image": ["rtx_frames_success_40/62_robo_set#episode_1478/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_1479/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[273, 476], [372, 579], [488, 630], [637, 648]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[288, 446], [300, 549], [345, 618], [396, 657]]</point>"}]}
{"id": 5030, "image": ["rtx_frames_success_40/62_robo_set#episode_13938/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_13957/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[237, 465], [354, 554], [467, 583], [566, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the strainer."}, {"from": "gpt", "value": "<point>[[271, 472], [350, 542], [446, 585], [542, 571]]</point>"}]}
{"id": 5032, "image": ["rtx_frames_success_40/62_robo_set#episode_13969/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_13969/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[284, 472], [325, 532], [367, 565], [429, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug from left to right."}, {"from": "gpt", "value": "<point>[[427, 608], [478, 641], [545, 654], [604, 631]]</point>"}]}
{"id": 5034, "image": ["rtx_frames_success_40/62_robo_set#episode_13972/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_13973/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[267, 439], [290, 518], [329, 571], [391, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[263, 485], [299, 545], [346, 565], [395, 551]]</point>"}]}
{"id": 5036, "image": ["rtx_frames_success_40/62_robo_set#episode_13993/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_13993/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[243, 458], [301, 525], [361, 571], [429, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[453, 651], [508, 658], [564, 651], [607, 611]]</point>"}]}
{"id": 5038, "image": ["rtx_frames_success_40/62_robo_set#episode_14006/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14016/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[256, 468], [361, 548], [481, 578], [602, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[274, 478], [368, 565], [451, 585], [560, 588]]</point>"}]}
{"id": 5040, "image": ["rtx_frames_success_40/62_robo_set#episode_14045/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14058/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[291, 478], [350, 382], [425, 319], [523, 279]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[258, 468], [340, 542], [459, 598], [594, 598]]</point>"}]}
{"id": 5042, "image": ["rtx_frames_success_40/62_robo_set#episode_14070/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14075/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[282, 482], [308, 545], [365, 591], [448, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[258, 472], [322, 372], [410, 316], [517, 309]]</point>"}]}
{"id": 5044, "image": ["rtx_frames_success_40/62_robo_set#episode_14076/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14079/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[286, 478], [335, 402], [425, 359], [551, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[250, 445], [384, 522], [536, 545], [692, 505]]</point>"}]}
{"id": 5046, "image": ["rtx_frames_success_40/62_robo_set#episode_14097/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14128/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[301, 449], [312, 548], [352, 615], [406, 618]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[265, 452], [291, 545], [338, 605], [401, 671]]</point>"}]}
{"id": 5048, "image": ["rtx_frames_success_40/62_robo_set#episode_14131/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14145/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[267, 452], [359, 532], [476, 575], [602, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[265, 482], [350, 538], [444, 542], [570, 502]]</point>"}]}
{"id": 5050, "image": ["rtx_frames_success_40/62_robo_set#episode_14162/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14168/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the lid of the container."}, {"from": "gpt", "value": "<point>[[275, 465], [401, 505], [519, 475], [651, 379]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 439], [376, 528], [532, 565], [752, 581]]</point>"}]}
{"id": 5052, "image": ["rtx_frames_success_40/62_robo_set#episode_14171/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14175/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[305, 482], [361, 402], [440, 342], [527, 326]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the oven."}, {"from": "gpt", "value": "<point>[[237, 458], [382, 512], [517, 485], [671, 412]]</point>"}]}
{"id": 5054, "image": ["rtx_frames_success_40/62_robo_set#episode_14178/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14186/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[275, 452], [297, 532], [340, 578], [389, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[256, 465], [365, 542], [483, 585], [594, 585]]</point>"}]}
{"id": 5056, "image": ["rtx_frames_success_40/62_robo_set#episode_14209/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14212/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[252, 458], [354, 525], [446, 551], [545, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[280, 455], [361, 518], [453, 545], [570, 551]]</point>"}]}
{"id": 5058, "image": ["rtx_frames_success_40/62_robo_set#episode_14218/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14220/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[263, 492], [331, 405], [425, 339], [534, 292]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[244, 478], [314, 392], [406, 326], [515, 299]]</point>"}]}
{"id": 5060, "image": ["rtx_frames_success_40/62_robo_set#episode_14224/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14228/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[265, 485], [361, 555], [444, 581], [572, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[260, 462], [359, 545], [483, 568], [606, 538]]</point>"}]}
{"id": 5062, "image": ["rtx_frames_success_40/62_robo_set#episode_14229/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14241/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven flap."}, {"from": "gpt", "value": "<point>[[263, 439], [387, 468], [512, 468], [662, 419]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 452], [376, 515], [496, 545], [592, 545]]</point>"}]}
{"id": 5064, "image": ["rtx_frames_success_40/62_robo_set#episode_14245/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14247/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[267, 488], [369, 548], [463, 551], [562, 505]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[265, 478], [367, 545], [449, 565], [594, 548]]</point>"}]}
{"id": 5066, "image": ["rtx_frames_success_40/62_robo_set#episode_14270/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14275/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[271, 472], [372, 551], [470, 591], [590, 608]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[301, 458], [312, 528], [354, 591], [387, 611]]</point>"}]}
{"id": 5068, "image": ["rtx_frames_success_40/62_robo_set#episode_14275/frame_15.png", "rtx_frames_success_40/62_robo_set#episode_14278/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[370, 611], [352, 532], [350, 458], [361, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[265, 468], [363, 465], [470, 505], [568, 595]]</point>"}]}
{"id": 5070, "image": ["rtx_frames_success_40/62_robo_set#episode_14286/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14288/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[239, 462], [410, 538], [559, 575], [782, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[265, 492], [322, 412], [414, 352], [519, 322]]</point>"}]}
{"id": 5072, "image": ["rtx_frames_success_40/62_robo_set#episode_14292/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14302/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle placed on the oven."}, {"from": "gpt", "value": "<point>[[244, 478], [391, 542], [560, 558], [763, 478]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[267, 478], [387, 558], [512, 585], [632, 571]]</point>"}]}
{"id": 5074, "image": ["rtx_frames_success_40/62_robo_set#episode_14308/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14309/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[312, 485], [395, 551], [491, 581], [583, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[303, 468], [312, 551], [346, 611], [401, 664]]</point>"}]}
{"id": 5076, "image": ["rtx_frames_success_40/62_robo_set#episode_14315/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14316/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[335, 439], [416, 532], [500, 555], [604, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the tray."}, {"from": "gpt", "value": "<point>[[260, 498], [419, 571], [606, 581], [816, 571]]</point>"}]}
{"id": 5078, "image": ["rtx_frames_success_40/62_robo_set#episode_14322/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14328/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[293, 488], [365, 379], [444, 332], [536, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana in the container."}, {"from": "gpt", "value": "<point>[[290, 478], [421, 558], [583, 585], [780, 538]]</point>"}]}
{"id": 5080, "image": ["rtx_frames_success_40/62_robo_set#episode_14330/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14335/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[280, 458], [386, 532], [480, 565], [581, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 492], [346, 375], [455, 272], [592, 186]]</point>"}]}
{"id": 5082, "image": ["rtx_frames_success_40/62_robo_set#episode_14335/frame_15.png", "rtx_frames_success_40/62_robo_set#episode_14336/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[592, 302], [583, 419], [592, 518], [624, 635]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[269, 485], [368, 581], [479, 651], [622, 664]]</point>"}]}
{"id": 5084, "image": ["rtx_frames_success_40/62_robo_set#episode_14361/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14376/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 482], [395, 578], [494, 591], [615, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[305, 449], [327, 518], [365, 575], [414, 618]]</point>"}]}
{"id": 5086, "image": ["rtx_frames_success_40/62_robo_set#episode_14388/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14402/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[274, 492], [397, 561], [485, 581], [609, 588]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[282, 465], [331, 538], [374, 585], [414, 605]]</point>"}]}
{"id": 5088, "image": ["rtx_frames_success_40/62_robo_set#episode_14402/frame_15.png", "rtx_frames_success_40/62_robo_set#episode_14405/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[417, 615], [480, 618], [534, 608], [589, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[265, 478], [370, 601], [478, 668], [626, 698]]</point>"}]}
{"id": 5090, "image": ["rtx_frames_success_40/62_robo_set#episode_14406/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14408/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 478], [369, 542], [470, 558], [579, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 458], [369, 518], [453, 545], [553, 545]]</point>"}]}
{"id": 5092, "image": ["rtx_frames_success_40/62_robo_set#episode_14409/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14416/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle placed on the oven."}, {"from": "gpt", "value": "<point>[[276, 468], [433, 545], [602, 528], [820, 455]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[258, 465], [376, 551], [463, 551], [575, 551]]</point>"}]}
{"id": 5094, "image": ["rtx_frames_success_40/62_robo_set#episode_14429/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14434/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[267, 492], [365, 585], [474, 611], [587, 611]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 462], [372, 522], [474, 538], [606, 535]]</point>"}]}
{"id": 5096, "image": ["rtx_frames_success_40/62_robo_set#episode_14446/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14453/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[265, 482], [365, 561], [461, 585], [568, 581]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 472], [380, 551], [470, 585], [587, 571]]</point>"}]}
{"id": 5098, "image": ["rtx_frames_success_40/62_robo_set#episode_14473/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14477/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[267, 478], [370, 538], [466, 551], [566, 515]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[293, 475], [421, 532], [572, 561], [779, 502]]</point>"}]}
{"id": 5100, "image": ["rtx_frames_success_40/62_robo_set#episode_14479/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14487/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[243, 455], [391, 538], [560, 561], [692, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 478], [314, 542], [359, 591], [410, 615]]</point>"}]}
{"id": 5102, "image": ["rtx_frames_success_40/62_robo_set#episode_14488/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14488/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[309, 471], [330, 538], [363, 600], [389, 600]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[375, 617], [366, 500], [370, 400], [387, 317]]</point>"}]}
{"id": 5104, "image": ["rtx_frames_success_40/62_robo_set#episode_14492/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14496/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 458], [367, 542], [459, 575], [574, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[271, 446], [350, 452], [423, 401], [519, 311]]</point>"}]}
{"id": 5106, "image": ["rtx_frames_success_40/62_robo_set#episode_14498/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14504/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[276, 476], [341, 338], [457, 238], [608, 217]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[258, 461], [348, 476], [445, 431], [517, 335]]</point>"}]}
{"id": 5108, "image": ["rtx_frames_success_40/62_robo_set#episode_14526/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14527/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[261, 490], [311, 445], [345, 423], [395, 401]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[336, 460], [437, 475], [538, 505], [622, 601]]</point>"}]}
{"id": 5110, "image": ["rtx_frames_success_40/62_robo_set#episode_14529/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14535/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[319, 542], [370, 557], [399, 586], [416, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[336, 482], [412, 512], [483, 534], [597, 586]]</point>"}]}
{"id": 5112, "image": ["rtx_frames_success_40/62_robo_set#episode_14537/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14540/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[319, 497], [408, 482], [483, 505], [555, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[336, 490], [353, 557], [378, 631], [412, 690]]</point>"}]}
{"id": 5114, "image": ["rtx_frames_success_40/62_robo_set#episode_14547/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14561/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[319, 505], [374, 527], [420, 542], [437, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[412, 223], [336, 304], [286, 401], [261, 527]]</point>"}]}
{"id": 5116, "image": ["rtx_frames_success_40/62_robo_set#episode_14571/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14572/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[261, 475], [408, 475], [504, 527], [592, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[295, 482], [381, 531], [469, 555], [590, 564]]</point>"}]}
{"id": 5118, "image": ["rtx_frames_success_40/62_robo_set#episode_14579/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14599/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[273, 512], [408, 497], [508, 557], [622, 690]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[315, 475], [361, 542], [395, 594], [399, 690]]</point>"}]}
{"id": 5120, "image": ["rtx_frames_success_40/62_robo_set#episode_14603/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14604/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[277, 475], [387, 512], [483, 534], [563, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[496, 52], [429, 208], [399, 297], [366, 438]]</point>"}]}
{"id": 5122, "image": ["rtx_frames_success_40/62_robo_set#episode_14605/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14611/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 512], [391, 468], [559, 482], [723, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[286, 438], [391, 468], [487, 527], [613, 616]]</point>"}]}
{"id": 5124, "image": ["rtx_frames_success_40/62_robo_set#episode_14621/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14624/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[25, 141], [143, 186], [261, 245], [370, 349]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[299, 497], [375, 558], [464, 564], [597, 552]]</point>"}]}
{"id": 5126, "image": ["rtx_frames_success_40/62_robo_set#episode_14631/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14655/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 497], [408, 497], [500, 527], [588, 594]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the oven."}, {"from": "gpt", "value": "<point>[[273, 470], [408, 525], [556, 494], [669, 410]]</point>"}]}
{"id": 5128, "image": ["rtx_frames_success_40/62_robo_set#episode_14680/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14702/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[266, 479], [334, 371], [432, 314], [526, 320]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[247, 488], [276, 437], [312, 425], [375, 425]]</point>"}]}
{"id": 5130, "image": ["rtx_frames_success_40/62_robo_set#episode_14702/frame_15.png", "rtx_frames_success_40/62_robo_set#episode_14707/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[358, 531], [343, 615], [345, 693], [389, 820]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[282, 503], [379, 573], [466, 579], [539, 573]]</point>"}]}
{"id": 5132, "image": ["rtx_frames_success_40/62_robo_set#episode_14711/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14717/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[292, 452], [355, 359], [439, 326], [532, 329]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 470], [377, 567], [457, 591], [560, 591]]</point>"}]}
{"id": 5134, "image": ["rtx_frames_success_40/62_robo_set#episode_14717/frame_15.png", "rtx_frames_success_40/62_robo_set#episode_14718/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the ketchup bottle towards the plate."}, {"from": "gpt", "value": "<point>[[503, 618], [476, 519], [491, 374], [524, 277]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[258, 470], [323, 374], [416, 295], [526, 274]]</point>"}]}
{"id": 5136, "image": ["rtx_frames_success_40/62_robo_set#episode_14721/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14732/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 464], [358, 549], [468, 570], [597, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[276, 464], [352, 516], [461, 534], [561, 497]]</point>"}]}
{"id": 5138, "image": ["rtx_frames_success_40/62_robo_set#episode_14738/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14771/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[278, 482], [334, 389], [411, 344], [522, 323]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the lid."}, {"from": "gpt", "value": "<point>[[490, 27], [430, 109], [406, 220], [413, 380]]</point>"}]}
{"id": 5140, "image": ["rtx_frames_success_40/62_robo_set#episode_14773/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14784/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[312, 479], [384, 374], [469, 298], [604, 289]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[290, 455], [365, 555], [478, 600], [601, 594]]</point>"}]}
{"id": 5142, "image": ["rtx_frames_success_40/62_robo_set#episode_14794/frame_0.png", "rtx_frames_success_40/62_robo_set#episode_14798/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[317, 479], [341, 564], [387, 615], [454, 621]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[307, 470], [416, 573], [577, 585], [703, 528]]</point>"}]}
{"id": 5144, "image": ["rtx_frames_success_40/62_robo_set#episode_14800/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_1485/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[334, 446], [367, 350], [437, 317], [509, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[327, 460], [382, 562], [457, 654]]</point>"}]}
{"id": 5146, "image": ["rtx_frames_success_41/62_robo_set#episode_1512/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_1516/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[309, 465], [448, 516], [561, 562], [656, 669]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[271, 492], [314, 549], [367, 602], [426, 636]]</point>"}]}
{"id": 5148, "image": ["rtx_frames_success_41/62_robo_set#episode_1516/frame_15.png", "rtx_frames_success_41/62_robo_set#episode_1522/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer to the right."}, {"from": "gpt", "value": "<point>[[430, 622], [479, 629], [523, 629], [568, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[267, 485], [375, 519], [470, 555], [557, 609]]</point>"}]}
{"id": 5150, "image": ["rtx_frames_success_41/62_robo_set#episode_1524/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_1532/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[271, 482], [379, 539], [475, 572], [591, 596]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[508, 321], [466, 378], [426, 455], [398, 529]]</point>"}]}
{"id": 5152, "image": ["rtx_frames_success_41/62_robo_set#episode_1539/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_1555/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[250, 462], [360, 492], [464, 539], [562, 596]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[267, 495], [371, 505], [473, 529], [589, 586]]</point>"}]}
{"id": 5154, "image": ["rtx_frames_success_41/62_robo_set#episode_1559/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_1559/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[292, 482], [328, 559], [362, 616], [413, 666]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[369, 669], [420, 606], [491, 569], [559, 545]]</point>"}]}
{"id": 5156, "image": ["rtx_frames_success_41/62_robo_set#episode_1562/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_14819/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 468], [330, 542], [367, 596], [407, 642]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[303, 460], [410, 490], [509, 551], [607, 597], [746, 603]]</point>"}]}
{"id": 5158, "image": ["rtx_frames_success_41/62_robo_set#episode_14821/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_14826/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[312, 470], [393, 480], [483, 526], [610, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[306, 485], [428, 531], [552, 557], [668, 603], [775, 613]]</point>"}]}
{"id": 5160, "image": ["rtx_frames_success_41/62_robo_set#episode_14830/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_14839/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[312, 444], [419, 490], [523, 500], [590, 505], [679, 526]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[312, 485], [387, 408], [535, 342]]</point>"}]}
{"id": 5162, "image": ["rtx_frames_success_41/62_robo_set#episode_14841/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_14852/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[321, 470], [416, 378], [486, 322], [526, 296]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[321, 465], [416, 541], [526, 572], [610, 603]]</point>"}]}
{"id": 5164, "image": ["rtx_frames_success_41/62_robo_set#episode_14888/frame_15.png", "rtx_frames_success_41/62_robo_set#episode_14902/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[382, 255], [347, 378], [309, 480]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[332, 470], [445, 393], [512, 363], [630, 286]]</point>"}]}
{"id": 5166, "image": ["rtx_frames_success_41/62_robo_set#episode_14925/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_14942/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[295, 449], [422, 521], [494, 557], [607, 603]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[309, 444], [425, 521], [497, 546], [592, 603]]</point>"}]}
{"id": 5168, "image": ["rtx_frames_success_41/62_robo_set#episode_14943/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_14943/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[335, 470], [436, 505], [538, 536], [613, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[486, 643], [480, 511], [500, 357], [514, 271]]</point>"}]}
{"id": 5170, "image": ["rtx_frames_success_41/62_robo_set#episode_14981/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15011/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[315, 470], [408, 526], [500, 551], [607, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[312, 475], [425, 526], [486, 541], [566, 587]]</point>"}]}
{"id": 5172, "image": ["rtx_frames_success_41/62_robo_set#episode_15066/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15067/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[312, 470], [419, 511], [509, 567], [590, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 465], [486, 495], [630, 531], [728, 521]]</point>"}]}
{"id": 5174, "image": ["rtx_frames_success_41/62_robo_set#episode_15074/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15088/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[324, 470], [390, 500], [460, 531], [590, 603]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[321, 460], [405, 516], [468, 557], [535, 577], [598, 577]]</point>"}]}
{"id": 5176, "image": ["rtx_frames_success_41/62_robo_set#episode_15104/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15131/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[321, 485], [439, 531], [520, 557], [604, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[454, 337], [419, 408], [382, 485], [341, 557]]</point>"}]}
{"id": 5178, "image": ["rtx_frames_success_41/62_robo_set#episode_15135/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15151/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[267, 478], [316, 462], [362, 458], [405, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[284, 462], [316, 535], [360, 599], [409, 656]]</point>"}]}
{"id": 5180, "image": ["rtx_frames_success_41/62_robo_set#episode_15153/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15163/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[508, 23], [591, 70], [678, 164], [741, 248]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[265, 495], [436, 549], [621, 582], [816, 596]]</point>"}]}
{"id": 5182, "image": ["rtx_frames_success_41/62_robo_set#episode_15196/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15206/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[271, 502], [356, 432], [447, 365], [527, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[254, 475], [377, 539], [483, 576], [608, 596]]</point>"}]}
{"id": 5184, "image": ["rtx_frames_success_41/62_robo_set#episode_15220/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15226/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[267, 468], [379, 532], [479, 569], [595, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[273, 489], [364, 422], [438, 371], [528, 341]]</point>"}]}
{"id": 5186, "image": ["rtx_frames_success_41/62_robo_set#episode_15240/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15247/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[259, 472], [375, 512], [479, 555], [585, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[267, 472], [383, 512], [477, 555], [574, 606]]</point>"}]}
{"id": 5188, "image": ["rtx_frames_success_41/62_robo_set#episode_15259/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15268/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[267, 495], [352, 435], [439, 385], [540, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana in the container."}, {"from": "gpt", "value": "<point>[[261, 478], [436, 482], [616, 515], [828, 555]]</point>"}]}
{"id": 5190, "image": ["rtx_frames_success_41/62_robo_set#episode_15279/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15280/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[292, 465], [383, 519], [473, 562], [576, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[259, 478], [369, 535], [470, 572], [593, 596]]</point>"}]}
{"id": 5192, "image": ["rtx_frames_success_41/62_robo_set#episode_15281/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15303/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[290, 442], [322, 535], [360, 609], [413, 689]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[248, 475], [328, 519], [381, 569], [447, 636]]</point>"}]}
{"id": 5194, "image": ["rtx_frames_success_41/62_robo_set#episode_15305/frame_15.png", "rtx_frames_success_41/62_robo_set#episode_15306/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[432, 368], [390, 428], [347, 509], [311, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[277, 499], [398, 522], [498, 545], [604, 582]]</point>"}]}
{"id": 5196, "image": ["rtx_frames_success_41/62_robo_set#episode_15327/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15328/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the tray."}, {"from": "gpt", "value": "<point>[[227, 472], [443, 539], [623, 572], [856, 589]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[426, 385], [394, 428], [367, 468], [343, 529]]</point>"}]}
{"id": 5198, "image": ["rtx_frames_success_41/62_robo_set#episode_15336/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15343/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[294, 462], [403, 492], [509, 542], [602, 599]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[301, 489], [373, 415], [449, 368], [519, 341]]</point>"}]}
{"id": 5200, "image": ["rtx_frames_success_41/62_robo_set#episode_15367/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15384/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[261, 478], [367, 532], [464, 565], [566, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[267, 482], [403, 485], [536, 509], [691, 542]]</point>"}]}
{"id": 5202, "image": ["rtx_frames_success_41/62_robo_set#episode_15400/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15401/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[288, 495], [386, 532], [502, 555], [608, 562]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[280, 475], [411, 489], [542, 532], [627, 576]]</point>"}]}
{"id": 5204, "image": ["rtx_frames_success_41/62_robo_set#episode_15405/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15415/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 458], [398, 482], [508, 515], [614, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[269, 482], [390, 532], [506, 559], [627, 572]]</point>"}]}
{"id": 5206, "image": ["rtx_frames_success_41/62_robo_set#episode_15419/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15424/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[259, 462], [438, 452], [604, 465], [794, 495]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[297, 458], [333, 559], [371, 629], [411, 699]]</point>"}]}
{"id": 5208, "image": ["rtx_frames_success_41/62_robo_set#episode_15429/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15440/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[271, 462], [305, 512], [350, 559], [405, 599]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[295, 458], [324, 559], [364, 642], [409, 706]]</point>"}]}
{"id": 5210, "image": ["rtx_frames_success_41/62_robo_set#episode_15475/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15484/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[252, 468], [316, 529], [377, 582], [456, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[261, 472], [379, 509], [460, 545], [555, 612]]</point>"}]}
{"id": 5212, "image": ["rtx_frames_success_41/62_robo_set#episode_15485/frame_15.png", "rtx_frames_success_41/62_robo_set#episode_15499/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[386, 351], [354, 425], [331, 492], [316, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[290, 492], [371, 438], [468, 381], [540, 361]]</point>"}]}
{"id": 5214, "image": ["rtx_frames_success_41/62_robo_set#episode_15511/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15511/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[269, 499], [309, 482], [350, 472], [388, 482]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[381, 532], [371, 609], [373, 706], [388, 776]]</point>"}]}
{"id": 5216, "image": ["rtx_frames_success_41/62_robo_set#episode_15537/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15537/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[292, 472], [339, 519], [383, 569], [419, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the bowl across the table."}, {"from": "gpt", "value": "<point>[[417, 616], [466, 629], [511, 629], [564, 616]]</point>"}]}
{"id": 5218, "image": ["rtx_frames_success_41/62_robo_set#episode_15555/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15571/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[237, 495], [460, 458], [608, 445], [812, 448]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 458], [312, 549], [356, 629], [402, 689]]</point>"}]}
{"id": 5220, "image": ["rtx_frames_success_41/62_robo_set#episode_15581/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15588/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 489], [367, 505], [477, 529], [570, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[383, 542], [379, 649], [381, 743], [398, 816]]</point>"}]}
{"id": 5222, "image": ["rtx_frames_success_41/62_robo_set#episode_15590/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15590/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[290, 485], [409, 375], [489, 288], [625, 201]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[629, 268], [625, 438], [629, 592], [646, 733]]</point>"}]}
{"id": 5224, "image": ["rtx_frames_success_41/62_robo_set#episode_15596/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15596/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[277, 482], [316, 542], [360, 609], [400, 652]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[422, 619], [479, 602], [521, 599], [576, 606]]</point>"}]}
{"id": 5226, "image": ["rtx_frames_success_41/62_robo_set#episode_15598/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15604/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[297, 482], [400, 549], [491, 589], [587, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[292, 472], [390, 555], [483, 602], [593, 626]]</point>"}]}
{"id": 5228, "image": ["rtx_frames_success_41/62_robo_set#episode_15605/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15614/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 468], [402, 482], [511, 509], [604, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[254, 462], [464, 455], [623, 458], [837, 489]]</point>"}]}
{"id": 5230, "image": ["rtx_frames_success_41/62_robo_set#episode_15628/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15631/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[288, 478], [430, 525], [553, 549], [697, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[277, 489], [441, 505], [617, 549], [795, 602]]</point>"}]}
{"id": 5232, "image": ["rtx_frames_success_41/62_robo_set#episode_15643/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15647/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the table."}, {"from": "gpt", "value": "<point>[[259, 448], [381, 492], [487, 552], [585, 619]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 509], [411, 482], [525, 428], [623, 358]]</point>"}]}
{"id": 5234, "image": ["rtx_frames_success_41/62_robo_set#episode_15656/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15660/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the lid of the tea container."}, {"from": "gpt", "value": "<point>[[506, 20], [591, 74], [691, 187], [739, 278]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 462], [398, 509], [485, 535], [608, 549]]</point>"}]}
{"id": 5236, "image": ["rtx_frames_success_41/62_robo_set#episode_15669/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15673/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[259, 482], [367, 408], [496, 338], [619, 298]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[267, 495], [381, 495], [489, 525], [585, 589]]</point>"}]}
{"id": 5238, "image": ["rtx_frames_success_41/62_robo_set#episode_15674/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15677/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[275, 482], [362, 412], [443, 365], [530, 335]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[258, 489], [362, 512], [470, 539], [555, 579]]</point>"}]}
{"id": 5240, "image": ["rtx_frames_success_41/62_robo_set#episode_15686/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15688/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the banana on the table."}, {"from": "gpt", "value": "<point>[[309, 482], [413, 535], [517, 572], [621, 596]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[449, 355], [407, 422], [392, 468], [367, 565]]</point>"}]}
{"id": 5242, "image": ["rtx_frames_success_41/62_robo_set#episode_15689/frame_0.png", "rtx_frames_success_41/62_robo_set#episode_15689/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[288, 462], [384, 519], [483, 562], [585, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the bowl."}, {"from": "gpt", "value": "<point>[[422, 596], [553, 542], [684, 522], [831, 512]]</point>"}]}
{"id": 5244, "image": ["rtx_frames_success_42/62_robo_set#episode_160/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_1572/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 484], [463, 636], [568, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[349, 514], [512, 529], [571, 636]]</point>"}]}
{"id": 5246, "image": ["rtx_frames_success_42/62_robo_set#episode_1575/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_1577/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 509], [443, 475], [568, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 494], [726, 685], [881, 568]]</point>"}]}
{"id": 5248, "image": ["rtx_frames_success_42/62_robo_set#episode_1591/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_1592/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[319, 509], [449, 362], [529, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[343, 509], [501, 641], [601, 607]]</point>"}]}
{"id": 5250, "image": ["rtx_frames_success_42/62_robo_set#episode_1601/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_1608/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[335, 514], [540, 529], [607, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[335, 494], [476, 568], [587, 563]]</point>"}]}
{"id": 5252, "image": ["rtx_frames_success_42/62_robo_set#episode_1618/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_1619/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[355, 514], [501, 450], [515, 333]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[332, 514], [526, 661], [623, 602]]</point>"}]}
{"id": 5254, "image": ["rtx_frames_success_42/62_robo_set#episode_1636/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_1636/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[341, 529], [380, 666], [449, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[465, 592], [576, 529], [657, 675]]</point>"}]}
{"id": 5256, "image": ["rtx_frames_success_42/62_robo_set#episode_1639/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_1645/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[352, 514], [501, 475], [535, 294]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[360, 519], [626, 612], [701, 563]]</point>"}]}
{"id": 5258, "image": ["rtx_frames_success_42/62_robo_set#episode_1647/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_1647/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[346, 504], [393, 626], [457, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug to the right."}, {"from": "gpt", "value": "<point>[[460, 592], [565, 529], [626, 587]]</point>"}]}
{"id": 5260, "image": ["rtx_frames_success_42/62_robo_set#episode_1654/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15697/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[346, 494], [352, 631], [410, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 509], [529, 514], [604, 563]]</point>"}]}
{"id": 5262, "image": ["rtx_frames_success_42/62_robo_set#episode_15702/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15706/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 480], [521, 514], [593, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[335, 533], [391, 612], [440, 607]]</point>"}]}
{"id": 5264, "image": ["rtx_frames_success_42/62_robo_set#episode_15722/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15726/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[332, 504], [366, 636], [452, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 509], [496, 484], [579, 592]]</point>"}]}
{"id": 5266, "image": ["rtx_frames_success_42/62_robo_set#episode_15728/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15747/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[363, 480], [366, 612], [416, 690]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[341, 514], [377, 661], [429, 636]]</point>"}]}
{"id": 5268, "image": ["rtx_frames_success_42/62_robo_set#episode_15758/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15758/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[371, 504], [440, 333], [626, 210]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[609, 303], [579, 602], [645, 700]]</point>"}]}
{"id": 5270, "image": ["rtx_frames_success_42/62_robo_set#episode_15760/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15763/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[338, 504], [521, 470], [598, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 504], [518, 475], [632, 568]]</point>"}]}
{"id": 5272, "image": ["rtx_frames_success_42/62_robo_set#episode_15763/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_15764/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[596, 622], [596, 416], [535, 333]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[343, 509], [515, 538], [607, 622]]</point>"}]}
{"id": 5274, "image": ["rtx_frames_success_42/62_robo_set#episode_15781/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15786/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[368, 499], [490, 553], [571, 563]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 509], [488, 602], [565, 612]]</point>"}]}
{"id": 5276, "image": ["rtx_frames_success_42/62_robo_set#episode_15792/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15795/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[360, 499], [465, 641], [548, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the oven."}, {"from": "gpt", "value": "<point>[[346, 514], [560, 582], [670, 431]]</point>"}]}
{"id": 5278, "image": ["rtx_frames_success_42/62_robo_set#episode_15800/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15804/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[371, 504], [349, 646], [407, 680]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[357, 480], [343, 622], [413, 646]]</point>"}]}
{"id": 5280, "image": ["rtx_frames_success_42/62_robo_set#episode_15804/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_15812/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[407, 651], [499, 548], [573, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[313, 519], [463, 401], [521, 338]]</point>"}]}
{"id": 5282, "image": ["rtx_frames_success_42/62_robo_set#episode_15824/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15825/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[335, 524], [499, 641], [604, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 519], [582, 710], [643, 646]]</point>"}]}
{"id": 5284, "image": ["rtx_frames_success_42/62_robo_set#episode_15825/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_15840/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[637, 494], [670, 372], [657, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[335, 514], [535, 626], [601, 582]]</point>"}]}
{"id": 5286, "image": ["rtx_frames_success_42/62_robo_set#episode_15845/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15855/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 475], [626, 484], [670, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[302, 519], [540, 499], [609, 592]]</point>"}]}
{"id": 5288, "image": ["rtx_frames_success_42/62_robo_set#episode_15855/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_15856/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the strainer."}, {"from": "gpt", "value": "<point>[[573, 622], [657, 519], [643, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 519], [349, 597]]</point>"}]}
{"id": 5290, "image": ["rtx_frames_success_42/62_robo_set#episode_15856/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_15860/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the strainer from left to right."}, {"from": "gpt", "value": "<point>[[488, 538], [579, 480], [634, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 509], [576, 661], [604, 587]]</point>"}]}
{"id": 5292, "image": ["rtx_frames_success_42/62_robo_set#episode_15863/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15867/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[352, 509], [421, 685], [463, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[352, 509], [645, 440], [803, 607]]</point>"}]}
{"id": 5294, "image": ["rtx_frames_success_42/62_robo_set#episode_15884/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15890/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 484], [510, 670], [593, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[341, 504], [529, 666], [576, 622]]</point>"}]}
{"id": 5296, "image": ["rtx_frames_success_42/62_robo_set#episode_15893/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15915/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 499], [385, 680], [449, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 494], [488, 514], [582, 597]]</point>"}]}
{"id": 5298, "image": ["rtx_frames_success_42/62_robo_set#episode_15933/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15940/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[330, 509], [457, 509], [598, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[335, 484], [571, 670], [670, 670]]</point>"}]}
{"id": 5300, "image": ["rtx_frames_success_42/62_robo_set#episode_15940/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_15945/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[637, 450], [659, 343], [609, 299]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[449, 343], [288, 470], [294, 538]]</point>"}]}
{"id": 5302, "image": ["rtx_frames_success_42/62_robo_set#episode_15952/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15953/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the container lid."}, {"from": "gpt", "value": "<point>[[349, 509], [537, 392], [673, 396]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the container lid."}, {"from": "gpt", "value": "<point>[[346, 484], [579, 401], [679, 392]]</point>"}]}
{"id": 5304, "image": ["rtx_frames_success_42/62_robo_set#episode_15976/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15982/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[366, 489], [366, 700], [413, 739]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[338, 504], [543, 519], [604, 592]]</point>"}]}
{"id": 5306, "image": ["rtx_frames_success_42/62_robo_set#episode_15987/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_15997/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[349, 499], [587, 465], [701, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[382, 494], [612, 352], [629, 269]]</point>"}]}
{"id": 5308, "image": ["rtx_frames_success_42/62_robo_set#episode_15997/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16006/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[620, 470], [665, 577], [654, 700]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[288, 499], [637, 357], [803, 489]]</point>"}]}
{"id": 5310, "image": ["rtx_frames_success_42/62_robo_set#episode_16010/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16010/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[346, 509], [634, 318], [632, 206]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[609, 299], [673, 592], [648, 710]]</point>"}]}
{"id": 5312, "image": ["rtx_frames_success_42/62_robo_set#episode_16011/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16018/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[335, 519], [524, 504], [579, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[346, 509], [618, 480], [698, 587]]</point>"}]}
{"id": 5314, "image": ["rtx_frames_success_42/62_robo_set#episode_16020/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16043/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[529, 445], [341, 450], [285, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[377, 489], [371, 641], [418, 675]]</point>"}]}
{"id": 5316, "image": ["rtx_frames_success_42/62_robo_set#episode_16049/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16049/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 504], [355, 587], [382, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[388, 470], [418, 396], [385, 328]]</point>"}]}
{"id": 5318, "image": ["rtx_frames_success_42/62_robo_set#episode_16094/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16097/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 494], [501, 651], [579, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[357, 509], [471, 622], [596, 592]]</point>"}]}
{"id": 5320, "image": ["rtx_frames_success_42/62_robo_set#episode_16097/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16098/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[512, 622], [587, 440], [548, 284]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[349, 489], [612, 367], [623, 206]]</point>"}]}
{"id": 5322, "image": ["rtx_frames_success_42/62_robo_set#episode_16098/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16106/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[612, 382], [579, 577], [651, 710]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[338, 519], [479, 675], [587, 617]]</point>"}]}
{"id": 5324, "image": ["rtx_frames_success_42/62_robo_set#episode_16116/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16127/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[404, 357], [263, 411], [238, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 514], [618, 651], [776, 538]]</point>"}]}
{"id": 5326, "image": ["rtx_frames_success_42/62_robo_set#episode_16130/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16133/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 504], [521, 714], [662, 666]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 524], [488, 631], [587, 592]]</point>"}]}
{"id": 5328, "image": ["rtx_frames_success_42/62_robo_set#episode_16148/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16158/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[352, 494], [485, 465], [540, 308]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[357, 489], [346, 617], [388, 626]]</point>"}]}
{"id": 5330, "image": ["rtx_frames_success_42/62_robo_set#episode_16160/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16160/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 519], [341, 587], [380, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[377, 455], [404, 392], [374, 323]]</point>"}]}
{"id": 5332, "image": ["rtx_frames_success_42/62_robo_set#episode_16161/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16164/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[321, 519], [504, 553], [582, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[343, 514], [651, 675], [776, 602]]</point>"}]}
{"id": 5334, "image": ["rtx_frames_success_42/62_robo_set#episode_16170/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16176/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[319, 524], [482, 484], [512, 352]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 504], [357, 612], [413, 685]]</point>"}]}
{"id": 5336, "image": ["rtx_frames_success_42/62_robo_set#episode_16181/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16187/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 494], [507, 431], [537, 333]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 504], [584, 524], [651, 661]]</point>"}]}
{"id": 5338, "image": ["rtx_frames_success_42/62_robo_set#episode_16191/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16204/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[343, 499], [490, 666], [598, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[349, 538], [352, 666], [446, 622]]</point>"}]}
{"id": 5340, "image": ["rtx_frames_success_42/62_robo_set#episode_16204/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16211/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[449, 568], [557, 514], [637, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 509], [551, 440], [623, 313]]</point>"}]}
{"id": 5342, "image": ["rtx_frames_success_42/62_robo_set#episode_16211/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16215/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[632, 440], [618, 573], [645, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 484], [510, 563], [604, 563]]</point>"}]}
{"id": 5344, "image": ["rtx_frames_success_42/62_robo_set#episode_16219/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16238/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[352, 504], [596, 641], [695, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[324, 504], [587, 597], [698, 538]]</point>"}]}
{"id": 5346, "image": ["rtx_frames_success_42/62_robo_set#episode_16239/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16252/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[321, 494], [438, 558], [601, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[338, 499], [476, 299], [632, 245]]</point>"}]}
{"id": 5348, "image": ["rtx_frames_success_42/62_robo_set#episode_16252/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16259/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[626, 426], [582, 607], [657, 710]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 499], [499, 612], [596, 533]]</point>"}]}
{"id": 5350, "image": ["rtx_frames_success_42/62_robo_set#episode_16264/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16264/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[338, 524], [562, 685], [643, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[654, 465], [665, 299], [612, 250]]</point>"}]}
{"id": 5352, "image": ["rtx_frames_success_42/62_robo_set#episode_16270/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16282/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[324, 514], [374, 602], [440, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[346, 529], [590, 406], [654, 210]]</point>"}]}
{"id": 5354, "image": ["rtx_frames_success_42/62_robo_set#episode_16282/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16284/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[623, 357], [590, 548], [643, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[341, 509], [476, 470], [535, 294]]</point>"}]}
{"id": 5356, "image": ["rtx_frames_success_42/62_robo_set#episode_16293/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16299/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[341, 509], [573, 597], [704, 563]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[557, 426], [380, 436], [299, 519]]</point>"}]}
{"id": 5358, "image": ["rtx_frames_success_42/62_robo_set#episode_16300/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16304/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[346, 509], [463, 661], [551, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[319, 499], [440, 294], [540, 343]]</point>"}]}
{"id": 5360, "image": ["rtx_frames_success_42/62_robo_set#episode_16312/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16312/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[360, 509], [382, 670], [410, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the strainer across the table."}, {"from": "gpt", "value": "<point>[[446, 563], [562, 484], [632, 558]]</point>"}]}
{"id": 5362, "image": ["rtx_frames_success_42/62_robo_set#episode_16318/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16331/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[363, 524], [504, 641], [576, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[338, 524], [460, 675], [571, 602]]</point>"}]}
{"id": 5364, "image": ["rtx_frames_success_42/62_robo_set#episode_16335/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16356/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[382, 489], [368, 636], [418, 636], [413, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[366, 504], [443, 592], [576, 568]]</point>"}]}
{"id": 5366, "image": ["rtx_frames_success_42/62_robo_set#episode_16359/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16365/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 514], [562, 705], [654, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[307, 519], [496, 455], [546, 274]]</point>"}]}
{"id": 5368, "image": ["rtx_frames_success_42/62_robo_set#episode_16367/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16379/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[341, 514], [546, 666], [612, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[366, 484], [476, 646], [573, 573]]</point>"}]}
{"id": 5370, "image": ["rtx_frames_success_42/62_robo_set#episode_16383/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16392/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[360, 509], [476, 661], [612, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 494], [640, 705], [751, 577]]</point>"}]}
{"id": 5372, "image": ["rtx_frames_success_42/62_robo_set#episode_16397/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16403/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[335, 519], [526, 631], [609, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[357, 504], [460, 670], [557, 626]]</point>"}]}
{"id": 5374, "image": ["rtx_frames_success_42/62_robo_set#episode_16407/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16412/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[343, 509], [615, 470], [670, 411]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[349, 514], [604, 719], [806, 607]]</point>"}]}
{"id": 5376, "image": ["rtx_frames_success_42/62_robo_set#episode_16422/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16428/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[343, 504], [499, 631], [568, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[360, 519], [449, 299], [515, 333]]</point>"}]}
{"id": 5378, "image": ["rtx_frames_success_42/62_robo_set#episode_16429/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16436/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 504], [504, 431], [543, 308]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 504], [532, 626], [604, 568]]</point>"}]}
{"id": 5380, "image": ["rtx_frames_success_42/62_robo_set#episode_16442/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16454/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 509], [607, 484], [676, 392]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[346, 494], [604, 480], [676, 411]]</point>"}]}
{"id": 5382, "image": ["rtx_frames_success_42/62_robo_set#episode_16455/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16455/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[343, 519], [371, 631], [432, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the strainer from left to right."}, {"from": "gpt", "value": "<point>[[468, 499], [584, 475], [643, 607]]</point>"}]}
{"id": 5384, "image": ["rtx_frames_success_42/62_robo_set#episode_16460/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16464/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[349, 514], [521, 636], [601, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[338, 494], [620, 641], [698, 563]]</point>"}]}
{"id": 5386, "image": ["rtx_frames_success_42/62_robo_set#episode_16468/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16485/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[321, 524], [490, 622], [562, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[360, 509], [615, 524], [701, 582]]</point>"}]}
{"id": 5388, "image": ["rtx_frames_success_42/62_robo_set#episode_16488/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16494/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[368, 509], [612, 573], [654, 670]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 509], [490, 602], [582, 577]]</point>"}]}
{"id": 5390, "image": ["rtx_frames_success_42/62_robo_set#episode_16496/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16497/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[352, 504], [607, 617], [701, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[507, 470], [366, 465], [244, 499]]</point>"}]}
{"id": 5392, "image": ["rtx_frames_success_42/62_robo_set#episode_16507/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16517/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 489], [512, 675], [604, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[349, 519], [615, 514], [706, 582]]</point>"}]}
{"id": 5394, "image": ["rtx_frames_success_42/62_robo_set#episode_16527/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16534/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to position the strainer over the bowl."}, {"from": "gpt", "value": "<point>[[446, 568], [579, 543], [637, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[368, 504], [355, 568], [402, 617]]</point>"}]}
{"id": 5396, "image": ["rtx_frames_success_42/62_robo_set#episode_16538/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16541/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[429, 406], [316, 426], [285, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[346, 499], [518, 651], [604, 602]]</point>"}]}
{"id": 5398, "image": ["rtx_frames_success_42/62_robo_set#episode_16545/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16547/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[371, 519], [604, 504], [654, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[357, 499], [380, 641], [418, 641]]</point>"}]}
{"id": 5400, "image": ["rtx_frames_success_42/62_robo_set#episode_16547/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16549/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[385, 470], [396, 387], [374, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[363, 489], [471, 489], [532, 318]]</point>"}]}
{"id": 5402, "image": ["rtx_frames_success_42/62_robo_set#episode_16552/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16559/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 499], [623, 612], [742, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[368, 489], [346, 563], [393, 607]]</point>"}]}
{"id": 5404, "image": ["rtx_frames_success_42/62_robo_set#episode_16559/frame_15.png", "rtx_frames_success_42/62_robo_set#episode_16560/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[388, 568], [424, 480], [385, 367]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[341, 509], [493, 675], [565, 607]]</point>"}]}
{"id": 5406, "image": ["rtx_frames_success_42/62_robo_set#episode_16579/frame_0.png", "rtx_frames_success_42/62_robo_set#episode_16582/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 509], [496, 617], [573, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[346, 519], [499, 626], [582, 592]]</point>"}]}
{"id": 5408, "image": ["rtx_frames_success_43/62_robo_set#episode_1662/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_1665/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the banana."}, {"from": "gpt", "value": "<point>[[288, 465], [398, 485], [479, 529], [574, 606]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[366, 281], [360, 315], [354, 348], [352, 402]]</point>"}]}
{"id": 5410, "image": ["rtx_frames_success_43/62_robo_set#episode_1674/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_1677/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[271, 458], [388, 482], [483, 502], [593, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[284, 472], [436, 428], [608, 415], [801, 448]]</point>"}]}
{"id": 5412, "image": ["rtx_frames_success_43/62_robo_set#episode_1680/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_1700/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 482], [415, 448], [534, 425], [669, 412]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[250, 482], [415, 458], [551, 482], [695, 515]]</point>"}]}
{"id": 5414, "image": ["rtx_frames_success_43/62_robo_set#episode_1705/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_1705/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[259, 472], [305, 539], [367, 592], [436, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[438, 572], [492, 569], [555, 572], [621, 596]]</point>"}]}
{"id": 5416, "image": ["rtx_frames_success_43/62_robo_set#episode_1707/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_1723/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[275, 495], [358, 422], [441, 365], [547, 311]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[328, 504], [459, 493], [564, 508], [695, 561]]</point>"}]}
{"id": 5418, "image": ["rtx_frames_success_43/62_robo_set#episode_1731/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_1744/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[264, 481], [356, 398], [455, 337], [545, 326]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[258, 485], [418, 478], [592, 497], [796, 535]]</point>"}]}
{"id": 5420, "image": ["rtx_frames_success_43/62_robo_set#episode_16585/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16585/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[420, 646], [357, 564], [282, 497], [210, 453]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the strainer from left to right."}, {"from": "gpt", "value": "<point>[[639, 586], [555, 586], [475, 586], [403, 616]]</point>"}]}
{"id": 5422, "image": ["rtx_frames_success_43/62_robo_set#episode_16586/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16587/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[588, 586], [471, 490], [353, 468], [256, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[563, 505], [471, 505], [378, 490], [269, 468]]</point>"}]}
{"id": 5424, "image": ["rtx_frames_success_43/62_robo_set#episode_16594/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16596/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[727, 534], [571, 497], [437, 482], [265, 482]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[567, 557], [462, 542], [366, 512], [252, 497]]</point>"}]}
{"id": 5426, "image": ["rtx_frames_success_43/62_robo_set#episode_16601/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16607/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[849, 564], [571, 497], [391, 453], [218, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[412, 624], [361, 594], [311, 534], [239, 482]]</point>"}]}
{"id": 5428, "image": ["rtx_frames_success_43/62_robo_set#episode_16607/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16618/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer from left to right."}, {"from": "gpt", "value": "<point>[[630, 512], [576, 542], [521, 557], [437, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 529], [369, 415], [447, 348], [544, 308]]</point>"}]}
{"id": 5430, "image": ["rtx_frames_success_43/62_robo_set#episode_16621/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16621/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[333, 442], [352, 545], [377, 649], [419, 739]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[407, 729], [432, 616], [473, 569], [553, 529]]</point>"}]}
{"id": 5432, "image": ["rtx_frames_success_43/62_robo_set#episode_16643/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16643/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[265, 502], [328, 522], [394, 555], [453, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[464, 616], [517, 599], [568, 596], [633, 602]]</point>"}]}
{"id": 5434, "image": ["rtx_frames_success_43/62_robo_set#episode_16645/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16645/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[286, 458], [324, 515], [354, 562], [394, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[373, 579], [369, 505], [367, 425], [369, 348]]</point>"}]}
{"id": 5436, "image": ["rtx_frames_success_43/62_robo_set#episode_16646/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16653/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[282, 438], [316, 542], [360, 616], [409, 683]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[252, 475], [362, 492], [464, 532], [580, 579]]</point>"}]}
{"id": 5438, "image": ["rtx_frames_success_43/62_robo_set#episode_16658/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16660/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[259, 468], [366, 529], [472, 576], [581, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[263, 478], [309, 529], [352, 569], [409, 596]]</point>"}]}
{"id": 5440, "image": ["rtx_frames_success_43/62_robo_set#episode_16666/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16667/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[261, 489], [360, 418], [432, 371], [532, 345]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 448], [377, 475], [458, 505], [553, 552]]</point>"}]}
{"id": 5442, "image": ["rtx_frames_success_43/62_robo_set#episode_16691/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16692/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana on the plate."}, {"from": "gpt", "value": "<point>[[284, 542], [352, 452], [426, 395], [519, 348]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[290, 472], [328, 525], [377, 569], [432, 602]]</point>"}]}
{"id": 5444, "image": ["rtx_frames_success_43/62_robo_set#episode_16694/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16698/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[445, 328], [419, 381], [396, 442], [371, 519]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[282, 485], [312, 532], [352, 562], [403, 572]]</point>"}]}
{"id": 5446, "image": ["rtx_frames_success_43/62_robo_set#episode_16705/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16712/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[273, 455], [377, 489], [481, 555], [574, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[267, 442], [309, 512], [347, 555], [403, 596]]</point>"}]}
{"id": 5448, "image": ["rtx_frames_success_43/62_robo_set#episode_16718/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16721/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 465], [394, 529], [475, 569], [589, 596]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[449, 368], [403, 442], [375, 495], [339, 586]]</point>"}]}
{"id": 5450, "image": ["rtx_frames_success_43/62_robo_set#episode_16730/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16731/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the tea box lid."}, {"from": "gpt", "value": "<point>[[316, 338], [286, 381], [267, 432], [252, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[292, 478], [419, 408], [509, 335], [623, 221]]</point>"}]}
{"id": 5452, "image": ["rtx_frames_success_43/62_robo_set#episode_16731/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16746/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[614, 294], [612, 418], [623, 549], [642, 723]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 455], [386, 509], [470, 555], [568, 599]]</point>"}]}
{"id": 5454, "image": ["rtx_frames_success_43/62_robo_set#episode_16747/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16755/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[282, 478], [318, 458], [352, 455], [392, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[305, 465], [438, 418], [551, 395], [680, 385]]</point>"}]}
{"id": 5456, "image": ["rtx_frames_success_43/62_robo_set#episode_16783/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16783/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[295, 495], [400, 502], [508, 515], [604, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the oven."}, {"from": "gpt", "value": "<point>[[487, 562], [608, 532], [712, 535], [807, 569]]</point>"}]}
{"id": 5458, "image": ["rtx_frames_success_43/62_robo_set#episode_16790/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16792/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[369, 304], [345, 365], [324, 435], [316, 509]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[267, 468], [373, 482], [487, 509], [608, 569]]</point>"}]}
{"id": 5460, "image": ["rtx_frames_success_43/62_robo_set#episode_16795/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16797/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[244, 475], [411, 495], [574, 535], [703, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[286, 458], [420, 462], [544, 502], [617, 565]]</point>"}]}
{"id": 5462, "image": ["rtx_frames_success_43/62_robo_set#episode_16807/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16809/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[273, 475], [316, 552], [360, 579], [398, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[290, 472], [396, 522], [504, 545], [598, 555]]</point>"}]}
{"id": 5464, "image": ["rtx_frames_success_43/62_robo_set#episode_16809/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16815/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[528, 629], [521, 509], [521, 402], [532, 274]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 462], [379, 512], [473, 549], [585, 562]]</point>"}]}
{"id": 5466, "image": ["rtx_frames_success_43/62_robo_set#episode_16816/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16819/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[275, 482], [392, 502], [473, 535], [564, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[267, 482], [384, 502], [470, 529], [561, 596]]</point>"}]}
{"id": 5468, "image": ["rtx_frames_success_43/62_robo_set#episode_16824/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16840/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[261, 458], [373, 468], [483, 505], [619, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 492], [403, 412], [506, 331], [627, 211]]</point>"}]}
{"id": 5470, "image": ["rtx_frames_success_43/62_robo_set#episode_16840/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16842/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[621, 321], [616, 422], [612, 532], [619, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[259, 455], [400, 489], [513, 515], [617, 559]]</point>"}]}
{"id": 5472, "image": ["rtx_frames_success_43/62_robo_set#episode_16852/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16852/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[271, 442], [305, 515], [345, 582], [407, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[375, 696], [424, 629], [496, 579], [581, 562]]</point>"}]}
{"id": 5474, "image": ["rtx_frames_success_43/62_robo_set#episode_16854/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16857/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[271, 462], [369, 515], [483, 559], [593, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[491, 385], [439, 432], [396, 495], [358, 569]]</point>"}]}
{"id": 5476, "image": ["rtx_frames_success_43/62_robo_set#episode_16866/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16868/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[261, 489], [384, 539], [513, 562], [619, 559]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[485, 325], [458, 358], [434, 415], [419, 468]]</point>"}]}
{"id": 5478, "image": ["rtx_frames_success_43/62_robo_set#episode_16878/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16881/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[309, 499], [364, 428], [445, 348], [515, 294]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 455], [388, 489], [491, 542], [578, 616]]</point>"}]}
{"id": 5480, "image": ["rtx_frames_success_43/62_robo_set#episode_16882/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16883/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[286, 478], [339, 522], [381, 552], [438, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[271, 468], [409, 529], [545, 559], [699, 572]]</point>"}]}
{"id": 5482, "image": ["rtx_frames_success_43/62_robo_set#episode_16914/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16915/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[456, 338], [422, 402], [403, 482], [386, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[487, 375], [462, 428], [428, 485], [392, 542]]</point>"}]}
{"id": 5484, "image": ["rtx_frames_success_43/62_robo_set#episode_16916/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16918/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[248, 478], [411, 458], [530, 428], [676, 381]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[237, 489], [392, 462], [583, 442], [816, 468]]</point>"}]}
{"id": 5486, "image": ["rtx_frames_success_43/62_robo_set#episode_16925/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16928/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[277, 475], [330, 525], [386, 569], [430, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[282, 478], [333, 549], [367, 572], [424, 589]]</point>"}]}
{"id": 5488, "image": ["rtx_frames_success_43/62_robo_set#episode_16947/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16947/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[284, 485], [333, 458], [364, 442], [398, 455]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[373, 542], [377, 656], [384, 726], [398, 810]]</point>"}]}
{"id": 5490, "image": ["rtx_frames_success_43/62_robo_set#episode_16952/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16955/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[259, 482], [356, 422], [456, 381], [544, 355]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[273, 495], [419, 448], [544, 428], [669, 422]]</point>"}]}
{"id": 5492, "image": ["rtx_frames_success_43/62_robo_set#episode_16958/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16958/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[259, 468], [371, 445], [475, 412], [608, 361]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[629, 542], [636, 596], [636, 656], [629, 719]]</point>"}]}
{"id": 5494, "image": ["rtx_frames_success_43/62_robo_set#episode_16969/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16975/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[292, 482], [350, 545], [403, 596], [456, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[271, 455], [341, 502], [398, 555], [449, 626]]</point>"}]}
{"id": 5496, "image": ["rtx_frames_success_43/62_robo_set#episode_16975/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_16987/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[468, 619], [525, 622], [593, 606], [634, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 482], [405, 492], [498, 515], [616, 589]]</point>"}]}
{"id": 5498, "image": ["rtx_frames_success_43/62_robo_set#episode_16993/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_16994/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[282, 452], [375, 532], [462, 576], [581, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[259, 485], [424, 475], [600, 499], [818, 545]]</point>"}]}
{"id": 5500, "image": ["rtx_frames_success_43/62_robo_set#episode_16995/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17015/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[277, 478], [354, 415], [424, 375], [517, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 448], [312, 539], [350, 619], [402, 676]]</point>"}]}
{"id": 5502, "image": ["rtx_frames_success_43/62_robo_set#episode_17021/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17027/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[254, 475], [388, 492], [515, 515], [627, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[294, 489], [407, 495], [508, 505], [616, 545]]</point>"}]}
{"id": 5504, "image": ["rtx_frames_success_43/62_robo_set#episode_17032/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17034/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[259, 472], [362, 515], [475, 555], [598, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[263, 468], [381, 502], [468, 539], [581, 596]]</point>"}]}
{"id": 5506, "image": ["rtx_frames_success_43/62_robo_set#episode_17034/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_17035/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to place the banana on the oven."}, {"from": "gpt", "value": "<point>[[453, 589], [555, 565], [669, 552], [812, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[305, 455], [330, 535], [358, 609], [407, 646]]</point>"}]}
{"id": 5508, "image": ["rtx_frames_success_43/62_robo_set#episode_17038/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17039/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 455], [314, 535], [354, 582], [405, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[244, 482], [324, 495], [383, 532], [432, 589]]</point>"}]}
{"id": 5510, "image": ["rtx_frames_success_43/62_robo_set#episode_17041/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17041/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[286, 472], [402, 388], [530, 328], [642, 301]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[627, 468], [616, 555], [619, 629], [634, 699]]</point>"}]}
{"id": 5512, "image": ["rtx_frames_success_43/62_robo_set#episode_17043/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17045/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[275, 512], [347, 415], [430, 361], [544, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[267, 489], [343, 438], [428, 381], [521, 304]]</point>"}]}
{"id": 5514, "image": ["rtx_frames_success_43/62_robo_set#episode_17055/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17055/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[294, 478], [339, 549], [377, 589], [434, 619]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[451, 606], [504, 612], [551, 616], [593, 596]]</point>"}]}
{"id": 5516, "image": ["rtx_frames_success_43/62_robo_set#episode_17058/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17071/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[297, 475], [409, 482], [498, 512], [616, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[267, 468], [419, 489], [566, 522], [703, 579]]</point>"}]}
{"id": 5518, "image": ["rtx_frames_success_43/62_robo_set#episode_17072/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17086/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[271, 495], [426, 499], [600, 529], [792, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[271, 478], [309, 545], [366, 589], [430, 606]]</point>"}]}
{"id": 5520, "image": ["rtx_frames_success_43/62_robo_set#episode_17086/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_17098/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the strainer from left to right."}, {"from": "gpt", "value": "<point>[[460, 599], [517, 579], [568, 589], [608, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[278, 509], [369, 562], [466, 589], [578, 616]]</point>"}]}
{"id": 5522, "image": ["rtx_frames_success_43/62_robo_set#episode_17105/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17108/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[246, 452], [350, 512], [460, 559], [578, 589]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 435], [405, 495], [511, 525], [629, 529]]</point>"}]}
{"id": 5524, "image": ["rtx_frames_success_43/62_robo_set#episode_17114/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_17131/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[466, 348], [434, 412], [409, 478], [390, 562]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[275, 468], [405, 515], [500, 579], [633, 666]]</point>"}]}
{"id": 5526, "image": ["rtx_frames_success_43/62_robo_set#episode_17131/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_17140/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[640, 438], [640, 348], [642, 288], [638, 234]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[254, 472], [369, 515], [487, 549], [638, 565]]</point>"}]}
{"id": 5528, "image": ["rtx_frames_success_43/62_robo_set#episode_17152/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17153/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[286, 482], [419, 502], [544, 529], [638, 562]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[275, 485], [362, 412], [453, 358], [534, 325]]</point>"}]}
{"id": 5530, "image": ["rtx_frames_success_43/62_robo_set#episode_17171/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17174/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 475], [384, 475], [498, 499], [598, 559]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[269, 465], [345, 405], [434, 351], [532, 315]]</point>"}]}
{"id": 5532, "image": ["rtx_frames_success_43/62_robo_set#episode_17182/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17208/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[267, 472], [388, 485], [491, 515], [602, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[267, 430], [303, 527], [350, 607], [400, 681]]</point>"}]}
{"id": 5534, "image": ["rtx_frames_success_43/62_robo_set#episode_17210/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17228/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[288, 452], [409, 422], [530, 402], [665, 395]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 463], [386, 478], [483, 512], [601, 565]]</point>"}]}
{"id": 5536, "image": ["rtx_frames_success_43/62_robo_set#episode_17245/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17246/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[264, 478], [403, 493], [541, 519], [689, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[466, 337], [433, 394], [412, 440], [384, 531]]</point>"}]}
{"id": 5538, "image": ["rtx_frames_success_43/62_robo_set#episode_17249/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17256/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[270, 478], [386, 500], [474, 516], [592, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[270, 470], [391, 481], [498, 523], [612, 569]]</point>"}]}
{"id": 5540, "image": ["rtx_frames_success_43/62_robo_set#episode_17259/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17262/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[318, 459], [393, 387], [446, 349], [541, 318]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[253, 459], [380, 470], [476, 516], [575, 599]]</point>"}]}
{"id": 5542, "image": ["rtx_frames_success_43/62_robo_set#episode_17267/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17274/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[262, 470], [395, 512], [543, 584], [657, 644]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[279, 489], [373, 417], [448, 379], [541, 356]]</point>"}]}
{"id": 5544, "image": ["rtx_frames_success_43/62_robo_set#episode_17275/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17291/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[283, 463], [403, 500], [506, 561], [590, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[279, 451], [320, 542], [361, 618], [403, 644]]</point>"}]}
{"id": 5546, "image": ["rtx_frames_success_43/62_robo_set#episode_17291/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_17295/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[384, 569], [376, 485], [373, 425], [378, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[290, 489], [410, 512], [513, 538], [599, 584]]</point>"}]}
{"id": 5548, "image": ["rtx_frames_success_43/62_robo_set#episode_17304/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17310/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[279, 466], [429, 481], [519, 512], [620, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[315, 466], [433, 497], [502, 519], [612, 580]]</point>"}]}
{"id": 5550, "image": ["rtx_frames_success_43/62_robo_set#episode_17317/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17323/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[288, 489], [414, 444], [545, 425], [674, 413]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[397, 364], [371, 402], [350, 455], [333, 538]]</point>"}]}
{"id": 5552, "image": ["rtx_frames_success_43/62_robo_set#episode_17325/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17326/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[253, 500], [363, 402], [457, 349], [534, 311]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the container lid."}, {"from": "gpt", "value": "<point>[[290, 485], [429, 447], [541, 432], [682, 417]]</point>"}]}
{"id": 5554, "image": ["rtx_frames_success_43/62_robo_set#episode_17337/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_17348/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer closed."}, {"from": "gpt", "value": "<point>[[298, 326], [298, 394], [296, 447], [281, 512]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[296, 493], [380, 421], [438, 368], [521, 326]]</point>"}]}
{"id": 5556, "image": ["rtx_frames_success_43/62_robo_set#episode_17357/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17360/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[264, 478], [408, 485], [530, 516], [648, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[296, 485], [427, 489], [519, 523], [622, 584]]</point>"}]}
{"id": 5558, "image": ["rtx_frames_success_43/62_robo_set#episode_17365/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17365/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[260, 474], [318, 489], [378, 542], [429, 599]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug from left to right."}, {"from": "gpt", "value": "<point>[[431, 626], [479, 599], [539, 595], [603, 599]]</point>"}]}
{"id": 5560, "image": ["rtx_frames_success_43/62_robo_set#episode_17367/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_17380/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[448, 311], [425, 356], [408, 402], [406, 447]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[281, 485], [354, 527], [401, 572], [451, 629]]</point>"}]}
{"id": 5562, "image": ["rtx_frames_success_43/62_robo_set#episode_17380/frame_15.png", "rtx_frames_success_43/62_robo_set#episode_17384/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[466, 603], [521, 580], [592, 576], [648, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[270, 493], [401, 451], [549, 417], [687, 398]]</point>"}]}
{"id": 5564, "image": ["rtx_frames_success_43/62_robo_set#episode_17388/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17400/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana placed on the oven."}, {"from": "gpt", "value": "<point>[[258, 481], [418, 478], [601, 500], [798, 580]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[266, 470], [429, 497], [562, 519], [706, 569]]</point>"}]}
{"id": 5566, "image": ["rtx_frames_success_43/62_robo_set#episode_17425/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17431/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[311, 474], [410, 497], [491, 523], [562, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[498, 447], [427, 493], [350, 554], [279, 637]]</point>"}]}
{"id": 5568, "image": ["rtx_frames_success_43/62_robo_set#episode_17433/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17442/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[279, 493], [399, 493], [494, 516], [601, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[305, 463], [388, 500], [457, 523], [554, 554]]</point>"}]}
{"id": 5570, "image": ["rtx_frames_success_43/62_robo_set#episode_17447/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17450/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[285, 489], [446, 508], [584, 538], [700, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[268, 485], [330, 531], [382, 554], [416, 546]]</point>"}]}
{"id": 5572, "image": ["rtx_frames_success_43/62_robo_set#episode_17452/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17455/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[258, 481], [386, 531], [506, 554], [620, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the container lid."}, {"from": "gpt", "value": "<point>[[240, 481], [412, 432], [539, 394], [689, 402]]</point>"}]}
{"id": 5574, "image": ["rtx_frames_success_43/62_robo_set#episode_17458/frame_0.png", "rtx_frames_success_43/62_robo_set#episode_17468/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[277, 470], [397, 516], [485, 561], [579, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[270, 466], [384, 489], [481, 516], [579, 584]]</point>"}]}
{"id": 5576, "image": ["rtx_frames_success_44/62_robo_set#episode_185/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_185/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[355, 475], [518, 617], [596, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the oven."}, {"from": "gpt", "value": "<point>[[515, 563], [745, 656], [806, 631]]</point>"}]}
{"id": 5578, "image": ["rtx_frames_success_44/62_robo_set#episode_189/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_194/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[360, 519], [438, 617], [485, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[341, 514], [515, 646], [584, 646]]</point>"}]}
{"id": 5580, "image": ["rtx_frames_success_44/62_robo_set#episode_1752/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1784/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 509], [565, 504], [618, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[396, 587], [380, 734], [424, 866]]</point>"}]}
{"id": 5582, "image": ["rtx_frames_success_44/62_robo_set#episode_1788/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_1789/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[391, 347], [299, 421], [319, 524]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[468, 323], [343, 416], [338, 489]]</point>"}]}
{"id": 5584, "image": ["rtx_frames_success_44/62_robo_set#episode_1801/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1811/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana on the plate."}, {"from": "gpt", "value": "<point>[[357, 509], [524, 455], [529, 362]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strawberry."}, {"from": "gpt", "value": "<point>[[363, 489], [598, 455], [676, 392]]</point>"}]}
{"id": 5586, "image": ["rtx_frames_success_44/62_robo_set#episode_1818/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1820/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[366, 494], [548, 602], [607, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[349, 499], [560, 622], [607, 577]]</point>"}]}
{"id": 5588, "image": ["rtx_frames_success_44/62_robo_set#episode_1824/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1829/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[343, 494], [726, 690], [823, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the bottle on the plate."}, {"from": "gpt", "value": "<point>[[360, 509], [512, 406], [535, 313]]</point>"}]}
{"id": 5590, "image": ["rtx_frames_success_44/62_robo_set#episode_1829/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_1830/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bottle towards the table."}, {"from": "gpt", "value": "<point>[[529, 308], [568, 577], [634, 685]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[349, 494], [504, 401], [532, 318]]</point>"}]}
{"id": 5592, "image": ["rtx_frames_success_44/62_robo_set#episode_1860/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1860/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 499], [601, 685], [665, 685]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[637, 421], [665, 303], [620, 235]]</point>"}]}
{"id": 5594, "image": ["rtx_frames_success_44/62_robo_set#episode_1865/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1866/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[352, 529], [524, 612], [573, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[363, 499], [479, 436], [535, 352]]</point>"}]}
{"id": 5596, "image": ["rtx_frames_success_44/62_robo_set#episode_1878/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1881/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 504], [537, 646], [590, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to close the tray."}, {"from": "gpt", "value": "<point>[[343, 328], [291, 465], [302, 538]]</point>"}]}
{"id": 5598, "image": ["rtx_frames_success_44/62_robo_set#episode_1890/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_1902/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[385, 440], [404, 347], [368, 284]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[343, 484], [560, 607], [587, 543]]</point>"}]}
{"id": 5600, "image": ["rtx_frames_success_44/62_robo_set#episode_1908/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1924/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 514], [560, 602], [596, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 494], [565, 641], [604, 602]]</point>"}]}
{"id": 5602, "image": ["rtx_frames_success_44/62_robo_set#episode_1926/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_1926/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 509], [540, 631], [607, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[524, 641], [571, 318], [532, 259]]</point>"}]}
{"id": 5604, "image": ["rtx_frames_success_44/62_robo_set#episode_1928/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17475/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 494], [537, 651], [573, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[343, 509], [571, 519], [634, 631]]</point>"}]}
{"id": 5606, "image": ["rtx_frames_success_44/62_robo_set#episode_17476/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17495/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[368, 504], [479, 514], [535, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[332, 489], [629, 431], [734, 533]]</point>"}]}
{"id": 5608, "image": ["rtx_frames_success_44/62_robo_set#episode_17502/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17534/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[349, 504], [598, 475], [698, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the drawer."}, {"from": "gpt", "value": "<point>[[510, 411], [352, 426], [211, 573]]</point>"}]}
{"id": 5610, "image": ["rtx_frames_success_44/62_robo_set#episode_17546/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17551/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[338, 514], [452, 445], [560, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[349, 514], [526, 484], [598, 577]]</point>"}]}
{"id": 5612, "image": ["rtx_frames_success_44/62_robo_set#episode_17564/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17570/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[349, 519], [548, 514], [618, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[399, 597], [366, 719], [399, 837]]</point>"}]}
{"id": 5614, "image": ["rtx_frames_success_44/62_robo_set#episode_17574/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_17575/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[393, 509], [407, 396], [374, 318]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 484], [496, 489], [587, 597]]</point>"}]}
{"id": 5616, "image": ["rtx_frames_success_44/62_robo_set#episode_17576/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17583/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[357, 504], [604, 489], [795, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 499], [493, 661], [562, 631]]</point>"}]}
{"id": 5618, "image": ["rtx_frames_success_44/62_robo_set#episode_17584/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_17603/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[402, 602], [377, 724], [402, 842]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[346, 514], [587, 392], [684, 406]]</point>"}]}
{"id": 5620, "image": ["rtx_frames_success_44/62_robo_set#episode_17604/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17654/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the container lid."}, {"from": "gpt", "value": "<point>[[338, 480], [576, 382], [698, 382]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[357, 499], [399, 626], [454, 622]]</point>"}]}
{"id": 5622, "image": ["rtx_frames_success_44/62_robo_set#episode_17655/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_17661/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[402, 617], [388, 739], [427, 852]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 494], [532, 529], [590, 612]]</point>"}]}
{"id": 5624, "image": ["rtx_frames_success_44/62_robo_set#episode_17662/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17692/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[338, 499], [651, 484], [798, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[404, 592], [382, 759], [396, 852]]</point>"}]}
{"id": 5626, "image": ["rtx_frames_success_44/62_robo_set#episode_17699/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17707/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[371, 499], [571, 651], [659, 656]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 484], [512, 685], [573, 622]]</point>"}]}
{"id": 5628, "image": ["rtx_frames_success_44/62_robo_set#episode_17708/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17721/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[366, 489], [518, 431], [543, 318]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[343, 509], [704, 617], [767, 460]]</point>"}]}
{"id": 5630, "image": ["rtx_frames_success_44/62_robo_set#episode_17727/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17727/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[332, 519], [620, 333], [640, 210]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[637, 338], [598, 548], [643, 705]]</point>"}]}
{"id": 5632, "image": ["rtx_frames_success_44/62_robo_set#episode_17740/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17743/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[363, 509], [399, 626], [429, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[399, 558], [413, 421], [368, 299]]</point>"}]}
{"id": 5634, "image": ["rtx_frames_success_44/62_robo_set#episode_17748/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17765/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 519], [515, 509], [593, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[352, 504], [493, 396], [521, 245]]</point>"}]}
{"id": 5636, "image": ["rtx_frames_success_44/62_robo_set#episode_17770/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17781/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[363, 499], [496, 661], [573, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[360, 519], [476, 646], [554, 617]]</point>"}]}
{"id": 5638, "image": ["rtx_frames_success_44/62_robo_set#episode_17783/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17791/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 519], [510, 646], [604, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[332, 499], [551, 514], [604, 582]]</point>"}]}
{"id": 5640, "image": ["rtx_frames_success_44/62_robo_set#episode_17801/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17823/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[377, 494], [526, 622], [596, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[346, 524], [543, 661], [604, 607]]</point>"}]}
{"id": 5642, "image": ["rtx_frames_success_44/62_robo_set#episode_17827/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17835/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 514], [485, 631], [560, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[418, 308], [313, 392], [316, 494]]</point>"}]}
{"id": 5644, "image": ["rtx_frames_success_44/62_robo_set#episode_17842/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17843/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[338, 504], [482, 436], [537, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 519], [485, 636], [568, 607]]</point>"}]}
{"id": 5646, "image": ["rtx_frames_success_44/62_robo_set#episode_17848/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17856/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 519], [501, 617], [584, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[355, 519], [604, 318], [620, 215]]</point>"}]}
{"id": 5648, "image": ["rtx_frames_success_44/62_robo_set#episode_17856/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_17873/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[609, 313], [579, 538], [629, 714]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[391, 533], [399, 392], [368, 308]]</point>"}]}
{"id": 5650, "image": ["rtx_frames_success_44/62_robo_set#episode_17877/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_17881/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[396, 612], [388, 763], [416, 856]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 499], [507, 519], [590, 631]]</point>"}]}
{"id": 5652, "image": ["rtx_frames_success_44/62_robo_set#episode_17882/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17892/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 509], [435, 533], [457, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[341, 499], [479, 313], [532, 323]]</point>"}]}
{"id": 5654, "image": ["rtx_frames_success_44/62_robo_set#episode_17893/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_17910/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[421, 695], [388, 783], [424, 876]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[366, 524], [501, 617], [546, 592]]</point>"}]}
{"id": 5656, "image": ["rtx_frames_success_44/62_robo_set#episode_17928/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17930/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[346, 489], [471, 308], [518, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[349, 533], [598, 396], [676, 396]]</point>"}]}
{"id": 5658, "image": ["rtx_frames_success_44/62_robo_set#episode_17931/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17931/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[346, 494], [416, 563], [449, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[471, 607], [571, 543], [629, 617]]</point>"}]}
{"id": 5660, "image": ["rtx_frames_success_44/62_robo_set#episode_17939/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_17950/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 489], [529, 670], [582, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[357, 494], [512, 401], [537, 308]]</point>"}]}
{"id": 5662, "image": ["rtx_frames_success_44/62_robo_set#episode_17970/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18005/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[357, 509], [607, 377], [687, 401]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 499], [363, 666], [410, 705]]</point>"}]}
{"id": 5664, "image": ["rtx_frames_success_44/62_robo_set#episode_18005/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_18017/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[413, 700], [560, 651], [607, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[421, 338], [305, 416], [294, 509]]</point>"}]}
{"id": 5666, "image": ["rtx_frames_success_44/62_robo_set#episode_18020/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_18022/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[399, 577], [413, 406], [374, 318]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[335, 499], [460, 328], [518, 357]]</point>"}]}
{"id": 5668, "image": ["rtx_frames_success_44/62_robo_set#episode_18028/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18033/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 480], [515, 602], [576, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[335, 524], [524, 617], [579, 597]]</point>"}]}
{"id": 5670, "image": ["rtx_frames_success_44/62_robo_set#episode_18039/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18043/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[335, 499], [407, 631], [457, 666]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[360, 504], [490, 592], [551, 563]]</point>"}]}
{"id": 5672, "image": ["rtx_frames_success_44/62_robo_set#episode_18050/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18050/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[341, 514], [546, 617], [598, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the oven."}, {"from": "gpt", "value": "<point>[[537, 577], [734, 455], [825, 529]]</point>"}]}
{"id": 5674, "image": ["rtx_frames_success_44/62_robo_set#episode_18053/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18075/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the table."}, {"from": "gpt", "value": "<point>[[363, 494], [535, 548], [582, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[357, 484], [524, 543], [573, 597]]</point>"}]}
{"id": 5676, "image": ["rtx_frames_success_44/62_robo_set#episode_18091/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18094/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[374, 494], [377, 622], [416, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[380, 499], [382, 685], [413, 734]]</point>"}]}
{"id": 5678, "image": ["rtx_frames_success_44/62_robo_set#episode_18103/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18103/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[366, 489], [576, 666], [659, 670]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[645, 406], [662, 254], [623, 206]]</point>"}]}
{"id": 5680, "image": ["rtx_frames_success_44/62_robo_set#episode_18116/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18124/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[349, 524], [731, 524], [812, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[349, 504], [548, 641], [593, 617]]</point>"}]}
{"id": 5682, "image": ["rtx_frames_success_44/62_robo_set#episode_18133/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18142/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 509], [612, 705], [648, 666]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[352, 519], [526, 626], [565, 597]]</point>"}]}
{"id": 5684, "image": ["rtx_frames_success_44/62_robo_set#episode_18150/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18155/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 484], [512, 661], [573, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[416, 695], [396, 798], [416, 871]]</point>"}]}
{"id": 5686, "image": ["rtx_frames_success_44/62_robo_set#episode_18166/frame_15.png", "rtx_frames_success_44/62_robo_set#episode_18172/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[402, 617], [380, 749], [418, 866]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[346, 509], [629, 460], [684, 396]]</point>"}]}
{"id": 5688, "image": ["rtx_frames_success_44/62_robo_set#episode_18180/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18186/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the oven."}, {"from": "gpt", "value": "<point>[[346, 509], [615, 445], [681, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[402, 592], [388, 768], [402, 861]]</point>"}]}
{"id": 5690, "image": ["rtx_frames_success_44/62_robo_set#episode_18196/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18201/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 519], [551, 622], [598, 563]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[368, 504], [548, 607], [601, 582]]</point>"}]}
{"id": 5692, "image": ["rtx_frames_success_44/62_robo_set#episode_18205/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18213/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[343, 494], [695, 656], [776, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[357, 504], [546, 646], [612, 587]]</point>"}]}
{"id": 5694, "image": ["rtx_frames_success_44/62_robo_set#episode_18215/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18215/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 514], [615, 289], [609, 210]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[623, 284], [609, 592], [643, 690]]</point>"}]}
{"id": 5696, "image": ["rtx_frames_success_44/62_robo_set#episode_18238/frame_0.png", "rtx_frames_success_44/62_robo_set#episode_18238/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[363, 509], [399, 597], [449, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the strainer from left to right."}, {"from": "gpt", "value": "<point>[[452, 548], [584, 504], [645, 597]]</point>"}]}
{"id": 5698, "image": ["rtx_frames_success_44/62_robo_set#episode_18245/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_196/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 509], [526, 577], [576, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[279, 481], [333, 554], [378, 599], [421, 618]]</point>"}]}
{"id": 5700, "image": ["rtx_frames_success_45/62_robo_set#episode_196/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_206/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[423, 644], [485, 644], [543, 644], [594, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[380, 519], [374, 597], [374, 646], [389, 672]]</point>"}]}
{"id": 5702, "image": ["rtx_frames_success_45/62_robo_set#episode_217/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_221/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana on the plate."}, {"from": "gpt", "value": "<point>[[365, 466], [410, 396], [441, 362], [511, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[368, 507], [422, 601], [477, 601], [566, 575]]</point>"}]}
{"id": 5704, "image": ["rtx_frames_success_45/62_robo_set#episode_227/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_240/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[372, 455], [414, 522], [471, 571], [551, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[348, 459], [422, 575], [469, 593], [549, 549]]</point>"}]}
{"id": 5706, "image": ["rtx_frames_success_45/62_robo_set#episode_250/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_272/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[365, 474], [444, 571], [505, 604], [587, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[359, 470], [431, 493], [486, 530], [566, 634]]</point>"}]}
{"id": 5708, "image": ["rtx_frames_success_45/62_robo_set#episode_276/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_277/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[355, 470], [391, 384], [439, 343], [507, 325]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[353, 478], [395, 496], [422, 537], [435, 601]]</point>"}]}
{"id": 5710, "image": ["rtx_frames_success_45/62_robo_set#episode_1941/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_1947/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 474], [468, 455], [616, 447], [813, 474]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[270, 466], [380, 542], [470, 595], [571, 644]]</point>"}]}
{"id": 5712, "image": ["rtx_frames_success_45/62_robo_set#episode_1951/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_1971/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[253, 474], [378, 489], [483, 519], [577, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the container lid."}, {"from": "gpt", "value": "<point>[[296, 474], [421, 428], [547, 402], [682, 398]]</point>"}]}
{"id": 5714, "image": ["rtx_frames_success_45/62_robo_set#episode_1986/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_1990/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[279, 485], [406, 493], [506, 519], [605, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[252, 478], [352, 505], [470, 552], [587, 619]]</point>"}]}
{"id": 5716, "image": ["rtx_frames_success_45/62_robo_set#episode_2003/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2008/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[386, 486], [454, 495], [528, 523], [566, 556]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[401, 517], [427, 553], [478, 602], [582, 596]]</point>"}]}
{"id": 5718, "image": ["rtx_frames_success_45/62_robo_set#episode_2008/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2033/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[583, 541], [561, 480], [546, 407], [540, 310]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[368, 491], [399, 538], [417, 571], [423, 632]]</point>"}]}
{"id": 5720, "image": ["rtx_frames_success_45/62_robo_set#episode_2033/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2047/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[509, 593], [575, 564], [628, 574], [648, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[372, 466], [419, 466], [505, 499], [579, 564]]</point>"}]}
{"id": 5722, "image": ["rtx_frames_success_45/62_robo_set#episode_2050/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2050/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[393, 515], [444, 578], [524, 619], [598, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[528, 537], [513, 466], [509, 403], [532, 321]]</point>"}]}
{"id": 5724, "image": ["rtx_frames_success_45/62_robo_set#episode_2057/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2080/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[382, 513], [435, 578], [537, 618], [604, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[384, 530], [425, 575], [507, 616], [596, 601]]</point>"}]}
{"id": 5726, "image": ["rtx_frames_success_45/62_robo_set#episode_2083/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2090/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[401, 526], [448, 604], [498, 619], [583, 623]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[380, 537], [448, 668], [524, 724], [604, 728]]</point>"}]}
{"id": 5728, "image": ["rtx_frames_success_45/62_robo_set#episode_2100/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2123/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 534], [416, 608], [498, 642], [572, 657]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[368, 515], [422, 578], [486, 604], [570, 586]]</point>"}]}
{"id": 5730, "image": ["rtx_frames_success_45/62_robo_set#episode_2130/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2132/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[380, 493], [475, 526], [591, 563], [777, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[380, 507], [446, 586], [488, 586], [553, 563]]</point>"}]}
{"id": 5732, "image": ["rtx_frames_success_45/62_robo_set#episode_2140/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2151/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[389, 504], [446, 571], [490, 586], [589, 590]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[372, 537], [458, 687], [517, 705], [608, 687]]</point>"}]}
{"id": 5734, "image": ["rtx_frames_success_45/62_robo_set#episode_2151/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2164/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[648, 407], [636, 313], [642, 272], [648, 190]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[353, 519], [456, 604], [593, 604], [815, 567]]</point>"}]}
{"id": 5736, "image": ["rtx_frames_success_45/62_robo_set#episode_2178/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2181/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 526], [401, 616], [469, 627], [558, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[368, 534], [406, 683], [492, 716], [619, 698]]</point>"}]}
{"id": 5738, "image": ["rtx_frames_success_45/62_robo_set#episode_2181/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2183/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[644, 470], [623, 369], [627, 257], [653, 194]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[365, 507], [368, 560], [378, 612], [391, 623]]</point>"}]}
{"id": 5740, "image": ["rtx_frames_success_45/62_robo_set#episode_2190/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2205/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[372, 433], [406, 373], [460, 343], [496, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[363, 459], [397, 410], [452, 328], [520, 313]]</point>"}]}
{"id": 5742, "image": ["rtx_frames_success_45/62_robo_set#episode_2209/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2228/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[351, 530], [380, 608], [433, 616], [532, 593], [368, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[384, 440], [416, 366], [448, 347], [507, 332]]</point>"}]}
{"id": 5744, "image": ["rtx_frames_success_45/62_robo_set#episode_2247/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2247/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[391, 425], [431, 291], [498, 213], [568, 187]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[613, 299], [579, 485], [579, 582], [615, 690]]</point>"}]}
{"id": 5746, "image": ["rtx_frames_success_45/62_robo_set#episode_2252/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2253/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[365, 489], [410, 575], [471, 590], [562, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[346, 463], [357, 552], [368, 612], [397, 634]]</point>"}]}
{"id": 5748, "image": ["rtx_frames_success_45/62_robo_set#episode_2282/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2285/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[378, 485], [435, 582], [503, 612], [532, 575]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[368, 500], [410, 590], [501, 608], [598, 567]]</point>"}]}
{"id": 5750, "image": ["rtx_frames_success_45/62_robo_set#episode_2289/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2292/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[361, 507], [416, 616], [490, 646], [574, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[351, 455], [351, 519], [355, 582], [387, 612]]</point>"}]}
{"id": 5752, "image": ["rtx_frames_success_45/62_robo_set#episode_2300/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2304/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[348, 478], [439, 582], [539, 616], [617, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[361, 493], [393, 601], [456, 619], [536, 582]]</point>"}]}
{"id": 5754, "image": ["rtx_frames_success_45/62_robo_set#episode_2311/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2311/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[342, 485], [393, 575], [452, 604], [560, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[507, 560], [598, 612], [689, 601], [773, 526]]</point>"}]}
{"id": 5756, "image": ["rtx_frames_success_45/62_robo_set#episode_2325/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2358/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[342, 481], [382, 567], [463, 597], [541, 541]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[363, 493], [427, 575], [501, 616], [591, 590]]</point>"}]}
{"id": 5758, "image": ["rtx_frames_success_45/62_robo_set#episode_2359/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2362/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[346, 481], [397, 560], [463, 586], [587, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 463], [439, 567], [604, 545], [794, 489]]</point>"}]}
{"id": 5760, "image": ["rtx_frames_success_45/62_robo_set#episode_2379/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2385/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[340, 493], [387, 582], [465, 649], [560, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[368, 463], [389, 381], [441, 336], [513, 299]]</point>"}]}
{"id": 5762, "image": ["rtx_frames_success_45/62_robo_set#episode_2407/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2415/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[368, 478], [425, 534], [482, 560], [583, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[342, 470], [420, 590], [501, 619], [553, 556]]</point>"}]}
{"id": 5764, "image": ["rtx_frames_success_45/62_robo_set#episode_2428/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2432/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[342, 470], [393, 571], [488, 616], [587, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[359, 470], [456, 563], [583, 560], [750, 496]]</point>"}]}
{"id": 5766, "image": ["rtx_frames_success_45/62_robo_set#episode_2465/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2472/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[351, 485], [401, 578], [490, 612], [574, 590]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[342, 448], [408, 313], [492, 209], [608, 172]]</point>"}]}
{"id": 5768, "image": ["rtx_frames_success_45/62_robo_set#episode_2472/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2473/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[615, 377], [589, 549], [619, 687], [648, 713]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[348, 489], [406, 586], [460, 612], [547, 575]]</point>"}]}
{"id": 5770, "image": ["rtx_frames_success_45/62_robo_set#episode_2495/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2506/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[351, 485], [410, 560], [473, 597], [589, 575]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[351, 459], [399, 396], [463, 336], [503, 340]]</point>"}]}
{"id": 5772, "image": ["rtx_frames_success_45/62_robo_set#episode_2520/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2545/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[353, 466], [456, 582], [587, 612], [771, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[355, 500], [359, 552], [368, 627], [393, 627]]</point>"}]}
{"id": 5774, "image": ["rtx_frames_success_45/62_robo_set#episode_2545/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2546/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[448, 608], [492, 657], [560, 668], [615, 627]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[351, 451], [380, 563], [452, 616], [553, 578]]</point>"}]}
{"id": 5776, "image": ["rtx_frames_success_45/62_robo_set#episode_2551/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2556/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[348, 463], [342, 549], [346, 619], [389, 679]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[365, 451], [353, 537], [355, 597], [389, 642]]</point>"}]}
{"id": 5778, "image": ["rtx_frames_success_45/62_robo_set#episode_2557/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2569/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[368, 463], [439, 455], [498, 485], [581, 593]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[357, 470], [355, 534], [361, 604], [395, 619]]</point>"}]}
{"id": 5780, "image": ["rtx_frames_success_45/62_robo_set#episode_2569/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2587/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[471, 590], [517, 552], [577, 545], [619, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[338, 463], [363, 593], [460, 668], [598, 694]]</point>"}]}
{"id": 5782, "image": ["rtx_frames_success_45/62_robo_set#episode_2587/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2610/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[625, 448], [610, 362], [619, 243], [648, 127]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[334, 470], [389, 478], [448, 511], [536, 575]]</point>"}]}
{"id": 5784, "image": ["rtx_frames_success_45/62_robo_set#episode_2619/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2620/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 451], [408, 470], [465, 500], [532, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 474], [446, 466], [596, 485], [771, 571]]</point>"}]}
{"id": 5786, "image": ["rtx_frames_success_45/62_robo_set#episode_2632/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2636/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[355, 463], [427, 485], [530, 567], [587, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[336, 470], [376, 489], [408, 537], [431, 612]]</point>"}]}
{"id": 5788, "image": ["rtx_frames_success_45/62_robo_set#episode_2637/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2682/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 440], [416, 440], [520, 493], [591, 560]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[346, 425], [410, 328], [503, 239], [608, 220]]</point>"}]}
{"id": 5790, "image": ["rtx_frames_success_45/62_robo_set#episode_2682/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2683/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[627, 287], [606, 414], [600, 552], [640, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[353, 455], [389, 496], [414, 560], [422, 616]]</point>"}]}
{"id": 5792, "image": ["rtx_frames_success_45/62_robo_set#episode_2683/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2686/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[488, 586], [536, 541], [610, 560], [663, 604]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[355, 485], [439, 478], [505, 515], [600, 597]]</point>"}]}
{"id": 5794, "image": ["rtx_frames_success_45/62_robo_set#episode_2697/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2706/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[329, 470], [391, 478], [452, 515], [558, 627]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[336, 470], [401, 455], [475, 481], [570, 575]]</point>"}]}
{"id": 5796, "image": ["rtx_frames_success_45/62_robo_set#episode_2713/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2735/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 466], [427, 440], [463, 474], [562, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[359, 463], [351, 552], [368, 612], [397, 679]]</point>"}]}
{"id": 5798, "image": ["rtx_frames_success_45/62_robo_set#episode_2735/frame_15.png", "rtx_frames_success_45/62_robo_set#episode_2744/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[382, 504], [351, 403], [348, 343], [368, 284]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[355, 470], [498, 463], [619, 515], [769, 619]]</point>"}]}
{"id": 5800, "image": ["rtx_frames_success_45/62_robo_set#episode_2746/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2763/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[351, 448], [418, 437], [498, 481], [553, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[351, 466], [410, 597], [503, 638], [579, 623]]</point>"}]}
{"id": 5802, "image": ["rtx_frames_success_45/62_robo_set#episode_2766/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2772/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[342, 455], [372, 373], [425, 336], [501, 328]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 455], [422, 455], [469, 478], [553, 571]]</point>"}]}
{"id": 5804, "image": ["rtx_frames_success_45/62_robo_set#episode_2784/frame_0.png", "rtx_frames_success_45/62_robo_set#episode_2799/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 474], [407, 463], [475, 493], [565, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[342, 448], [437, 470], [492, 511], [583, 612]]</point>"}]}
{"id": 5806, "image": ["rtx_frames_success_45/62_robo_set#episode_2800/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_288/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[355, 440], [346, 537], [357, 612], [397, 679]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the croissant."}, {"from": "gpt", "value": "<point>[[265, 468], [383, 522], [500, 562], [593, 576]]</point>"}]}
{"id": 5808, "image": ["rtx_frames_success_46/62_robo_set#episode_294/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_300/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[278, 489], [466, 478], [627, 495], [794, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 472], [384, 532], [487, 576], [578, 592]]</point>"}]}
{"id": 5810, "image": ["rtx_frames_success_46/62_robo_set#episode_318/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_327/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 468], [424, 428], [538, 418], [674, 415]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[381, 532], [381, 622], [386, 689], [403, 783]]</point>"}]}
{"id": 5812, "image": ["rtx_frames_success_46/62_robo_set#episode_329/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_333/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[303, 445], [343, 555], [381, 632], [426, 696]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[250, 478], [384, 502], [491, 529], [585, 582]]</point>"}]}
{"id": 5814, "image": ["rtx_frames_success_46/62_robo_set#episode_356/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_364/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[527, 455], [445, 529], [366, 576], [248, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backward."}, {"from": "gpt", "value": "<point>[[377, 519], [379, 622], [383, 703], [394, 796]]</point>"}]}
{"id": 5816, "image": ["rtx_frames_success_46/62_robo_set#episode_2836/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_2836/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 468], [314, 549], [369, 622], [415, 666]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[369, 646], [426, 589], [521, 562], [581, 565]]</point>"}]}
{"id": 5818, "image": ["rtx_frames_success_46/62_robo_set#episode_2845/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_2857/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[259, 485], [398, 448], [536, 425], [676, 412]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[271, 502], [333, 525], [409, 576], [453, 642]]</point>"}]}
{"id": 5820, "image": ["rtx_frames_success_46/62_robo_set#episode_2857/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_2870/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[509, 639], [566, 626], [617, 629], [659, 639]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[290, 462], [316, 535], [348, 586], [400, 626]]</point>"}]}
{"id": 5822, "image": ["rtx_frames_success_46/62_robo_set#episode_2873/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_2886/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[259, 478], [384, 519], [494, 576], [627, 662]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[254, 468], [379, 489], [487, 529], [562, 589]]</point>"}]}
{"id": 5824, "image": ["rtx_frames_success_46/62_robo_set#episode_2892/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_2897/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[267, 482], [407, 502], [498, 522], [623, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[267, 435], [381, 478], [472, 519], [570, 569]]</point>"}]}
{"id": 5826, "image": ["rtx_frames_success_46/62_robo_set#episode_2900/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_2924/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[261, 489], [413, 505], [540, 525], [706, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[295, 482], [335, 529], [377, 565], [428, 576]]</point>"}]}
{"id": 5828, "image": ["rtx_frames_success_46/62_robo_set#episode_2924/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_2936/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the strainer from left to right."}, {"from": "gpt", "value": "<point>[[447, 616], [498, 602], [538, 599], [597, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[277, 482], [394, 545], [481, 589], [593, 629]]</point>"}]}
{"id": 5830, "image": ["rtx_frames_success_46/62_robo_set#episode_2943/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_2945/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[263, 458], [384, 509], [479, 549], [572, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[277, 478], [390, 545], [477, 589], [585, 622]]</point>"}]}
{"id": 5832, "image": ["rtx_frames_success_46/62_robo_set#episode_2946/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_2951/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 478], [392, 549], [489, 586], [593, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 482], [483, 485], [627, 495], [805, 515]]</point>"}]}
{"id": 5834, "image": ["rtx_frames_success_46/62_robo_set#episode_2953/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_2962/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[381, 522], [381, 616], [381, 703], [384, 776]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[307, 455], [381, 509], [468, 549], [564, 576]]</point>"}]}
{"id": 5836, "image": ["rtx_frames_success_46/62_robo_set#episode_2976/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_2980/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[502, 341], [477, 418], [447, 482], [407, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[271, 499], [347, 415], [428, 365], [528, 328]]</point>"}]}
{"id": 5838, "image": ["rtx_frames_success_46/62_robo_set#episode_2982/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_2986/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[294, 468], [335, 542], [362, 576], [417, 599]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 482], [383, 522], [491, 562], [597, 616]]</point>"}]}
{"id": 5840, "image": ["rtx_frames_success_46/62_robo_set#episode_3009/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3010/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[261, 468], [398, 478], [523, 515], [623, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[273, 468], [337, 499], [392, 542], [449, 612]]</point>"}]}
{"id": 5842, "image": ["rtx_frames_success_46/62_robo_set#episode_3010/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3011/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug to the right."}, {"from": "gpt", "value": "<point>[[473, 629], [536, 599], [597, 586], [642, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[284, 472], [419, 499], [506, 525], [616, 565]]</point>"}]}
{"id": 5844, "image": ["rtx_frames_success_46/62_robo_set#episode_3016/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3024/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 462], [381, 475], [485, 502], [585, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 465], [377, 381], [453, 348], [542, 338]]</point>"}]}
{"id": 5846, "image": ["rtx_frames_success_46/62_robo_set#episode_3043/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3051/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 485], [383, 529], [460, 552], [564, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[290, 475], [422, 535], [527, 596], [650, 676]]</point>"}]}
{"id": 5848, "image": ["rtx_frames_success_46/62_robo_set#episode_3051/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3058/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[634, 549], [642, 432], [642, 355], [634, 248]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green strainer."}, {"from": "gpt", "value": "<point>[[381, 555], [381, 646], [384, 713], [396, 793]]</point>"}]}
{"id": 5850, "image": ["rtx_frames_success_46/62_robo_set#episode_3062/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3080/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[267, 455], [352, 388], [434, 348], [519, 321]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[244, 468], [438, 489], [625, 532], [837, 596]]</point>"}]}
{"id": 5852, "image": ["rtx_frames_success_46/62_robo_set#episode_3091/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3092/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[379, 549], [379, 659], [381, 749], [392, 830]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[242, 475], [345, 502], [464, 539], [570, 582]]</point>"}]}
{"id": 5854, "image": ["rtx_frames_success_46/62_robo_set#episode_3095/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3112/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 435], [367, 499], [470, 552], [574, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[259, 475], [316, 515], [373, 562], [445, 599]]</point>"}]}
{"id": 5856, "image": ["rtx_frames_success_46/62_robo_set#episode_3116/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3120/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[252, 468], [379, 482], [473, 502], [600, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[290, 482], [381, 428], [460, 388], [513, 338]]</point>"}]}
{"id": 5858, "image": ["rtx_frames_success_46/62_robo_set#episode_3129/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3137/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the oven."}, {"from": "gpt", "value": "<point>[[246, 489], [392, 448], [523, 422], [670, 388]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[282, 485], [369, 405], [449, 355], [544, 318]]</point>"}]}
{"id": 5860, "image": ["rtx_frames_success_46/62_robo_set#episode_3138/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3141/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana on the tray."}, {"from": "gpt", "value": "<point>[[250, 452], [464, 475], [646, 522], [839, 589]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the handle of the oven."}, {"from": "gpt", "value": "<point>[[278, 462], [413, 485], [557, 519], [703, 579]]</point>"}]}
{"id": 5862, "image": ["rtx_frames_success_46/62_robo_set#episode_3142/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3147/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[271, 515], [415, 472], [544, 448], [689, 428]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[250, 462], [364, 478], [468, 515], [580, 565]]</point>"}]}
{"id": 5864, "image": ["rtx_frames_success_46/62_robo_set#episode_3148/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3156/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[468, 315], [441, 365], [428, 405], [409, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[275, 489], [377, 402], [434, 361], [528, 301]]</point>"}]}
{"id": 5866, "image": ["rtx_frames_success_46/62_robo_set#episode_3168/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3176/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[292, 472], [320, 535], [358, 586], [407, 639]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[263, 495], [388, 512], [479, 532], [559, 582]]</point>"}]}
{"id": 5868, "image": ["rtx_frames_success_46/62_robo_set#episode_3182/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3199/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[250, 455], [366, 489], [475, 522], [566, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[271, 455], [307, 525], [330, 562], [384, 592]]</point>"}]}
{"id": 5870, "image": ["rtx_frames_success_46/62_robo_set#episode_3225/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3236/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[241, 472], [415, 492], [547, 515], [697, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 462], [375, 519], [460, 542], [570, 569]]</point>"}]}
{"id": 5872, "image": ["rtx_frames_success_46/62_robo_set#episode_3250/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3261/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[256, 478], [388, 529], [519, 592], [646, 659]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[294, 472], [352, 502], [413, 549], [458, 602]]</point>"}]}
{"id": 5874, "image": ["rtx_frames_success_46/62_robo_set#episode_3261/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3268/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[472, 629], [547, 612], [604, 606], [659, 609]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[449, 341], [424, 412], [400, 468], [358, 535]]</point>"}]}
{"id": 5876, "image": ["rtx_frames_success_46/62_robo_set#episode_3280/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3283/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[241, 495], [438, 495], [610, 505], [835, 509]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[273, 475], [424, 539], [555, 565], [697, 569]]</point>"}]}
{"id": 5878, "image": ["rtx_frames_success_46/62_robo_set#episode_3284/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3285/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[263, 462], [403, 482], [504, 502], [623, 559]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[250, 475], [400, 529], [494, 572], [610, 636]]</point>"}]}
{"id": 5880, "image": ["rtx_frames_success_46/62_robo_set#episode_3312/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3312/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[278, 485], [347, 515], [398, 559], [449, 619]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[449, 646], [513, 606], [589, 582], [646, 586]]</point>"}]}
{"id": 5882, "image": ["rtx_frames_success_46/62_robo_set#episode_3321/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3361/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[267, 468], [392, 505], [491, 532], [604, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the flap."}, {"from": "gpt", "value": "<point>[[244, 489], [386, 452], [525, 435], [672, 408]]</point>"}]}
{"id": 5884, "image": ["rtx_frames_success_46/62_robo_set#episode_3366/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3372/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[290, 489], [426, 485], [545, 502], [695, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[267, 458], [384, 492], [485, 529], [593, 582]]</point>"}]}
{"id": 5886, "image": ["rtx_frames_success_46/62_robo_set#episode_3375/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3390/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[261, 468], [373, 495], [458, 515], [570, 549]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the banana."}, {"from": "gpt", "value": "<point>[[267, 499], [392, 545], [477, 562], [597, 549]]</point>"}]}
{"id": 5888, "image": ["rtx_frames_success_46/62_robo_set#episode_3391/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3405/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[294, 475], [383, 442], [456, 408], [536, 338]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[258, 462], [377, 492], [472, 529], [568, 579]]</point>"}]}
{"id": 5890, "image": ["rtx_frames_success_46/62_robo_set#episode_3410/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3419/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 482], [413, 519], [500, 545], [612, 599]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer closed."}, {"from": "gpt", "value": "<point>[[398, 435], [377, 468], [358, 502], [331, 569]]</point>"}]}
{"id": 5892, "image": ["rtx_frames_success_46/62_robo_set#episode_3421/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3424/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[256, 495], [373, 505], [460, 529], [570, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[280, 468], [396, 509], [485, 549], [593, 616]]</point>"}]}
{"id": 5894, "image": ["rtx_frames_success_46/62_robo_set#episode_3426/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3453/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[267, 455], [305, 549], [347, 629], [402, 689]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[265, 472], [373, 509], [468, 539], [578, 542]]</point>"}]}
{"id": 5896, "image": ["rtx_frames_success_46/62_robo_set#episode_3455/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3455/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[259, 485], [335, 515], [392, 555], [460, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[485, 622], [530, 602], [574, 599], [629, 606]]</point>"}]}
{"id": 5898, "image": ["rtx_frames_success_46/62_robo_set#episode_3459/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3468/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[278, 482], [390, 475], [517, 505], [638, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[277, 468], [405, 485], [509, 512], [627, 559]]</point>"}]}
{"id": 5900, "image": ["rtx_frames_success_46/62_robo_set#episode_3476/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3476/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[271, 475], [320, 529], [371, 579], [434, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer to the right."}, {"from": "gpt", "value": "<point>[[434, 629], [498, 596], [555, 582], [616, 586]]</point>"}]}
{"id": 5902, "image": ["rtx_frames_success_46/62_robo_set#episode_3482/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3501/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 452], [386, 478], [477, 515], [570, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[284, 489], [330, 539], [373, 576], [417, 602]]</point>"}]}
{"id": 5904, "image": ["rtx_frames_success_46/62_robo_set#episode_3501/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3514/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[456, 602], [509, 592], [557, 592], [612, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 485], [381, 515], [489, 549], [587, 596]]</point>"}]}
{"id": 5906, "image": ["rtx_frames_success_46/62_robo_set#episode_3522/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3523/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 489], [322, 545], [369, 586], [422, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[290, 442], [381, 472], [481, 515], [578, 572]]</point>"}]}
{"id": 5908, "image": ["rtx_frames_success_46/62_robo_set#episode_3525/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3526/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[263, 482], [347, 432], [419, 402], [513, 361]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[528, 428], [479, 458], [419, 515], [367, 569]]</point>"}]}
{"id": 5910, "image": ["rtx_frames_success_46/62_robo_set#episode_3532/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3533/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 478], [422, 482], [578, 489], [805, 509]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[259, 462], [343, 415], [441, 361], [513, 328]]</point>"}]}
{"id": 5912, "image": ["rtx_frames_success_46/62_robo_set#episode_3539/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3543/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[278, 462], [384, 502], [472, 539], [580, 589]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[263, 465], [388, 502], [487, 552], [580, 612]]</point>"}]}
{"id": 5914, "image": ["rtx_frames_success_46/62_robo_set#episode_3550/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3560/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 458], [367, 391], [441, 348], [517, 318]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[322, 472], [337, 549], [364, 602], [411, 669]]</point>"}]}
{"id": 5916, "image": ["rtx_frames_success_46/62_robo_set#episode_3560/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3574/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pour ketchup on the table."}, {"from": "gpt", "value": "<point>[[400, 689], [447, 612], [498, 586], [576, 552]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[244, 472], [422, 492], [604, 525], [799, 589]]</point>"}]}
{"id": 5918, "image": ["rtx_frames_success_46/62_robo_set#episode_3594/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3598/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[286, 489], [409, 499], [515, 529], [623, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[307, 465], [333, 525], [358, 602], [403, 646]]</point>"}]}
{"id": 5920, "image": ["rtx_frames_success_46/62_robo_set#episode_3598/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3616/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[377, 656], [438, 609], [496, 582], [578, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 465], [390, 489], [473, 529], [581, 599]]</point>"}]}
{"id": 5922, "image": ["rtx_frames_success_46/62_robo_set#episode_3619/frame_15.png", "rtx_frames_success_46/62_robo_set#episode_3630/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[366, 545], [375, 639], [381, 719], [379, 800]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 478], [369, 532], [468, 569], [566, 579]]</point>"}]}
{"id": 5924, "image": ["rtx_frames_success_46/62_robo_set#episode_3636/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3642/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[290, 468], [453, 492], [610, 535], [795, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[252, 499], [371, 512], [470, 539], [583, 592]]</point>"}]}
{"id": 5926, "image": ["rtx_frames_success_46/62_robo_set#episode_3653/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3655/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[259, 495], [379, 505], [492, 542], [604, 596]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana in the container."}, {"from": "gpt", "value": "<point>[[286, 485], [462, 502], [625, 522], [824, 555]]</point>"}]}
{"id": 5928, "image": ["rtx_frames_success_46/62_robo_set#episode_3668/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3678/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[259, 468], [362, 512], [473, 542], [595, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[319, 468], [391, 401], [462, 327], [534, 275]]</point>"}]}
{"id": 5930, "image": ["rtx_frames_success_46/62_robo_set#episode_3683/frame_0.png", "rtx_frames_success_46/62_robo_set#episode_3693/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[286, 512], [349, 557], [395, 586], [437, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[298, 527], [349, 572], [382, 616], [403, 698]]</point>"}]}
{"id": 5932, "image": ["rtx_frames_success_46/62_robo_set#episode_3693/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_42/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[353, 586], [357, 468], [357, 364], [387, 304]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana on the tray."}, {"from": "gpt", "value": "<point>[[247, 459], [461, 489], [648, 531], [858, 599]]</point>"}]}
{"id": 5934, "image": ["rtx_frames_success_47/62_robo_set#episode_381/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_386/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[298, 459], [403, 463], [509, 481], [588, 523]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[425, 318], [401, 379], [376, 444], [345, 538]]</point>"}]}
{"id": 5936, "image": ["rtx_frames_success_47/62_robo_set#episode_387/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_400/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[277, 459], [470, 444], [624, 440], [818, 470]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[300, 455], [365, 485], [416, 531], [453, 603]]</point>"}]}
{"id": 5938, "image": ["rtx_frames_success_47/62_robo_set#episode_400/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_402/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[464, 637], [521, 622], [569, 618], [629, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[288, 466], [369, 463], [427, 478], [506, 508]]</point>"}]}
{"id": 5940, "image": ["rtx_frames_success_47/62_robo_set#episode_410/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_410/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[300, 485], [423, 432], [534, 334], [607, 235]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[629, 296], [639, 444], [637, 584], [633, 735]]</point>"}]}
{"id": 5942, "image": ["rtx_frames_success_47/62_robo_set#episode_417/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_423/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 500], [388, 531], [459, 538], [571, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[292, 500], [380, 440], [446, 379], [526, 277]]</point>"}]}
{"id": 5944, "image": ["rtx_frames_success_47/62_robo_set#episode_423/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_429/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the bottle towards the table."}, {"from": "gpt", "value": "<point>[[517, 303], [556, 402], [592, 493], [614, 599]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[281, 481], [382, 512], [481, 542], [601, 557]]</point>"}]}
{"id": 5946, "image": ["rtx_frames_success_47/62_robo_set#episode_430/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_441/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[279, 485], [438, 489], [599, 538], [768, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[320, 463], [348, 550], [382, 591], [423, 610]]</point>"}]}
{"id": 5948, "image": ["rtx_frames_success_47/62_robo_set#episode_441/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_453/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the strainer."}, {"from": "gpt", "value": "<point>[[408, 614], [472, 591], [524, 591], [582, 599]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[288, 478], [429, 493], [586, 527], [732, 576]]</point>"}]}
{"id": 5950, "image": ["rtx_frames_success_47/62_robo_set#episode_454/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_454/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[277, 470], [386, 368], [476, 307], [614, 235]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[616, 307], [629, 425], [633, 527], [629, 637]]</point>"}]}
{"id": 5952, "image": ["rtx_frames_success_47/62_robo_set#episode_3726/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3731/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[268, 463], [333, 489], [399, 546], [455, 610]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer handle."}, {"from": "gpt", "value": "<point>[[496, 466], [423, 519], [356, 591], [283, 690]]</point>"}]}
{"id": 5954, "image": ["rtx_frames_success_47/62_robo_set#episode_3736/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_3737/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[388, 372], [354, 428], [335, 497], [318, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 493], [466, 504], [618, 519], [815, 580]]</point>"}]}
{"id": 5956, "image": ["rtx_frames_success_47/62_robo_set#episode_3741/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_3753/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[457, 318], [423, 379], [401, 425], [373, 489]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[288, 485], [425, 519], [547, 580], [652, 652]]</point>"}]}
{"id": 5958, "image": ["rtx_frames_success_47/62_robo_set#episode_3753/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_3764/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[633, 542], [644, 463], [642, 375], [631, 288]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[283, 485], [391, 493], [496, 523], [609, 584]]</point>"}]}
{"id": 5960, "image": ["rtx_frames_success_47/62_robo_set#episode_3766/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3766/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[279, 485], [425, 409], [530, 326], [635, 197]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[624, 262], [620, 425], [622, 576], [644, 717]]</point>"}]}
{"id": 5962, "image": ["rtx_frames_success_47/62_robo_set#episode_3767/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3775/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[288, 481], [395, 493], [476, 519], [579, 588]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer closed."}, {"from": "gpt", "value": "<point>[[459, 478], [397, 504], [348, 550], [283, 610]]</point>"}]}
{"id": 5964, "image": ["rtx_frames_success_47/62_robo_set#episode_3782/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3785/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[266, 447], [313, 451], [354, 455], [395, 470]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[262, 463], [412, 478], [554, 516], [704, 572]]</point>"}]}
{"id": 5966, "image": ["rtx_frames_success_47/62_robo_set#episode_3786/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3792/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[270, 481], [361, 398], [446, 356], [521, 345]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 478], [446, 485], [605, 508], [824, 554]]</point>"}]}
{"id": 5968, "image": ["rtx_frames_success_47/62_robo_set#episode_3800/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3800/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[264, 504], [324, 565], [378, 614], [425, 633]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer to the right."}, {"from": "gpt", "value": "<point>[[448, 591], [506, 572], [569, 569], [624, 591]]</point>"}]}
{"id": 5970, "image": ["rtx_frames_success_47/62_robo_set#episode_3808/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3815/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[281, 463], [391, 493], [483, 523], [575, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[266, 508], [444, 504], [620, 542], [800, 595]]</point>"}]}
{"id": 5972, "image": ["rtx_frames_success_47/62_robo_set#episode_3826/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3836/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[268, 493], [425, 481], [545, 497], [700, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[283, 474], [365, 409], [446, 379], [526, 360]]</point>"}]}
{"id": 5974, "image": ["rtx_frames_success_47/62_robo_set#episode_3838/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3840/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[292, 463], [333, 550], [365, 595], [412, 629]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[283, 463], [356, 402], [423, 368], [524, 334]]</point>"}]}
{"id": 5976, "image": ["rtx_frames_success_47/62_robo_set#episode_3845/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3849/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[268, 485], [393, 481], [511, 500], [618, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the strainer."}, {"from": "gpt", "value": "<point>[[262, 474], [318, 447], [348, 436], [393, 436]]</point>"}]}
{"id": 5978, "image": ["rtx_frames_success_47/62_robo_set#episode_3850/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3857/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[279, 447], [382, 519], [487, 572], [588, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[266, 463], [391, 493], [481, 535], [575, 607]]</point>"}]}
{"id": 5980, "image": ["rtx_frames_success_47/62_robo_set#episode_3861/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3862/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[266, 463], [403, 481], [496, 508], [594, 554]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[285, 474], [442, 523], [614, 538], [798, 516]]</point>"}]}
{"id": 5982, "image": ["rtx_frames_success_47/62_robo_set#episode_3864/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3864/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[266, 463], [341, 508], [399, 557], [451, 626]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[481, 622], [526, 584], [586, 569], [642, 572]]</point>"}]}
{"id": 5984, "image": ["rtx_frames_success_47/62_robo_set#episode_3865/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3924/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 474], [378, 485], [470, 508], [571, 550]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[296, 478], [380, 394], [436, 364], [528, 349]]</point>"}]}
{"id": 5986, "image": ["rtx_frames_success_47/62_robo_set#episode_3931/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3931/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[266, 463], [318, 493], [361, 535], [416, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the bowl to the right."}, {"from": "gpt", "value": "<point>[[444, 607], [487, 610], [536, 622], [577, 603]]</point>"}]}
{"id": 5988, "image": ["rtx_frames_success_47/62_robo_set#episode_3953/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3960/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[305, 474], [395, 493], [472, 519], [567, 554]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[279, 481], [401, 527], [511, 584], [644, 656]]</point>"}]}
{"id": 5990, "image": ["rtx_frames_success_47/62_robo_set#episode_3960/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_3966/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[633, 512], [637, 436], [639, 372], [631, 296]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[260, 478], [412, 493], [536, 516], [700, 584]]</point>"}]}
{"id": 5992, "image": ["rtx_frames_success_47/62_robo_set#episode_3978/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_3982/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[279, 474], [361, 390], [429, 334], [532, 292]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[296, 481], [429, 489], [558, 508], [695, 569]]</point>"}]}
{"id": 5994, "image": ["rtx_frames_success_47/62_robo_set#episode_3995/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4002/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[266, 504], [384, 516], [470, 531], [584, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[288, 463], [395, 447], [498, 455], [588, 489]]</point>"}]}
{"id": 5996, "image": ["rtx_frames_success_47/62_robo_set#episode_4010/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4026/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[225, 478], [416, 478], [577, 493], [811, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[262, 470], [399, 485], [476, 516], [588, 569]]</point>"}]}
{"id": 5998, "image": ["rtx_frames_success_47/62_robo_set#episode_4036/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4041/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[258, 489], [326, 554], [380, 595], [455, 637]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[238, 466], [382, 478], [455, 497], [588, 546]]</point>"}]}
{"id": 6000, "image": ["rtx_frames_success_47/62_robo_set#episode_4055/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4069/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[285, 485], [401, 531], [502, 595], [603, 682]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[262, 470], [313, 463], [363, 440], [386, 387]]</point>"}]}
{"id": 6002, "image": ["rtx_frames_success_47/62_robo_set#episode_4069/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4071/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[386, 516], [384, 626], [384, 682], [397, 770]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[281, 470], [328, 523], [391, 572], [455, 603]]</point>"}]}
{"id": 6004, "image": ["rtx_frames_success_47/62_robo_set#episode_4071/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4076/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[483, 641], [539, 626], [584, 618], [644, 637]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[275, 485], [324, 466], [361, 455], [403, 466]]</point>"}]}
{"id": 6006, "image": ["rtx_frames_success_47/62_robo_set#episode_4076/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4077/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[388, 531], [386, 614], [384, 690], [393, 754]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the ketchup bottle."}, {"from": "gpt", "value": "<point>[[315, 463], [328, 535], [365, 603], [408, 652]]</point>"}]}
{"id": 6008, "image": ["rtx_frames_success_47/62_robo_set#episode_4085/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4089/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[262, 463], [373, 497], [481, 554], [582, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[266, 493], [444, 493], [590, 493], [798, 538]]</point>"}]}
{"id": 6010, "image": ["rtx_frames_success_47/62_robo_set#episode_4095/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4096/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[262, 459], [438, 432], [597, 455], [811, 516]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[283, 455], [446, 444], [614, 459], [760, 504]]</point>"}]}
{"id": 6012, "image": ["rtx_frames_success_47/62_robo_set#episode_4099/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4106/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to close the drawer."}, {"from": "gpt", "value": "<point>[[431, 390], [401, 432], [380, 493], [361, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[294, 474], [414, 485], [519, 516], [607, 569]]</point>"}]}
{"id": 6014, "image": ["rtx_frames_success_47/62_robo_set#episode_4106/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4111/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[453, 622], [470, 481], [491, 379], [534, 273]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[260, 466], [371, 455], [485, 481], [594, 535]]</point>"}]}
{"id": 6016, "image": ["rtx_frames_success_47/62_robo_set#episode_4138/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4139/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[290, 466], [397, 478], [494, 504], [592, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[285, 497], [388, 512], [468, 523], [579, 561]]</point>"}]}
{"id": 6018, "image": ["rtx_frames_success_47/62_robo_set#episode_4141/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4166/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[476, 372], [453, 436], [410, 519], [363, 569]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[294, 463], [401, 481], [494, 508], [590, 561]]</point>"}]}
{"id": 6020, "image": ["rtx_frames_success_47/62_robo_set#episode_4167/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4170/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[270, 485], [416, 493], [513, 516], [631, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 489], [367, 481], [455, 504], [549, 546]]</point>"}]}
{"id": 6022, "image": ["rtx_frames_success_47/62_robo_set#episode_4174/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4186/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[273, 485], [328, 542], [373, 569], [421, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[273, 504], [363, 428], [451, 383], [536, 360]]</point>"}]}
{"id": 6024, "image": ["rtx_frames_success_47/62_robo_set#episode_4191/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4196/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 470], [376, 353], [479, 277], [614, 209]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[270, 493], [412, 402], [515, 307], [618, 186]]</point>"}]}
{"id": 6026, "image": ["rtx_frames_success_47/62_robo_set#episode_4196/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4207/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[631, 258], [648, 409], [652, 561], [650, 720]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[264, 470], [318, 436], [358, 436], [388, 459]]</point>"}]}
{"id": 6028, "image": ["rtx_frames_success_47/62_robo_set#episode_4207/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4208/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[388, 569], [388, 663], [393, 743], [401, 838]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[262, 478], [378, 481], [485, 512], [614, 569]]</point>"}]}
{"id": 6030, "image": ["rtx_frames_success_47/62_robo_set#episode_4225/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4228/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[279, 485], [410, 497], [554, 523], [695, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[270, 447], [373, 447], [464, 481], [554, 546]]</point>"}]}
{"id": 6032, "image": ["rtx_frames_success_47/62_robo_set#episode_4237/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4243/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[303, 470], [397, 538], [491, 588], [579, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[262, 493], [356, 398], [429, 349], [530, 326]]</point>"}]}
{"id": 6034, "image": ["rtx_frames_success_47/62_robo_set#episode_4245/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4250/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[258, 466], [373, 485], [468, 516], [594, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[240, 470], [311, 500], [380, 561], [442, 626]]</point>"}]}
{"id": 6036, "image": ["rtx_frames_success_47/62_robo_set#episode_4255/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4262/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[313, 459], [330, 531], [361, 603], [416, 679]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[309, 481], [427, 508], [530, 569], [639, 644]]</point>"}]}
{"id": 6038, "image": ["rtx_frames_success_47/62_robo_set#episode_4270/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4272/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[281, 470], [367, 402], [446, 364], [552, 334]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[255, 478], [324, 500], [391, 538], [455, 622]]</point>"}]}
{"id": 6040, "image": ["rtx_frames_success_47/62_robo_set#episode_4272/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4279/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[479, 614], [532, 591], [582, 584], [650, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[309, 516], [401, 523], [506, 542], [618, 584]]</point>"}]}
{"id": 6042, "image": ["rtx_frames_success_47/62_robo_set#episode_4286/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4291/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 466], [378, 489], [483, 523], [582, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[249, 466], [412, 447], [579, 459], [798, 516]]</point>"}]}
{"id": 6044, "image": ["rtx_frames_success_47/62_robo_set#episode_4296/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4297/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[251, 478], [442, 466], [609, 493], [813, 531]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the lid of the tea container."}, {"from": "gpt", "value": "<point>[[489, 27], [609, 53], [700, 118], [766, 212]]</point>"}]}
{"id": 6046, "image": ["rtx_frames_success_47/62_robo_set#episode_4310/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4317/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle in the strainer."}, {"from": "gpt", "value": "<point>[[313, 455], [330, 531], [365, 607], [406, 648]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[305, 459], [333, 535], [363, 591], [403, 637]]</point>"}]}
{"id": 6048, "image": ["rtx_frames_success_47/62_robo_set#episode_4322/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4333/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[268, 463], [320, 531], [361, 561], [412, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[296, 466], [388, 538], [481, 565], [605, 576]]</point>"}]}
{"id": 6050, "image": ["rtx_frames_success_47/62_robo_set#episode_4336/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4336/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[292, 478], [418, 421], [515, 330], [612, 201]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[616, 326], [637, 432], [646, 538], [657, 682]]</point>"}]}
{"id": 6052, "image": ["rtx_frames_success_47/62_robo_set#episode_4343/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4352/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[283, 466], [401, 508], [491, 542], [592, 588]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[273, 474], [345, 406], [438, 341], [521, 311]]</point>"}]}
{"id": 6054, "image": ["rtx_frames_success_47/62_robo_set#episode_4359/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4379/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[279, 463], [446, 463], [629, 489], [800, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[283, 497], [393, 478], [487, 485], [599, 504]]</point>"}]}
{"id": 6056, "image": ["rtx_frames_success_47/62_robo_set#episode_4394/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4394/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[277, 470], [378, 368], [474, 292], [609, 224]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[612, 311], [620, 425], [624, 527], [622, 660]]</point>"}]}
{"id": 6058, "image": ["rtx_frames_success_47/62_robo_set#episode_4400/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4404/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[481, 303], [470, 353], [459, 394], [423, 440]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[273, 466], [343, 485], [403, 531], [453, 610]]</point>"}]}
{"id": 6060, "image": ["rtx_frames_success_47/62_robo_set#episode_4406/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4409/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 485], [382, 493], [476, 512], [567, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 478], [468, 474], [614, 481], [800, 535]]</point>"}]}
{"id": 6062, "image": ["rtx_frames_success_47/62_robo_set#episode_4412/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4415/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle cap."}, {"from": "gpt", "value": "<point>[[288, 470], [433, 425], [549, 398], [670, 387]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[283, 459], [322, 531], [365, 576], [403, 595]]</point>"}]}
{"id": 6064, "image": ["rtx_frames_success_47/62_robo_set#episode_4415/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4425/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[365, 546], [376, 481], [371, 394], [361, 341]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[270, 493], [341, 425], [431, 364], [528, 322]]</point>"}]}
{"id": 6066, "image": ["rtx_frames_success_47/62_robo_set#episode_4433/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4434/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the red ball."}, {"from": "gpt", "value": "<point>[[288, 489], [421, 481], [547, 493], [682, 523]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[494, 23], [584, 68], [672, 129], [747, 243]]</point>"}]}
{"id": 6068, "image": ["rtx_frames_success_47/62_robo_set#episode_4442/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4454/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 463], [363, 478], [481, 538], [584, 595]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[266, 481], [322, 470], [367, 463], [401, 489]]</point>"}]}
{"id": 6070, "image": ["rtx_frames_success_47/62_robo_set#episode_4454/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4456/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[378, 554], [367, 644], [376, 728], [386, 789]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the mug."}, {"from": "gpt", "value": "<point>[[258, 478], [395, 402], [515, 315], [622, 216]]</point>"}]}
{"id": 6072, "image": ["rtx_frames_success_47/62_robo_set#episode_4456/frame_15.png", "rtx_frames_success_47/62_robo_set#episode_4465/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[614, 284], [633, 421], [646, 561], [644, 686]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[249, 459], [380, 481], [481, 523], [573, 584]]</point>"}]}
{"id": 6074, "image": ["rtx_frames_success_47/62_robo_set#episode_4480/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4487/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[266, 493], [403, 478], [506, 489], [609, 546]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[277, 447], [318, 425], [358, 432], [391, 455]]</point>"}]}
{"id": 6076, "image": ["rtx_frames_success_47/62_robo_set#episode_4499/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4501/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[288, 478], [352, 508], [395, 550], [436, 614]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[455, 341], [429, 402], [406, 455], [363, 523]]</point>"}]}
{"id": 6078, "image": ["rtx_frames_success_47/62_robo_set#episode_4502/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4506/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[288, 500], [425, 527], [532, 572], [639, 660]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[277, 478], [373, 489], [461, 535], [556, 607]]</point>"}]}
{"id": 6080, "image": ["rtx_frames_success_47/62_robo_set#episode_4513/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4517/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 478], [397, 478], [517, 516], [631, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the lid."}, {"from": "gpt", "value": "<point>[[502, 23], [597, 42], [678, 72], [751, 121]]</point>"}]}
{"id": 6082, "image": ["rtx_frames_success_47/62_robo_set#episode_4528/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4535/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[288, 470], [358, 402], [425, 356], [517, 326]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 463], [408, 466], [494, 489], [609, 546]]</point>"}]}
{"id": 6084, "image": ["rtx_frames_success_47/62_robo_set#episode_4539/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4550/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[260, 470], [378, 489], [461, 519], [562, 584]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[277, 478], [354, 504], [414, 554], [464, 614]]</point>"}]}
{"id": 6086, "image": ["rtx_frames_success_47/62_robo_set#episode_4562/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4572/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[251, 447], [421, 417], [601, 417], [768, 463]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[279, 485], [330, 531], [373, 565], [421, 580]]</point>"}]}
{"id": 6088, "image": ["rtx_frames_success_47/62_robo_set#episode_4575/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4579/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 459], [300, 535], [345, 599], [401, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[307, 478], [335, 569], [367, 633], [416, 705]]</point>"}]}
{"id": 6090, "image": ["rtx_frames_success_47/62_robo_set#episode_4592/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4597/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[283, 493], [361, 421], [455, 349], [539, 307]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[285, 489], [393, 489], [485, 519], [582, 572]]</point>"}]}
{"id": 6092, "image": ["rtx_frames_success_47/62_robo_set#episode_4600/frame_0.png", "rtx_frames_success_47/62_robo_set#episode_4613/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle placed on the oven."}, {"from": "gpt", "value": "<point>[[270, 459], [438, 444], [618, 466], [783, 523]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[279, 478], [382, 474], [479, 508], [567, 569]]</point>"}]}
{"id": 6094, "image": ["rtx_frames_success_48/62_robo_set#episode_54/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_464/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[271, 472], [372, 392], [481, 339], [545, 339]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[273, 505], [320, 551], [361, 591], [406, 601]]</point>"}]}
{"id": 6096, "image": ["rtx_frames_success_48/62_robo_set#episode_472/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_472/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[271, 468], [323, 542], [387, 601], [447, 618]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[468, 615], [528, 585], [588, 575], [660, 591]]</point>"}]}
{"id": 6098, "image": ["rtx_frames_success_48/62_robo_set#episode_479/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_499/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[267, 478], [462, 532], [620, 545], [827, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[261, 458], [393, 458], [483, 478], [575, 542]]</point>"}]}
{"id": 6100, "image": ["rtx_frames_success_48/62_robo_set#episode_507/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_510/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[265, 458], [337, 532], [444, 585], [564, 605]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[275, 485], [359, 405], [431, 359], [525, 342]]</point>"}]}
{"id": 6102, "image": ["rtx_frames_success_48/62_robo_set#episode_520/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_520/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[263, 485], [399, 419], [498, 346], [615, 216]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[624, 286], [613, 409], [619, 542], [641, 645]]</point>"}]}
{"id": 6104, "image": ["rtx_frames_success_48/62_robo_set#episode_522/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_524/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[278, 472], [389, 472], [510, 502], [615, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[252, 475], [372, 485], [498, 515], [606, 565]]</point>"}]}
{"id": 6106, "image": ["rtx_frames_success_48/62_robo_set#episode_547/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_547/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 432], [306, 505], [363, 578], [408, 605]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[363, 585], [355, 492], [359, 409], [376, 349]]</point>"}]}
{"id": 6108, "image": ["rtx_frames_success_48/62_robo_set#episode_4623/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4623/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[260, 475], [369, 449], [487, 372], [607, 276]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[622, 419], [613, 525], [621, 631], [634, 714]]</point>"}]}
{"id": 6110, "image": ["rtx_frames_success_48/62_robo_set#episode_4625/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4633/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[271, 498], [414, 488], [515, 525], [590, 581]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[258, 475], [372, 352], [491, 276], [626, 226]]</point>"}]}
{"id": 6112, "image": ["rtx_frames_success_48/62_robo_set#episode_4633/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_4637/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[630, 289], [607, 412], [615, 618], [645, 694]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[303, 478], [399, 458], [495, 478], [594, 551]]</point>"}]}
{"id": 6114, "image": ["rtx_frames_success_48/62_robo_set#episode_4639/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4654/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[273, 475], [354, 405], [463, 339], [538, 312]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[284, 478], [399, 458], [508, 482], [596, 538]]</point>"}]}
{"id": 6116, "image": ["rtx_frames_success_48/62_robo_set#episode_4666/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4673/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[261, 482], [344, 525], [476, 558], [559, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana in the container."}, {"from": "gpt", "value": "<point>[[290, 498], [414, 475], [671, 495], [801, 535]]</point>"}]}
{"id": 6118, "image": ["rtx_frames_success_48/62_robo_set#episode_4675/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4678/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[263, 485], [391, 495], [502, 522], [598, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[286, 492], [327, 468], [374, 465], [397, 478]]</point>"}]}
{"id": 6120, "image": ["rtx_frames_success_48/62_robo_set#episode_4694/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4710/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[246, 485], [440, 535], [617, 528], [846, 502]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[521, 13], [472, 96], [421, 329], [421, 468]]</point>"}]}
{"id": 6122, "image": ["rtx_frames_success_48/62_robo_set#episode_4715/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4716/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 449], [397, 452], [496, 475], [583, 518]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the ketchup bottle."}, {"from": "gpt", "value": "<point>[[293, 492], [327, 581], [368, 661], [415, 711]]</point>"}]}
{"id": 6124, "image": ["rtx_frames_success_48/62_robo_set#episode_4716/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_4717/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the ketchup bottle."}, {"from": "gpt", "value": "<point>[[385, 704], [457, 668], [532, 648], [594, 644]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[271, 502], [367, 558], [472, 578], [583, 555]]</point>"}]}
{"id": 6126, "image": ["rtx_frames_success_48/62_robo_set#episode_4720/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4744/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 492], [382, 472], [553, 505], [651, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[289, 492], [391, 488], [489, 522], [585, 588]]</point>"}]}
{"id": 6128, "image": ["rtx_frames_success_48/62_robo_set#episode_4746/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4748/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[265, 485], [423, 465], [541, 482], [694, 532]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[301, 465], [423, 495], [508, 542], [605, 628]]</point>"}]}
{"id": 6130, "image": ["rtx_frames_success_48/62_robo_set#episode_4754/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4756/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[278, 502], [425, 495], [555, 525], [698, 575]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[289, 485], [382, 488], [496, 528], [570, 588]]</point>"}]}
{"id": 6132, "image": ["rtx_frames_success_48/62_robo_set#episode_4766/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4770/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 502], [310, 575], [365, 658], [406, 688]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[305, 472], [342, 535], [380, 588], [425, 631]]</point>"}]}
{"id": 6134, "image": ["rtx_frames_success_48/62_robo_set#episode_4771/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4771/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[273, 462], [395, 442], [526, 389], [617, 296]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[592, 465], [596, 542], [618, 641], [645, 698]]</point>"}]}
{"id": 6136, "image": ["rtx_frames_success_48/62_robo_set#episode_4773/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4777/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[278, 485], [352, 412], [436, 355], [538, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[289, 465], [325, 528], [376, 611], [414, 634]]</point>"}]}
{"id": 6138, "image": ["rtx_frames_success_48/62_robo_set#episode_4793/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4793/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[269, 495], [329, 571], [370, 611], [427, 615]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer from left to right."}, {"from": "gpt", "value": "<point>[[404, 611], [476, 585], [566, 575], [626, 591]]</point>"}]}
{"id": 6140, "image": ["rtx_frames_success_48/62_robo_set#episode_4794/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4801/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[278, 502], [400, 455], [555, 472], [684, 525]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[295, 465], [423, 472], [521, 512], [624, 585]]</point>"}]}
{"id": 6142, "image": ["rtx_frames_success_48/62_robo_set#episode_4807/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4816/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[267, 495], [440, 472], [660, 488], [844, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[295, 492], [425, 458], [541, 395], [639, 292]]</point>"}]}
{"id": 6144, "image": ["rtx_frames_success_48/62_robo_set#episode_4816/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_4819/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[624, 445], [615, 542], [624, 634], [648, 731]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[274, 488], [397, 395], [498, 352], [633, 336]]</point>"}]}
{"id": 6146, "image": ["rtx_frames_success_48/62_robo_set#episode_4819/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_4825/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[618, 542], [613, 625], [622, 694], [648, 764]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[280, 525], [382, 435], [489, 372], [628, 355]]</point>"}]}
{"id": 6148, "image": ["rtx_frames_success_48/62_robo_set#episode_4825/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_4836/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[636, 525], [624, 595], [624, 678], [641, 718]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[493, 27], [461, 156], [446, 322], [457, 465]]</point>"}]}
{"id": 6150, "image": ["rtx_frames_success_48/62_robo_set#episode_4837/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4837/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 478], [401, 419], [534, 312], [624, 193]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[615, 346], [596, 465], [602, 598], [630, 671]]</point>"}]}
{"id": 6152, "image": ["rtx_frames_success_48/62_robo_set#episode_4845/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4855/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[267, 508], [472, 399], [680, 392], [898, 439]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 478], [359, 525], [474, 555], [566, 538]]</point>"}]}
{"id": 6154, "image": ["rtx_frames_success_48/62_robo_set#episode_4857/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4857/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[293, 498], [387, 465], [528, 495], [620, 581]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[536, 664], [500, 515], [500, 399], [536, 279]]</point>"}]}
{"id": 6156, "image": ["rtx_frames_success_48/62_robo_set#episode_4870/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4875/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[265, 485], [425, 495], [534, 522], [643, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[271, 505], [355, 551], [445, 588], [549, 591]]</point>"}]}
{"id": 6158, "image": ["rtx_frames_success_48/62_robo_set#episode_4887/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4891/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[297, 462], [310, 542], [352, 611], [398, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 468], [387, 488], [485, 525], [572, 598]]</point>"}]}
{"id": 6160, "image": ["rtx_frames_success_48/62_robo_set#episode_4900/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4916/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[263, 505], [350, 419], [425, 372], [542, 342]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[258, 465], [355, 525], [462, 558], [553, 558]]</point>"}]}
{"id": 6162, "image": ["rtx_frames_success_48/62_robo_set#episode_4921/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4921/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[248, 462], [391, 458], [521, 402], [632, 286]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[636, 478], [622, 535], [628, 635], [647, 704]]</point>"}]}
{"id": 6164, "image": ["rtx_frames_success_48/62_robo_set#episode_4929/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4938/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[261, 475], [357, 535], [468, 558], [589, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[243, 478], [421, 452], [619, 475], [805, 545]]</point>"}]}
{"id": 6166, "image": ["rtx_frames_success_48/62_robo_set#episode_4951/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4956/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 465], [320, 555], [357, 618], [421, 684]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[274, 492], [368, 568], [468, 598], [590, 601]]</point>"}]}
{"id": 6168, "image": ["rtx_frames_success_48/62_robo_set#episode_4964/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4966/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[500, 43], [481, 199], [449, 332], [402, 435]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[271, 505], [384, 472], [455, 422], [536, 316]]</point>"}]}
{"id": 6170, "image": ["rtx_frames_success_48/62_robo_set#episode_4969/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4983/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[305, 475], [335, 555], [374, 628], [425, 674]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[307, 468], [395, 545], [470, 578], [560, 588]]</point>"}]}
{"id": 6172, "image": ["rtx_frames_success_48/62_robo_set#episode_4984/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4988/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[256, 498], [444, 498], [590, 518], [805, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the lid."}, {"from": "gpt", "value": "<point>[[506, 13], [478, 150], [474, 272], [478, 425]]</point>"}]}
{"id": 6174, "image": ["rtx_frames_success_48/62_robo_set#episode_4990/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_4993/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[250, 492], [395, 495], [506, 528], [609, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[256, 482], [376, 478], [525, 508], [624, 565]]</point>"}]}
{"id": 6176, "image": ["rtx_frames_success_48/62_robo_set#episode_5004/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5011/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[284, 478], [320, 568], [357, 621], [414, 671]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[254, 485], [380, 485], [468, 505], [581, 561]]</point>"}]}
{"id": 6178, "image": ["rtx_frames_success_48/62_robo_set#episode_5024/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5030/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[263, 492], [378, 485], [489, 505], [618, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[256, 485], [314, 542], [367, 578], [440, 618]]</point>"}]}
{"id": 6180, "image": ["rtx_frames_success_48/62_robo_set#episode_5030/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_5033/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug from left to right."}, {"from": "gpt", "value": "<point>[[468, 625], [538, 591], [583, 591], [643, 605]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[244, 492], [427, 458], [615, 492], [840, 585]]</point>"}]}
{"id": 6182, "image": ["rtx_frames_success_48/62_robo_set#episode_5038/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5044/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[269, 485], [383, 472], [496, 498], [596, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 478], [369, 482], [493, 518], [583, 568]]</point>"}]}
{"id": 6184, "image": ["rtx_frames_success_48/62_robo_set#episode_5047/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5065/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[273, 468], [325, 548], [395, 598], [453, 625]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 468], [427, 518], [594, 525], [797, 482]]</point>"}]}
{"id": 6186, "image": ["rtx_frames_success_48/62_robo_set#episode_5066/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5075/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 475], [455, 449], [618, 482], [821, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[261, 475], [350, 538], [436, 565], [553, 578]]</point>"}]}
{"id": 6188, "image": ["rtx_frames_success_48/62_robo_set#episode_5087/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5087/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[256, 478], [316, 505], [365, 508], [399, 498]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[380, 658], [370, 724], [370, 807], [393, 850]]</point>"}]}
{"id": 6190, "image": ["rtx_frames_success_48/62_robo_set#episode_5090/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5094/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[248, 502], [342, 399], [410, 359], [528, 346]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to open the drawer."}, {"from": "gpt", "value": "<point>[[280, 302], [269, 365], [271, 412], [291, 468]]</point>"}]}
{"id": 6192, "image": ["rtx_frames_success_48/62_robo_set#episode_5118/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5141/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[265, 468], [357, 375], [436, 342], [524, 336]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[274, 452], [314, 555], [359, 628], [419, 684]]</point>"}]}
{"id": 6194, "image": ["rtx_frames_success_48/62_robo_set#episode_5144/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5150/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[284, 465], [417, 505], [543, 568], [648, 661]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[494, 30], [588, 80], [680, 170], [737, 279]]</point>"}]}
{"id": 6196, "image": ["rtx_frames_success_48/62_robo_set#episode_5153/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5154/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[271, 452], [395, 478], [519, 538], [607, 605]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[504, 27], [492, 163], [470, 292], [419, 399]]</point>"}]}
{"id": 6198, "image": ["rtx_frames_success_48/62_robo_set#episode_5156/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5156/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[274, 482], [398, 545], [492, 551], [617, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the oven."}, {"from": "gpt", "value": "<point>[[352, 578], [508, 535], [647, 525], [805, 558]]</point>"}]}
{"id": 6200, "image": ["rtx_frames_success_48/62_robo_set#episode_5163/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5176/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[229, 465], [425, 442], [620, 458], [836, 512]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[259, 498], [383, 492], [491, 508], [626, 565]]</point>"}]}
{"id": 6202, "image": ["rtx_frames_success_48/62_robo_set#episode_5191/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5195/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[288, 458], [355, 505], [412, 548], [463, 628]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 488], [370, 492], [478, 515], [564, 568]]</point>"}]}
{"id": 6204, "image": ["rtx_frames_success_48/62_robo_set#episode_5220/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5224/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[271, 485], [382, 485], [504, 512], [613, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[280, 495], [419, 508], [538, 558], [630, 638]]</point>"}]}
{"id": 6206, "image": ["rtx_frames_success_48/62_robo_set#episode_5224/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_5242/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[626, 495], [613, 412], [622, 286], [647, 223]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 485], [384, 482], [506, 515], [609, 565]]</point>"}]}
{"id": 6208, "image": ["rtx_frames_success_48/62_robo_set#episode_5246/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5252/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[263, 478], [419, 488], [523, 518], [639, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[278, 488], [410, 505], [557, 578], [641, 664]]</point>"}]}
{"id": 6210, "image": ["rtx_frames_success_48/62_robo_set#episode_5255/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5262/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[265, 462], [348, 382], [440, 332], [528, 312]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[260, 452], [363, 442], [489, 472], [575, 545]]</point>"}]}
{"id": 6212, "image": ["rtx_frames_success_48/62_robo_set#episode_5269/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5269/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[297, 485], [337, 565], [372, 615], [416, 648]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[397, 668], [451, 601], [532, 555], [604, 548]]</point>"}]}
{"id": 6214, "image": ["rtx_frames_success_48/62_robo_set#episode_5272/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5277/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 458], [386, 462], [489, 505], [579, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[500, 27], [606, 73], [686, 153], [750, 276]]</point>"}]}
{"id": 6216, "image": ["rtx_frames_success_48/62_robo_set#episode_5284/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5284/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[284, 485], [329, 495], [372, 475], [397, 442]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[384, 528], [370, 621], [374, 698], [393, 771]]</point>"}]}
{"id": 6218, "image": ["rtx_frames_success_48/62_robo_set#episode_5296/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5304/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[261, 478], [438, 439], [598, 449], [809, 498]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[269, 498], [408, 522], [530, 571], [651, 674]]</point>"}]}
{"id": 6220, "image": ["rtx_frames_success_48/62_robo_set#episode_5310/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5321/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[265, 465], [369, 492], [483, 535], [579, 588]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[252, 475], [380, 475], [489, 505], [596, 578]]</point>"}]}
{"id": 6222, "image": ["rtx_frames_success_48/62_robo_set#episode_5331/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5338/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[267, 475], [414, 472], [542, 488], [700, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[271, 495], [384, 485], [500, 505], [606, 555]]</point>"}]}
{"id": 6224, "image": ["rtx_frames_success_48/62_robo_set#episode_5360/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5362/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the oven."}, {"from": "gpt", "value": "<point>[[239, 458], [419, 435], [652, 452], [812, 475]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[291, 468], [387, 465], [519, 502], [594, 545]]</point>"}]}
{"id": 6226, "image": ["rtx_frames_success_48/62_robo_set#episode_5363/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5365/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[248, 472], [433, 425], [592, 412], [818, 445]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[308, 458], [312, 561], [346, 631], [397, 644]]</point>"}]}
{"id": 6228, "image": ["rtx_frames_success_48/62_robo_set#episode_5373/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5385/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[244, 472], [331, 419], [427, 379], [525, 365]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[291, 478], [438, 432], [560, 412], [690, 415]]</point>"}]}
{"id": 6230, "image": ["rtx_frames_success_48/62_robo_set#episode_5388/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5402/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 472], [378, 485], [480, 525], [575, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[260, 485], [380, 435], [527, 405], [671, 399]]</point>"}]}
{"id": 6232, "image": ["rtx_frames_success_48/62_robo_set#episode_5412/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5412/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[286, 472], [323, 482], [357, 482], [387, 465]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[365, 532], [352, 621], [355, 701], [384, 774]]</point>"}]}
{"id": 6234, "image": ["rtx_frames_success_48/62_robo_set#episode_5413/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5419/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 465], [431, 432], [598, 429], [779, 472]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 455], [380, 449], [472, 492], [564, 558]]</point>"}]}
{"id": 6236, "image": ["rtx_frames_success_48/62_robo_set#episode_5420/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5436/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[269, 485], [395, 482], [506, 505], [636, 581]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 465], [354, 528], [449, 575], [545, 581]]</point>"}]}
{"id": 6238, "image": ["rtx_frames_success_48/62_robo_set#episode_5436/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_5439/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the ketchup bottle towards the plate."}, {"from": "gpt", "value": "<point>[[476, 645], [476, 538], [491, 442], [532, 342]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[282, 468], [314, 535], [355, 581], [398, 601]]</point>"}]}
{"id": 6240, "image": ["rtx_frames_success_48/62_robo_set#episode_5443/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5456/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[263, 465], [376, 445], [474, 458], [568, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 472], [384, 355], [508, 259], [624, 213]]</point>"}]}
{"id": 6242, "image": ["rtx_frames_success_48/62_robo_set#episode_5456/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_5457/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[606, 332], [600, 458], [621, 638], [656, 744]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[282, 498], [383, 538], [477, 561], [592, 568]]</point>"}]}
{"id": 6244, "image": ["rtx_frames_success_48/62_robo_set#episode_5462/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5462/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[233, 439], [282, 532], [350, 588], [423, 608]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer from left to right."}, {"from": "gpt", "value": "<point>[[480, 608], [527, 591], [585, 591], [639, 611]]</point>"}]}
{"id": 6246, "image": ["rtx_frames_success_48/62_robo_set#episode_5469/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5469/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[263, 508], [316, 498], [359, 468], [389, 425]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[382, 498], [380, 595], [382, 664], [393, 738]]</point>"}]}
{"id": 6248, "image": ["rtx_frames_success_48/62_robo_set#episode_5470/frame_0.png", "rtx_frames_success_48/62_robo_set#episode_5480/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 472], [303, 542], [338, 585], [384, 605]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[316, 505], [400, 385], [496, 289], [603, 219]]</point>"}]}
{"id": 6250, "image": ["rtx_frames_success_48/62_robo_set#episode_5480/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_5481/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[615, 299], [605, 402], [605, 568], [622, 711]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[276, 472], [299, 545], [338, 581], [382, 578]]</point>"}]}
{"id": 6252, "image": ["rtx_frames_success_48/62_robo_set#episode_5481/frame_15.png", "rtx_frames_success_48/62_robo_set#episode_5488/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[363, 528], [348, 458], [357, 392], [372, 342]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[299, 472], [325, 548], [374, 621], [421, 664]]</point>"}]}
{"id": 6254, "image": ["rtx_frames_success_49/62_robo_set#episode_59/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_556/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 489], [407, 542], [496, 561], [591, 523]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[365, 461], [448, 408], [490, 377], [540, 324]]</point>"}]}
{"id": 6256, "image": ["rtx_frames_success_49/62_robo_set#episode_572/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_596/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[268, 442], [362, 464], [513, 505], [586, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[247, 436], [455, 467], [541, 418], [600, 227]]</point>"}]}
{"id": 6258, "image": ["rtx_frames_success_49/62_robo_set#episode_603/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_613/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[238, 467], [317, 517], [460, 564], [614, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 452], [346, 530], [448, 598], [573, 626]]</point>"}]}
{"id": 6260, "image": ["rtx_frames_success_49/62_robo_set#episode_620/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_622/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[280, 424], [384, 517], [506, 576], [598, 530]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[277, 486], [427, 539], [586, 558], [815, 551]]</point>"}]}
{"id": 6262, "image": ["rtx_frames_success_49/62_robo_set#episode_625/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_630/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[270, 467], [409, 492], [506, 418], [550, 315]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[296, 467], [422, 523], [531, 573], [630, 561]]</point>"}]}
{"id": 6264, "image": ["rtx_frames_success_49/62_robo_set#episode_639/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5508/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[254, 470], [427, 570], [508, 570], [617, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[365, 458], [427, 474], [457, 530], [443, 623]]</point>"}]}
{"id": 6266, "image": ["rtx_frames_success_49/62_robo_set#episode_5522/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5553/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[383, 452], [462, 399], [582, 318], [637, 181]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[367, 467], [420, 446], [499, 389], [550, 293]]</point>"}]}
{"id": 6268, "image": ["rtx_frames_success_49/62_robo_set#episode_5558/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5561/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[363, 480], [489, 527], [564, 539], [582, 614]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[369, 474], [467, 458], [547, 483], [561, 545]]</point>"}]}
{"id": 6270, "image": ["rtx_frames_success_49/62_robo_set#episode_5569/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5573/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[342, 505], [414, 502], [473, 514], [589, 576]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[377, 492], [439, 477], [547, 511], [612, 583]]</point>"}]}
{"id": 6272, "image": ["rtx_frames_success_49/62_robo_set#episode_5573/frame_15.png", "rtx_frames_success_49/62_robo_set#episode_5574/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[601, 623], [612, 539], [593, 408], [533, 255]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[374, 477], [451, 486], [510, 486], [600, 558]]</point>"}]}
{"id": 6274, "image": ["rtx_frames_success_49/62_robo_set#episode_5580/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5586/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[265, 467], [363, 467], [446, 477], [568, 583]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 458], [358, 505], [397, 564], [411, 645]]</point>"}]}
{"id": 6276, "image": ["rtx_frames_success_49/62_robo_set#episode_5597/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5603/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[268, 483], [347, 527], [386, 589], [402, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[356, 470], [451, 455], [533, 461], [683, 442]]</point>"}]}
{"id": 6278, "image": ["rtx_frames_success_49/62_robo_set#episode_5608/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5608/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[332, 477], [460, 433], [554, 377], [626, 206]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[616, 287], [663, 389], [651, 545], [603, 704]]</point>"}]}
{"id": 6280, "image": ["rtx_frames_success_49/62_robo_set#episode_5619/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5635/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[358, 477], [446, 474], [549, 495], [568, 589]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the blue cup."}, {"from": "gpt", "value": "<point>[[354, 464], [481, 424], [571, 411], [672, 411]]</point>"}]}
{"id": 6282, "image": ["rtx_frames_success_49/62_robo_set#episode_5680/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5680/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[275, 474], [344, 470], [381, 511], [423, 620]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer from left to right."}, {"from": "gpt", "value": "<point>[[436, 617], [503, 611], [582, 604], [617, 664]]</point>"}]}
{"id": 6284, "image": ["rtx_frames_success_49/62_robo_set#episode_5689/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5724/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[284, 474], [434, 486], [660, 508], [859, 449]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[310, 452], [384, 495], [471, 555], [557, 536]]</point>"}]}
{"id": 6286, "image": ["rtx_frames_success_49/62_robo_set#episode_5726/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5740/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[284, 461], [379, 492], [478, 567], [568, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[298, 458], [376, 486], [490, 583], [628, 567]]</point>"}]}
{"id": 6288, "image": ["rtx_frames_success_49/62_robo_set#episode_5757/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5757/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[257, 477], [384, 495], [513, 433], [616, 231]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[610, 259], [596, 430], [556, 611], [598, 717]]</point>"}]}
{"id": 6290, "image": ["rtx_frames_success_49/62_robo_set#episode_5759/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5773/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[270, 489], [420, 499], [589, 523], [695, 442]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[293, 455], [397, 511], [492, 614], [594, 626]]</point>"}]}
{"id": 6292, "image": ["rtx_frames_success_49/62_robo_set#episode_5789/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5795/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[261, 470], [411, 555], [534, 620], [587, 595]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the ketchup bottle placed on the oven."}, {"from": "gpt", "value": "<point>[[249, 483], [367, 530], [571, 570], [825, 483]]</point>"}]}
{"id": 6294, "image": ["rtx_frames_success_49/62_robo_set#episode_5796/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5804/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the tray."}, {"from": "gpt", "value": "<point>[[252, 470], [407, 533], [547, 633], [852, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[309, 489], [425, 477], [492, 427], [522, 252]]</point>"}]}
{"id": 6296, "image": ["rtx_frames_success_49/62_robo_set#episode_5806/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5828/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[282, 489], [429, 564], [524, 601], [617, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[254, 477], [383, 561], [511, 664], [600, 648]]</point>"}]}
{"id": 6298, "image": ["rtx_frames_success_49/62_robo_set#episode_5829/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5845/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[268, 467], [400, 580], [483, 617], [550, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 461], [358, 527], [473, 586], [586, 573]]</point>"}]}
{"id": 6300, "image": ["rtx_frames_success_49/62_robo_set#episode_5847/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5847/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[332, 464], [353, 583], [372, 633], [409, 707]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[390, 695], [467, 726], [556, 729], [607, 633]]</point>"}]}
{"id": 6302, "image": ["rtx_frames_success_49/62_robo_set#episode_5864/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5894/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[259, 461], [354, 477], [439, 449], [520, 327]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 439], [455, 514], [589, 589], [861, 567]]</point>"}]}
{"id": 6304, "image": ["rtx_frames_success_49/62_robo_set#episode_5909/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5938/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[249, 480], [363, 567], [489, 611], [612, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 439], [328, 564], [376, 639], [416, 682]]</point>"}]}
{"id": 6306, "image": ["rtx_frames_success_49/62_robo_set#episode_5945/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5945/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[282, 477], [437, 604], [584, 673], [674, 685]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug forward."}, {"from": "gpt", "value": "<point>[[638, 558], [646, 470], [661, 377], [631, 227]]</point>"}]}
{"id": 6308, "image": ["rtx_frames_success_49/62_robo_set#episode_5947/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5973/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[259, 464], [339, 477], [466, 489], [543, 302]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 458], [374, 539], [496, 570], [601, 567]]</point>"}]}
{"id": 6310, "image": ["rtx_frames_success_49/62_robo_set#episode_5983/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_5984/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[235, 495], [434, 604], [638, 620], [825, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[257, 467], [414, 548], [547, 508], [686, 424]]</point>"}]}
{"id": 6312, "image": ["rtx_frames_success_49/62_robo_set#episode_6043/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6051/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the tray."}, {"from": "gpt", "value": "<point>[[295, 492], [448, 570], [600, 633], [843, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[242, 483], [374, 545], [511, 573], [607, 580]]</point>"}]}
{"id": 6314, "image": ["rtx_frames_success_49/62_robo_set#episode_6054/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6060/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[266, 480], [402, 551], [533, 548], [644, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[277, 470], [414, 486], [564, 439], [637, 243]]</point>"}]}
{"id": 6316, "image": ["rtx_frames_success_49/62_robo_set#episode_6085/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6086/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[279, 464], [386, 442], [473, 389], [522, 265]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[245, 470], [407, 576], [563, 589], [695, 542]]</point>"}]}
{"id": 6318, "image": ["rtx_frames_success_49/62_robo_set#episode_6109/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6118/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[323, 480], [400, 499], [497, 567], [608, 589]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[268, 449], [344, 499], [446, 558], [577, 576]]</point>"}]}
{"id": 6320, "image": ["rtx_frames_success_49/62_robo_set#episode_6130/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6132/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[263, 461], [374, 539], [460, 573], [593, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana."}, {"from": "gpt", "value": "<point>[[263, 474], [367, 542], [510, 620], [601, 586]]</point>"}]}
{"id": 6322, "image": ["rtx_frames_success_49/62_robo_set#episode_6137/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6153/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[310, 499], [409, 486], [489, 433], [522, 327]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle cap."}, {"from": "gpt", "value": "<point>[[316, 480], [460, 533], [582, 530], [670, 418]]</point>"}]}
{"id": 6324, "image": ["rtx_frames_success_49/62_robo_set#episode_6170/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6177/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[314, 446], [399, 520], [527, 558], [702, 530]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[245, 464], [384, 592], [559, 617], [797, 595]]</point>"}]}
{"id": 6326, "image": ["rtx_frames_success_49/62_robo_set#episode_6226/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6230/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[279, 489], [369, 517], [552, 446], [616, 209]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 480], [395, 520], [504, 576], [646, 558]]</point>"}]}
{"id": 6328, "image": ["rtx_frames_success_49/62_robo_set#episode_6233/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6239/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[303, 464], [411, 567], [504, 629], [580, 611]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[247, 470], [347, 495], [453, 480], [515, 355]]</point>"}]}
{"id": 6330, "image": ["rtx_frames_success_49/62_robo_set#episode_6244/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6246/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[273, 470], [422, 467], [480, 418], [527, 308]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[321, 480], [411, 539], [557, 586], [695, 514]]</point>"}]}
{"id": 6332, "image": ["rtx_frames_success_49/62_robo_set#episode_6248/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6253/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[333, 480], [418, 536], [501, 573], [651, 561]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[317, 492], [439, 558], [549, 604], [594, 583]]</point>"}]}
{"id": 6334, "image": ["rtx_frames_success_49/62_robo_set#episode_6259/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6275/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[252, 474], [384, 536], [559, 561], [691, 530]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the tray."}, {"from": "gpt", "value": "<point>[[282, 486], [411, 576], [619, 642], [829, 589]]</point>"}]}
{"id": 6336, "image": ["rtx_frames_success_49/62_robo_set#episode_6291/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6304/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[317, 483], [450, 489], [506, 446], [527, 343]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[303, 470], [416, 580], [527, 598], [626, 580]]</point>"}]}
{"id": 6338, "image": ["rtx_frames_success_49/62_robo_set#episode_6309/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6314/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 477], [370, 536], [496, 567], [570, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[312, 486], [437, 567], [536, 598], [653, 558]]</point>"}]}
{"id": 6340, "image": ["rtx_frames_success_49/62_robo_set#episode_6316/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6323/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[272, 511], [425, 551], [522, 551], [670, 424]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[287, 467], [377, 545], [460, 576], [591, 592]]</point>"}]}
{"id": 6342, "image": ["rtx_frames_success_49/62_robo_set#episode_6334/frame_0.png", "rtx_frames_success_49/62_robo_set#episode_6384/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[279, 464], [400, 595], [543, 620], [698, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[282, 477], [404, 576], [511, 651], [640, 651]]</point>"}]}
{"id": 6344, "image": ["rtx_frames_success_50/62_robo_set#episode_67/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_68/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[357, 509], [496, 426], [529, 318]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[335, 494], [518, 524], [590, 592]]</point>"}]}
{"id": 6346, "image": ["rtx_frames_success_50/62_robo_set#episode_70/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_72/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[341, 484], [413, 641], [463, 651]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 519], [693, 646], [806, 597]]</point>"}]}
{"id": 6348, "image": ["rtx_frames_success_50/62_robo_set#episode_663/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_666/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[357, 509], [612, 480], [698, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 504], [493, 680], [582, 626]]</point>"}]}
{"id": 6350, "image": ["rtx_frames_success_50/62_robo_set#episode_668/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_673/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[368, 494], [640, 597], [693, 533]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[501, 289], [424, 338], [463, 475]]</point>"}]}
{"id": 6352, "image": ["rtx_frames_success_50/62_robo_set#episode_674/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_683/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[463, 343], [327, 411], [285, 504]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[360, 499], [623, 470], [698, 533]]</point>"}]}
{"id": 6354, "image": ["rtx_frames_success_50/62_robo_set#episode_688/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_695/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[349, 499], [634, 445], [698, 519]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[343, 480], [507, 597], [568, 582]]</point>"}]}
{"id": 6356, "image": ["rtx_frames_success_50/62_robo_set#episode_698/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_699/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 509], [546, 636], [634, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[360, 509], [612, 303], [609, 191]]</point>"}]}
{"id": 6358, "image": ["rtx_frames_success_50/62_robo_set#episode_699/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_701/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[626, 352], [598, 592], [668, 773], [662, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[346, 494], [620, 622], [809, 582]]</point>"}]}
{"id": 6360, "image": ["rtx_frames_success_50/62_robo_set#episode_704/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_704/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[368, 504], [593, 299], [615, 210]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[615, 279], [590, 612], [657, 817]]</point>"}]}
{"id": 6362, "image": ["rtx_frames_success_50/62_robo_set#episode_705/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_706/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 514], [601, 710], [781, 641]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[343, 509], [532, 666], [587, 622]]</point>"}]}
{"id": 6364, "image": ["rtx_frames_success_50/62_robo_set#episode_713/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_715/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[360, 489], [634, 484], [693, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[363, 504], [521, 519], [579, 577]]</point>"}]}
{"id": 6366, "image": ["rtx_frames_success_50/62_robo_set#episode_722/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_725/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 509], [515, 622], [587, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 499], [748, 651], [789, 524]]</point>"}]}
{"id": 6368, "image": ["rtx_frames_success_50/62_robo_set#episode_726/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6403/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[368, 499], [546, 563], [593, 519]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[293, 477], [360, 520], [492, 611], [598, 576]]</point>"}]}
{"id": 6370, "image": ["rtx_frames_success_50/62_robo_set#episode_6405/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6406/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[257, 480], [370, 533], [573, 601], [794, 570]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[333, 474], [444, 551], [586, 558], [690, 533]]</point>"}]}
{"id": 6372, "image": ["rtx_frames_success_50/62_robo_set#episode_6407/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6407/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[323, 480], [414, 555], [504, 583], [605, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[540, 580], [563, 474], [552, 365], [533, 268]]</point>"}]}
{"id": 6374, "image": ["rtx_frames_success_50/62_robo_set#episode_6417/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6428/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle cap."}, {"from": "gpt", "value": "<point>[[332, 483], [478, 502], [619, 499], [683, 383]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[312, 480], [414, 561], [540, 601], [623, 576]]</point>"}]}
{"id": 6376, "image": ["rtx_frames_success_50/62_robo_set#episode_6428/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_6444/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[508, 614], [543, 530], [543, 405], [529, 287]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[305, 461], [400, 539], [494, 611], [578, 604]]</point>"}]}
{"id": 6378, "image": ["rtx_frames_success_50/62_robo_set#episode_6445/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6452/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[265, 461], [384, 523], [481, 586], [598, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[266, 464], [395, 570], [524, 580], [559, 520]]</point>"}]}
{"id": 6380, "image": ["rtx_frames_success_50/62_robo_set#episode_6481/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6481/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[235, 470], [390, 470], [563, 389], [631, 181]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[617, 352], [626, 505], [612, 639], [570, 729]]</point>"}]}
{"id": 6382, "image": ["rtx_frames_success_50/62_robo_set#episode_6501/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6515/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[247, 467], [386, 539], [480, 561], [570, 495]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[293, 464], [437, 564], [556, 592], [705, 542]]</point>"}]}
{"id": 6384, "image": ["rtx_frames_success_50/62_robo_set#episode_6523/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6527/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the container."}, {"from": "gpt", "value": "<point>[[265, 470], [413, 586], [578, 598], [847, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[250, 480], [379, 558], [610, 614], [850, 580]]</point>"}]}
{"id": 6386, "image": ["rtx_frames_success_50/62_robo_set#episode_6533/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6540/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[272, 483], [379, 551], [520, 580], [607, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[275, 470], [376, 545], [534, 583], [707, 558]]</point>"}]}
{"id": 6388, "image": ["rtx_frames_success_50/62_robo_set#episode_6551/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6571/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[287, 483], [409, 576], [531, 595], [616, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[280, 474], [476, 598], [649, 620], [832, 580]]</point>"}]}
{"id": 6390, "image": ["rtx_frames_success_50/62_robo_set#episode_6574/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6591/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[257, 474], [390, 542], [559, 542], [670, 408]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[310, 474], [414, 589], [499, 604], [573, 601]]</point>"}]}
{"id": 6392, "image": ["rtx_frames_success_50/62_robo_set#episode_6596/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6609/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 474], [392, 576], [490, 583], [575, 573]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[343, 509], [482, 529], [557, 582]]</point>"}]}
{"id": 6394, "image": ["rtx_frames_success_50/62_robo_set#episode_6632/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6632/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[360, 509], [382, 597], [446, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the strainer from left to right."}, {"from": "gpt", "value": "<point>[[479, 538], [571, 489], [637, 563]]</point>"}]}
{"id": 6396, "image": ["rtx_frames_success_50/62_robo_set#episode_6650/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6658/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[335, 509], [468, 577], [568, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 524], [524, 592], [618, 573]]</point>"}]}
{"id": 6398, "image": ["rtx_frames_success_50/62_robo_set#episode_6660/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_6668/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[432, 367], [296, 431], [299, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[396, 524], [424, 401], [391, 294]]</point>"}]}
{"id": 6400, "image": ["rtx_frames_success_50/62_robo_set#episode_6671/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6681/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 524], [684, 656], [809, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[346, 489], [529, 626], [596, 597]]</point>"}]}
{"id": 6402, "image": ["rtx_frames_success_50/62_robo_set#episode_6682/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6688/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the tray."}, {"from": "gpt", "value": "<point>[[341, 494], [729, 499], [850, 617]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 514], [499, 450], [535, 343]]</point>"}]}
{"id": 6404, "image": ["rtx_frames_success_50/62_robo_set#episode_6692/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6693/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 499], [515, 646], [571, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 514], [512, 387], [557, 269]]</point>"}]}
{"id": 6406, "image": ["rtx_frames_success_50/62_robo_set#episode_6702/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6709/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[349, 504], [482, 480], [515, 377]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 484], [524, 695], [601, 622]]</point>"}]}
{"id": 6408, "image": ["rtx_frames_success_50/62_robo_set#episode_6711/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6713/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[349, 519], [529, 626], [596, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[374, 499], [366, 656], [407, 690]]</point>"}]}
{"id": 6410, "image": ["rtx_frames_success_50/62_robo_set#episode_6714/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6718/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[330, 504], [474, 592], [557, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[302, 279], [285, 362], [321, 475]]</point>"}]}
{"id": 6412, "image": ["rtx_frames_success_50/62_robo_set#episode_6727/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6734/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 509], [488, 617], [573, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[360, 484], [587, 382], [676, 396]]</point>"}]}
{"id": 6414, "image": ["rtx_frames_success_50/62_robo_set#episode_6737/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6738/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 519], [524, 631], [604, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[454, 343], [341, 431], [341, 529]]</point>"}]}
{"id": 6416, "image": ["rtx_frames_success_50/62_robo_set#episode_6739/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6749/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[360, 514], [512, 631], [596, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle cap."}, {"from": "gpt", "value": "<point>[[349, 524], [601, 382], [670, 396]]</point>"}]}
{"id": 6418, "image": ["rtx_frames_success_50/62_robo_set#episode_6750/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6755/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[366, 489], [512, 641], [593, 612]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[349, 484], [410, 617], [440, 592]]</point>"}]}
{"id": 6420, "image": ["rtx_frames_success_50/62_robo_set#episode_6760/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6766/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[335, 514], [512, 519], [573, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 494], [579, 440], [734, 514]]</point>"}]}
{"id": 6422, "image": ["rtx_frames_success_50/62_robo_set#episode_6776/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_6778/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[402, 607], [399, 759], [432, 886]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[396, 563], [377, 724], [407, 871]]</point>"}]}
{"id": 6424, "image": ["rtx_frames_success_50/62_robo_set#episode_6786/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6793/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[366, 489], [501, 524], [573, 622]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[366, 504], [524, 259], [623, 181]]</point>"}]}
{"id": 6426, "image": ["rtx_frames_success_50/62_robo_set#episode_6793/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_6796/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[632, 279], [668, 538], [643, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[371, 489], [465, 303], [526, 313]]</point>"}]}
{"id": 6428, "image": ["rtx_frames_success_50/62_robo_set#episode_6808/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_6828/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[429, 347], [352, 411], [335, 519]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 499], [598, 382], [765, 489]]</point>"}]}
{"id": 6430, "image": ["rtx_frames_success_50/62_robo_set#episode_6835/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6840/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[357, 480], [504, 631], [576, 607]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[330, 514], [457, 274], [540, 333]]</point>"}]}
{"id": 6432, "image": ["rtx_frames_success_50/62_robo_set#episode_6841/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6844/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[341, 504], [524, 494], [604, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[482, 450], [294, 445], [224, 543]]</point>"}]}
{"id": 6434, "image": ["rtx_frames_success_50/62_robo_set#episode_6846/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6859/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[355, 514], [623, 480], [690, 529]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[343, 504], [515, 538], [568, 617]]</point>"}]}
{"id": 6436, "image": ["rtx_frames_success_50/62_robo_set#episode_6867/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6871/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[346, 489], [449, 323], [526, 352]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the strainer."}, {"from": "gpt", "value": "<point>[[360, 489], [391, 592], [432, 582]]</point>"}]}
{"id": 6438, "image": ["rtx_frames_success_50/62_robo_set#episode_6871/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_6873/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[438, 548], [512, 470], [562, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[360, 509], [521, 504], [601, 577]]</point>"}]}
{"id": 6440, "image": ["rtx_frames_success_50/62_robo_set#episode_6873/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_6880/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[548, 592], [587, 392], [535, 284]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 499], [518, 524], [584, 617]]</point>"}]}
{"id": 6442, "image": ["rtx_frames_success_50/62_robo_set#episode_6881/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6887/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 504], [529, 617], [607, 666]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[332, 509], [474, 494], [551, 582]]</point>"}]}
{"id": 6444, "image": ["rtx_frames_success_50/62_robo_set#episode_6905/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6906/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[366, 489], [479, 440], [546, 333]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[385, 504], [371, 597], [424, 636], [432, 558]]</point>"}]}
{"id": 6446, "image": ["rtx_frames_success_50/62_robo_set#episode_6907/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6907/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 504], [510, 254], [615, 196]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[626, 289], [587, 573], [645, 710]]</point>"}]}
{"id": 6448, "image": ["rtx_frames_success_50/62_robo_set#episode_6908/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6913/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[352, 509], [476, 308], [540, 318]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[341, 333], [285, 431], [352, 558]]</point>"}]}
{"id": 6450, "image": ["rtx_frames_success_50/62_robo_set#episode_6927/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6935/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[368, 499], [496, 612], [565, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[352, 484], [504, 524], [571, 602]]</point>"}]}
{"id": 6452, "image": ["rtx_frames_success_50/62_robo_set#episode_6936/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6937/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[360, 524], [598, 362], [681, 392]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 494], [706, 646], [806, 548]]</point>"}]}
{"id": 6454, "image": ["rtx_frames_success_50/62_robo_set#episode_6944/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6951/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[363, 509], [568, 509], [609, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[341, 514], [465, 607], [571, 592]]</point>"}]}
{"id": 6456, "image": ["rtx_frames_success_50/62_robo_set#episode_6958/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6964/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[377, 494], [377, 617], [418, 636]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 494], [471, 626], [576, 612]]</point>"}]}
{"id": 6458, "image": ["rtx_frames_success_50/62_robo_set#episode_6971/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_6973/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[355, 484], [510, 685], [579, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 484], [584, 582], [781, 514]]</point>"}]}
{"id": 6460, "image": ["rtx_frames_success_50/62_robo_set#episode_6991/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_7025/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the drawer towards its closed position."}, {"from": "gpt", "value": "<point>[[385, 289], [307, 352], [277, 455]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[366, 509], [518, 587], [604, 563]]</point>"}]}
{"id": 6462, "image": ["rtx_frames_success_50/62_robo_set#episode_7059/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7075/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[355, 494], [623, 592], [701, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle placed on the oven."}, {"from": "gpt", "value": "<point>[[355, 494], [687, 612], [787, 514]]</point>"}]}
{"id": 6464, "image": ["rtx_frames_success_50/62_robo_set#episode_7076/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_7080/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[388, 313], [338, 426], [335, 514]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 489], [482, 612], [537, 548]]</point>"}]}
{"id": 6466, "image": ["rtx_frames_success_50/62_robo_set#episode_7087/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7088/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[355, 509], [551, 524], [609, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[341, 504], [463, 323], [518, 343]]</point>"}]}
{"id": 6468, "image": ["rtx_frames_success_50/62_robo_set#episode_7096/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7098/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[360, 499], [609, 587], [693, 538]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[435, 357], [327, 538], [316, 626]]</point>"}]}
{"id": 6470, "image": ["rtx_frames_success_50/62_robo_set#episode_7105/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7107/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[349, 494], [463, 299], [521, 299]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[360, 494], [637, 577], [693, 533]]</point>"}]}
{"id": 6472, "image": ["rtx_frames_success_50/62_robo_set#episode_7109/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7124/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[366, 489], [493, 401], [521, 279]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[371, 494], [584, 607], [632, 553]]</point>"}]}
{"id": 6474, "image": ["rtx_frames_success_50/62_robo_set#episode_7149/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7161/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[366, 489], [524, 622], [584, 587]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[371, 504], [385, 582], [432, 587]]</point>"}]}
{"id": 6476, "image": ["rtx_frames_success_50/62_robo_set#episode_7169/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7169/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[368, 494], [427, 533], [446, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[482, 617], [582, 548], [645, 636]]</point>"}]}
{"id": 6478, "image": ["rtx_frames_success_50/62_robo_set#episode_7178/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7201/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[360, 494], [427, 607], [471, 656], [449, 553]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[349, 484], [529, 617], [593, 587]]</point>"}]}
{"id": 6480, "image": ["rtx_frames_success_50/62_robo_set#episode_7209/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7209/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 509], [532, 617], [607, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[548, 573], [571, 328], [529, 250]]</point>"}]}
{"id": 6482, "image": ["rtx_frames_success_50/62_robo_set#episode_7214/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7214/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[355, 514], [593, 362], [637, 254], [587, 254]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[634, 475], [618, 582], [648, 666]]</point>"}]}
{"id": 6484, "image": ["rtx_frames_success_50/62_robo_set#episode_7217/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7218/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[357, 489], [537, 577], [601, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana-like object."}, {"from": "gpt", "value": "<point>[[338, 504], [490, 622], [587, 587]]</point>"}]}
{"id": 6486, "image": ["rtx_frames_success_50/62_robo_set#episode_7220/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_7223/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[391, 450], [404, 343], [377, 274]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[360, 509], [590, 680], [665, 680]]</point>"}]}
{"id": 6488, "image": ["rtx_frames_success_50/62_robo_set#episode_7223/frame_15.png", "rtx_frames_success_50/62_robo_set#episode_7226/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug forwards."}, {"from": "gpt", "value": "<point>[[640, 440], [648, 279], [612, 235]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[349, 499], [615, 465], [681, 396]]</point>"}]}
{"id": 6490, "image": ["rtx_frames_success_50/62_robo_set#episode_7244/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7260/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[360, 489], [526, 436], [507, 313]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[346, 519], [548, 622], [623, 597]]</point>"}]}
{"id": 6492, "image": ["rtx_frames_success_50/62_robo_set#episode_7261/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7267/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[382, 499], [537, 622], [571, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[349, 499], [510, 612], [562, 592]]</point>"}]}
{"id": 6494, "image": ["rtx_frames_success_50/62_robo_set#episode_7270/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7271/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[346, 484], [535, 651], [568, 602]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[343, 494], [526, 524], [579, 582]]</point>"}]}
{"id": 6496, "image": ["rtx_frames_success_50/62_robo_set#episode_7277/frame_0.png", "rtx_frames_success_50/62_robo_set#episode_7279/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[355, 499], [551, 631], [607, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[360, 494], [352, 592], [396, 617]]</point>"}]}
{"id": 6498, "image": ["rtx_frames_success_51/62_robo_set#episode_729/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_732/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the banana."}, {"from": "gpt", "value": "<point>[[273, 472], [384, 465], [485, 512], [579, 581]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[244, 472], [386, 548], [551, 585], [707, 591]]</point>"}]}
{"id": 6500, "image": ["rtx_frames_success_51/62_robo_set#episode_751/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_785/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[271, 478], [487, 449], [685, 465], [842, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up a wooden lid."}, {"from": "gpt", "value": "<point>[[523, 14], [494, 133], [476, 282], [464, 422]]</point>"}]}
{"id": 6502, "image": ["rtx_frames_success_51/62_robo_set#episode_789/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7300/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[241, 485], [361, 498], [481, 542], [560, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[280, 512], [355, 558], [485, 598], [581, 598]]</point>"}]}
{"id": 6504, "image": ["rtx_frames_success_51/62_robo_set#episode_7314/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7330/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[269, 488], [389, 571], [496, 605], [594, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[256, 498], [382, 505], [466, 532], [572, 605]]</point>"}]}
{"id": 6506, "image": ["rtx_frames_success_51/62_robo_set#episode_7345/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7352/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 458], [446, 452], [611, 452], [807, 505]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 472], [387, 492], [493, 535], [579, 605]]</point>"}]}
{"id": 6508, "image": ["rtx_frames_success_51/62_robo_set#episode_7367/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7391/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[252, 472], [342, 399], [433, 352], [528, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[271, 478], [335, 505], [391, 545], [448, 618]]</point>"}]}
{"id": 6510, "image": ["rtx_frames_success_51/62_robo_set#episode_7391/frame_15.png", "rtx_frames_success_51/62_robo_set#episode_7392/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[461, 611], [521, 581], [583, 575], [653, 611]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the lid."}, {"from": "gpt", "value": "<point>[[480, 73], [583, 73], [677, 113], [748, 169]]</point>"}]}
{"id": 6512, "image": ["rtx_frames_success_51/62_robo_set#episode_7393/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7395/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[260, 458], [325, 555], [389, 598], [448, 605]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[297, 482], [384, 532], [476, 568], [574, 575]]</point>"}]}
{"id": 6514, "image": ["rtx_frames_success_51/62_robo_set#episode_7409/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7417/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[308, 472], [380, 385], [451, 342], [538, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[286, 478], [322, 455], [365, 445], [397, 472]]</point>"}]}
{"id": 6516, "image": ["rtx_frames_success_51/62_robo_set#episode_7417/frame_15.png", "rtx_frames_success_51/62_robo_set#episode_7424/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[376, 545], [372, 651], [380, 744], [401, 824]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[276, 488], [370, 548], [472, 595], [587, 611]]</point>"}]}
{"id": 6518, "image": ["rtx_frames_success_51/62_robo_set#episode_7429/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7431/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[273, 478], [429, 485], [547, 508], [703, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[278, 515], [414, 578], [564, 595], [707, 588]]</point>"}]}
{"id": 6520, "image": ["rtx_frames_success_51/62_robo_set#episode_7454/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7489/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[263, 468], [408, 538], [545, 568], [686, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the oven."}, {"from": "gpt", "value": "<point>[[248, 488], [397, 452], [523, 425], [679, 419]]</point>"}]}
{"id": 6522, "image": ["rtx_frames_success_51/62_robo_set#episode_7519/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7529/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[301, 458], [320, 561], [352, 631], [406, 684]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[293, 465], [316, 542], [370, 578], [417, 585]]</point>"}]}
{"id": 6524, "image": ["rtx_frames_success_51/62_robo_set#episode_7534/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7572/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[305, 445], [400, 449], [521, 498], [602, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[274, 462], [340, 366], [432, 329], [519, 322]]</point>"}]}
{"id": 6526, "image": ["rtx_frames_success_51/62_robo_set#episode_7592/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7624/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[254, 485], [348, 505], [476, 542], [541, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[263, 485], [301, 452], [346, 442], [382, 465]]</point>"}]}
{"id": 6528, "image": ["rtx_frames_success_51/62_robo_set#episode_7624/frame_15.png", "rtx_frames_success_51/62_robo_set#episode_7641/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the strainer backwards."}, {"from": "gpt", "value": "<point>[[376, 548], [367, 631], [370, 744], [387, 814]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[269, 465], [384, 472], [481, 522], [570, 605]]</point>"}]}
{"id": 6530, "image": ["rtx_frames_success_51/62_robo_set#episode_7674/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7677/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 482], [385, 538], [489, 585], [583, 595]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[308, 485], [434, 478], [511, 505], [605, 588]]</point>"}]}
{"id": 6532, "image": ["rtx_frames_success_51/62_robo_set#episode_7685/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7691/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[280, 472], [459, 435], [628, 432], [825, 445]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[316, 452], [331, 535], [370, 621], [410, 678]]</point>"}]}
{"id": 6534, "image": ["rtx_frames_success_51/62_robo_set#episode_7694/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7694/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 465], [314, 532], [346, 575], [387, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[359, 581], [363, 498], [370, 405], [383, 332]]</point>"}]}
{"id": 6536, "image": ["rtx_frames_success_51/62_robo_set#episode_7705/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7770/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle inside the container."}, {"from": "gpt", "value": "<point>[[273, 468], [305, 551], [344, 611], [398, 658]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 495], [374, 485], [478, 492], [570, 542]]</point>"}]}
{"id": 6538, "image": ["rtx_frames_success_51/62_robo_set#episode_7780/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7794/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 488], [464, 465], [637, 482], [833, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[274, 478], [365, 538], [470, 581], [570, 605]]</point>"}]}
{"id": 6540, "image": ["rtx_frames_success_51/62_robo_set#episode_7814/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7819/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[282, 449], [400, 452], [541, 518], [633, 611]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[282, 449], [361, 402], [459, 362], [536, 352]]</point>"}]}
{"id": 6542, "image": ["rtx_frames_success_51/62_robo_set#episode_7837/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7838/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[297, 492], [400, 482], [536, 502], [617, 532]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[295, 472], [357, 412], [453, 349], [534, 326]]</point>"}]}
{"id": 6544, "image": ["rtx_frames_success_51/62_robo_set#episode_7842/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7855/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[258, 462], [368, 458], [474, 492], [562, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 465], [376, 475], [464, 518], [539, 575]]</point>"}]}
{"id": 6546, "image": ["rtx_frames_success_51/62_robo_set#episode_7859/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7871/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[263, 502], [372, 505], [485, 528], [583, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 498], [363, 555], [470, 591], [587, 611]]</point>"}]}
{"id": 6548, "image": ["rtx_frames_success_51/62_robo_set#episode_7877/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7889/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven lid."}, {"from": "gpt", "value": "<point>[[260, 505], [389, 449], [532, 412], [677, 409]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[263, 485], [378, 458], [496, 472], [598, 558]]</point>"}]}
{"id": 6550, "image": ["rtx_frames_success_51/62_robo_set#episode_7894/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7917/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[282, 458], [361, 382], [459, 329], [519, 312]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[282, 485], [404, 425], [508, 342], [615, 226]]</point>"}]}
{"id": 6552, "image": ["rtx_frames_success_51/62_robo_set#episode_7917/frame_15.png", "rtx_frames_success_51/62_robo_set#episode_7922/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[630, 282], [626, 435], [637, 608], [656, 761]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[288, 475], [359, 512], [455, 548], [572, 561]]</point>"}]}
{"id": 6554, "image": ["rtx_frames_success_51/62_robo_set#episode_7924/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7958/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[316, 472], [340, 545], [376, 605], [414, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[271, 485], [338, 429], [436, 382], [528, 359]]</point>"}]}
{"id": 6556, "image": ["rtx_frames_success_51/62_robo_set#episode_7959/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7959/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[276, 445], [301, 508], [350, 558], [391, 568]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forwards."}, {"from": "gpt", "value": "<point>[[365, 542], [370, 435], [370, 369], [329, 372], [286, 375]]</point>"}]}
{"id": 6558, "image": ["rtx_frames_success_51/62_robo_set#episode_7961/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7965/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the tray."}, {"from": "gpt", "value": "<point>[[248, 472], [438, 472], [671, 508], [861, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[256, 472], [416, 468], [562, 488], [688, 525]]</point>"}]}
{"id": 6560, "image": ["rtx_frames_success_51/62_robo_set#episode_7970/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7979/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[258, 475], [436, 458], [662, 472], [844, 512]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 492], [331, 515], [401, 565], [448, 621]]</point>"}]}
{"id": 6562, "image": ["rtx_frames_success_51/62_robo_set#episode_7981/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_7982/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[293, 452], [357, 379], [431, 336], [519, 332]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 465], [372, 545], [476, 601], [586, 638]]</point>"}]}
{"id": 6564, "image": ["rtx_frames_success_51/62_robo_set#episode_8015/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_8045/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[233, 452], [352, 478], [463, 532], [568, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[260, 488], [387, 488], [574, 518], [701, 568]]</point>"}]}
{"id": 6566, "image": ["rtx_frames_success_51/62_robo_set#episode_8048/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_8050/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[274, 465], [352, 528], [464, 585], [553, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[263, 492], [316, 551], [363, 591], [421, 625]]</point>"}]}
{"id": 6568, "image": ["rtx_frames_success_51/62_robo_set#episode_8051/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_8051/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[261, 465], [312, 442], [367, 445], [397, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[387, 548], [382, 638], [383, 741], [395, 830]]</point>"}]}
{"id": 6570, "image": ["rtx_frames_success_51/62_robo_set#episode_8076/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_8085/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[256, 492], [417, 492], [560, 515], [684, 548]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[259, 485], [402, 405], [508, 332], [635, 226]]</point>"}]}
{"id": 6572, "image": ["rtx_frames_success_51/62_robo_set#episode_8085/frame_15.png", "rtx_frames_success_51/62_robo_set#episode_8086/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[624, 286], [618, 389], [630, 585], [647, 704]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[286, 478], [391, 379], [506, 269], [621, 216]]</point>"}]}
{"id": 6574, "image": ["rtx_frames_success_51/62_robo_set#episode_8086/frame_15.png", "rtx_frames_success_51/62_robo_set#episode_8090/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[626, 282], [651, 425], [660, 575], [662, 738]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[261, 485], [374, 462], [506, 485], [609, 548]]</point>"}]}
{"id": 6576, "image": ["rtx_frames_success_51/62_robo_set#episode_8112/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_8126/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[244, 478], [380, 472], [487, 498], [587, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid of the tea container."}, {"from": "gpt", "value": "<point>[[500, 17], [598, 60], [685, 146], [728, 292]]</point>"}]}
{"id": 6578, "image": ["rtx_frames_success_51/62_robo_set#episode_8129/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_8156/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[261, 452], [376, 482], [479, 515], [558, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[259, 472], [393, 485], [517, 522], [620, 561]]</point>"}]}
{"id": 6580, "image": ["rtx_frames_success_51/62_robo_set#episode_8177/frame_0.png", "rtx_frames_success_51/62_robo_set#episode_8178/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[242, 498], [378, 488], [521, 515], [635, 558]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[239, 455], [455, 435], [617, 432], [795, 445]]</point>"}]}
{"id": 6582, "image": ["rtx_frames_success_52/62_robo_set#episode_845/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_846/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[284, 482], [398, 542], [524, 565], [647, 565]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[242, 468], [423, 538], [600, 568], [827, 578]]</point>"}]}
{"id": 6584, "image": ["rtx_frames_success_52/62_robo_set#episode_854/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_862/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[235, 468], [353, 497], [471, 557], [605, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[311, 475], [395, 482], [471, 527], [576, 572]]</point>"}]}
{"id": 6586, "image": ["rtx_frames_success_52/62_robo_set#episode_864/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_865/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[294, 482], [403, 520], [504, 601], [601, 616]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[261, 497], [391, 512], [513, 527], [622, 564]]</point>"}]}
{"id": 6588, "image": ["rtx_frames_success_52/62_robo_set#episode_873/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_873/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[298, 482], [416, 408], [508, 304], [622, 171]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[597, 275], [597, 401], [605, 527], [613, 638]]</point>"}]}
{"id": 6590, "image": ["rtx_frames_success_52/62_robo_set#episode_874/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_884/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[311, 490], [424, 482], [521, 482], [613, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[294, 482], [395, 482], [487, 497], [584, 542]]</point>"}]}
{"id": 6592, "image": ["rtx_frames_success_52/62_robo_set#episode_885/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_889/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[319, 482], [370, 497], [416, 557], [416, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[298, 490], [399, 490], [496, 527], [605, 579]]</point>"}]}
{"id": 6594, "image": ["rtx_frames_success_52/62_robo_set#episode_893/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_894/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[286, 557], [429, 497], [588, 497], [739, 594]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[307, 527], [366, 527], [403, 557], [437, 631]]</point>"}]}
{"id": 6596, "image": ["rtx_frames_success_52/62_robo_set#episode_906/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8187/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[294, 482], [370, 393], [462, 327], [538, 312]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the drawer."}, {"from": "gpt", "value": "<point>[[493, 306], [470, 382], [406, 505], [354, 565]]</point>"}]}
{"id": 6598, "image": ["rtx_frames_success_52/62_robo_set#episode_8192/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8192/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[276, 458], [314, 465], [361, 458], [391, 439]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[378, 528], [355, 611], [357, 718], [386, 824]]</point>"}]}
{"id": 6600, "image": ["rtx_frames_success_52/62_robo_set#episode_8200/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8206/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[290, 475], [414, 528], [564, 542], [698, 515]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[271, 455], [350, 525], [448, 568], [570, 578]]</point>"}]}
{"id": 6602, "image": ["rtx_frames_success_52/62_robo_set#episode_8215/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8237/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[291, 465], [372, 522], [470, 545], [572, 545]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[301, 488], [357, 412], [434, 352], [513, 322]]</point>"}]}
{"id": 6604, "image": ["rtx_frames_success_52/62_robo_set#episode_8238/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8250/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 472], [359, 551], [451, 585], [598, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 488], [305, 442], [361, 442], [380, 465]]</point>"}]}
{"id": 6606, "image": ["rtx_frames_success_52/62_robo_set#episode_8250/frame_15.png", "rtx_frames_success_52/62_robo_set#episode_8255/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[374, 518], [357, 571], [361, 704], [384, 767]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[237, 502], [402, 538], [649, 532], [816, 488]]</point>"}]}
{"id": 6608, "image": ["rtx_frames_success_52/62_robo_set#episode_8258/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8261/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[282, 488], [376, 542], [481, 565], [611, 571]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[305, 465], [389, 558], [478, 615], [592, 651]]</point>"}]}
{"id": 6610, "image": ["rtx_frames_success_52/62_robo_set#episode_8274/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8280/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 465], [350, 525], [457, 568], [579, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[246, 522], [380, 561], [542, 558], [686, 515]]</point>"}]}
{"id": 6612, "image": ["rtx_frames_success_52/62_robo_set#episode_8286/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8288/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[271, 488], [308, 538], [386, 588], [425, 598]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[258, 492], [395, 445], [564, 452], [685, 512]]</point>"}]}
{"id": 6614, "image": ["rtx_frames_success_52/62_robo_set#episode_8292/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8293/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[282, 465], [355, 389], [429, 332], [517, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[263, 482], [342, 369], [463, 266], [606, 223]]</point>"}]}
{"id": 6616, "image": ["rtx_frames_success_52/62_robo_set#episode_8293/frame_15.png", "rtx_frames_success_52/62_robo_set#episode_8300/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backward."}, {"from": "gpt", "value": "<point>[[609, 302], [589, 439], [592, 618], [622, 738]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[256, 485], [376, 492], [556, 578], [641, 668]]</point>"}]}
{"id": 6618, "image": ["rtx_frames_success_52/62_robo_set#episode_8303/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8315/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[256, 458], [417, 532], [609, 555], [831, 532]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[252, 458], [355, 532], [474, 565], [615, 565]]</point>"}]}
{"id": 6620, "image": ["rtx_frames_success_52/62_robo_set#episode_8319/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8319/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[252, 472], [365, 359], [500, 259], [626, 219]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[577, 253], [566, 382], [571, 522], [602, 668]]</point>"}]}
{"id": 6622, "image": ["rtx_frames_success_52/62_robo_set#episode_8326/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8327/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[271, 478], [367, 558], [474, 591], [575, 605]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[284, 478], [325, 445], [372, 445], [395, 478]]</point>"}]}
{"id": 6624, "image": ["rtx_frames_success_52/62_robo_set#episode_8327/frame_15.png", "rtx_frames_success_52/62_robo_set#episode_8332/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer backwards."}, {"from": "gpt", "value": "<point>[[376, 551], [359, 605], [363, 691], [385, 754]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[276, 465], [400, 395], [515, 279], [577, 173]]</point>"}]}
{"id": 6626, "image": ["rtx_frames_success_52/62_robo_set#episode_8332/frame_15.png", "rtx_frames_success_52/62_robo_set#episode_8344/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backward."}, {"from": "gpt", "value": "<point>[[585, 253], [571, 395], [577, 535], [618, 664]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[271, 468], [357, 545], [483, 571], [594, 542]]</point>"}]}
{"id": 6628, "image": ["rtx_frames_success_52/62_robo_set#episode_8351/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8353/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[254, 485], [320, 402], [414, 359], [525, 349]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[265, 468], [293, 542], [335, 598], [389, 644]]</point>"}]}
{"id": 6630, "image": ["rtx_frames_success_52/62_robo_set#episode_8363/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8377/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[273, 488], [318, 512], [354, 505], [387, 492]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[275, 465], [397, 395], [513, 296], [611, 193]]</point>"}]}
{"id": 6632, "image": ["rtx_frames_success_52/62_robo_set#episode_8377/frame_15.png", "rtx_frames_success_52/62_robo_set#episode_8381/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[600, 213], [587, 362], [604, 532], [649, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[280, 485], [320, 545], [380, 598], [444, 621]]</point>"}]}
{"id": 6634, "image": ["rtx_frames_success_52/62_robo_set#episode_8381/frame_15.png", "rtx_frames_success_52/62_robo_set#episode_8385/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[457, 625], [500, 585], [547, 578], [609, 591]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[284, 505], [380, 551], [476, 565], [596, 565]]</point>"}]}
{"id": 6636, "image": ["rtx_frames_success_52/62_robo_set#episode_8400/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8404/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[273, 478], [299, 568], [350, 628], [399, 651]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[271, 465], [306, 545], [353, 615], [410, 651]]</point>"}]}
{"id": 6638, "image": ["rtx_frames_success_52/62_robo_set#episode_8410/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8415/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[295, 482], [378, 542], [472, 595], [602, 635]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[267, 485], [331, 542], [445, 595], [543, 618]]</point>"}]}
{"id": 6640, "image": ["rtx_frames_success_52/62_robo_set#episode_8418/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8418/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[284, 462], [322, 538], [352, 588], [395, 628]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[378, 551], [382, 482], [378, 412], [359, 359]]</point>"}]}
{"id": 6642, "image": ["rtx_frames_success_52/62_robo_set#episode_8425/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8430/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the oven."}, {"from": "gpt", "value": "<point>[[260, 488], [402, 561], [564, 598], [797, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[252, 478], [414, 465], [577, 495], [760, 551]]</point>"}]}
{"id": 6644, "image": ["rtx_frames_success_52/62_robo_set#episode_8432/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8439/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 485], [374, 465], [506, 485], [613, 535]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[327, 442], [374, 432], [433, 372], [453, 319]]</point>"}]}
{"id": 6646, "image": ["rtx_frames_success_52/62_robo_set#episode_8440/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8450/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 492], [355, 535], [459, 561], [557, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[274, 505], [335, 405], [427, 349], [528, 326]]</point>"}]}
{"id": 6648, "image": ["rtx_frames_success_52/62_robo_set#episode_8488/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8493/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[289, 532], [353, 429], [414, 366], [515, 302]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[278, 485], [382, 478], [483, 512], [590, 571]]</point>"}]}
{"id": 6650, "image": ["rtx_frames_success_52/62_robo_set#episode_8497/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8499/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[271, 492], [400, 558], [613, 581], [782, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[271, 485], [397, 505], [513, 568], [635, 658]]</point>"}]}
{"id": 6652, "image": ["rtx_frames_success_52/62_robo_set#episode_8505/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8525/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[267, 505], [397, 548], [547, 571], [694, 555]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 465], [374, 525], [485, 545], [602, 545]]</point>"}]}
{"id": 6654, "image": ["rtx_frames_success_52/62_robo_set#episode_8527/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8528/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[299, 458], [312, 528], [336, 575], [385, 615]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[259, 442], [306, 508], [370, 568], [419, 568]]</point>"}]}
{"id": 6656, "image": ["rtx_frames_success_52/62_robo_set#episode_8541/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8541/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[286, 453], [403, 475], [525, 512], [622, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move the banana towards the plate."}, {"from": "gpt", "value": "<point>[[559, 579], [534, 453], [534, 319], [529, 178]]</point>"}]}
{"id": 6658, "image": ["rtx_frames_success_52/62_robo_set#episode_8544/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8548/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[269, 497], [319, 542], [361, 594], [403, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[265, 445], [319, 542], [353, 586], [395, 675]]</point>"}]}
{"id": 6660, "image": ["rtx_frames_success_52/62_robo_set#episode_8556/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8567/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[286, 468], [370, 468], [454, 482], [576, 534]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[273, 468], [391, 490], [496, 512], [580, 549]]</point>"}]}
{"id": 6662, "image": ["rtx_frames_success_52/62_robo_set#episode_8570/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8584/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[290, 505], [399, 505], [492, 557], [618, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[273, 512], [382, 512], [500, 512], [605, 557]]</point>"}]}
{"id": 6664, "image": ["rtx_frames_success_52/62_robo_set#episode_8618/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8626/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[252, 460], [345, 416], [450, 341], [521, 289]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[256, 490], [374, 431], [483, 349], [630, 238]]</point>"}]}
{"id": 6666, "image": ["rtx_frames_success_52/62_robo_set#episode_8660/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8662/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[311, 497], [345, 557], [370, 586], [403, 624]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[286, 520], [336, 527], [387, 572], [437, 631]]</point>"}]}
{"id": 6668, "image": ["rtx_frames_success_52/62_robo_set#episode_8672/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8680/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[345, 475], [441, 527], [525, 564], [626, 690]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 482], [370, 512], [408, 527], [408, 638]]</point>"}]}
{"id": 6670, "image": ["rtx_frames_success_52/62_robo_set#episode_8688/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8694/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[286, 534], [319, 505], [361, 438], [395, 379]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[319, 475], [382, 386], [445, 334], [534, 289]]</point>"}]}
{"id": 6672, "image": ["rtx_frames_success_52/62_robo_set#episode_8701/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8710/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[261, 482], [395, 482], [517, 497], [647, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[311, 490], [441, 520], [513, 586], [613, 653]]</point>"}]}
{"id": 6674, "image": ["rtx_frames_success_52/62_robo_set#episode_8714/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8725/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[282, 512], [399, 482], [538, 490], [769, 468]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[315, 512], [424, 512], [496, 512], [588, 542]]</point>"}]}
{"id": 6676, "image": ["rtx_frames_success_52/62_robo_set#episode_8728/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8738/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[311, 542], [366, 564], [416, 586], [429, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 453], [361, 527], [395, 542], [424, 653]]</point>"}]}
{"id": 6678, "image": ["rtx_frames_success_52/62_robo_set#episode_8768/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8771/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[311, 482], [429, 497], [538, 520], [769, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[307, 505], [387, 438], [445, 393], [555, 349]]</point>"}]}
{"id": 6680, "image": ["rtx_frames_success_52/62_robo_set#episode_8777/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8792/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[324, 460], [408, 386], [471, 349], [534, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[357, 453], [391, 512], [408, 557], [412, 653]]</point>"}]}
{"id": 6682, "image": ["rtx_frames_success_52/62_robo_set#episode_8798/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8811/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[261, 505], [357, 497], [471, 497], [580, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[336, 475], [429, 438], [496, 512], [576, 557]]</point>"}]}
{"id": 6684, "image": ["rtx_frames_success_52/62_robo_set#episode_8812/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8825/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[303, 475], [387, 356], [462, 289], [550, 282]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[294, 453], [387, 453], [483, 460], [630, 572]]</point>"}]}
{"id": 6686, "image": ["rtx_frames_success_52/62_robo_set#episode_8830/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8843/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[319, 431], [420, 431], [496, 482], [571, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[256, 475], [361, 401], [454, 327], [563, 289]]</point>"}]}
{"id": 6688, "image": ["rtx_frames_success_52/62_robo_set#episode_8868/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8883/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[336, 505], [441, 520], [529, 586], [647, 727]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[303, 431], [387, 386], [466, 334], [550, 304]]</point>"}]}
{"id": 6690, "image": ["rtx_frames_success_52/62_robo_set#episode_8886/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8895/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[298, 468], [353, 512], [382, 534], [412, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[286, 512], [416, 512], [529, 512], [710, 527]]</point>"}]}
{"id": 6692, "image": ["rtx_frames_success_52/62_robo_set#episode_8901/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8907/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[290, 445], [370, 393], [437, 334], [529, 334]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[307, 423], [412, 379], [492, 334], [563, 282]]</point>"}]}
{"id": 6694, "image": ["rtx_frames_success_52/62_robo_set#episode_8912/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8912/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[328, 482], [378, 564], [408, 631], [408, 705]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer forward."}, {"from": "gpt", "value": "<point>[[374, 542], [370, 453], [353, 379], [357, 312]]</point>"}]}
{"id": 6696, "image": ["rtx_frames_success_52/62_robo_set#episode_8919/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8923/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[319, 505], [412, 512], [504, 534], [630, 646]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[315, 512], [395, 512], [504, 572], [597, 624]]</point>"}]}
{"id": 6698, "image": ["rtx_frames_success_52/62_robo_set#episode_8931/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8948/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[307, 482], [403, 497], [496, 520], [622, 586]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[303, 497], [445, 490], [563, 497], [765, 527]]</point>"}]}
{"id": 6700, "image": ["rtx_frames_success_52/62_robo_set#episode_8960/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8962/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[303, 564], [441, 534], [597, 512], [798, 594]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[328, 490], [429, 482], [500, 534], [588, 638]]</point>"}]}
{"id": 6702, "image": ["rtx_frames_success_52/62_robo_set#episode_8967/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8967/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[286, 468], [395, 364], [513, 238], [613, 141]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the mug backwards."}, {"from": "gpt", "value": "<point>[[605, 230], [613, 408], [626, 542], [622, 690]]</point>"}]}
{"id": 6704, "image": ["rtx_frames_success_52/62_robo_set#episode_8968/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8975/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[311, 520], [445, 512], [571, 512], [727, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the tray."}, {"from": "gpt", "value": "<point>[[315, 512], [475, 475], [613, 512], [798, 572]]</point>"}]}
{"id": 6706, "image": ["rtx_frames_success_52/62_robo_set#episode_8985/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8987/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[315, 512], [420, 482], [496, 497], [613, 542]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[290, 497], [433, 475], [567, 475], [765, 453]]</point>"}]}
{"id": 6708, "image": ["rtx_frames_success_52/62_robo_set#episode_8991/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8991/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[286, 512], [345, 460], [378, 379], [412, 393]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the strainer backwards."}, {"from": "gpt", "value": "<point>[[370, 512], [382, 609], [403, 690], [420, 794]]</point>"}]}
{"id": 6710, "image": ["rtx_frames_success_52/62_robo_set#episode_8992/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_8996/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[319, 542], [416, 520], [508, 549], [622, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[349, 512], [471, 527], [571, 631], [647, 727]]</point>"}]}
{"id": 6712, "image": ["rtx_frames_success_52/62_robo_set#episode_8999/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_9000/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the mug."}, {"from": "gpt", "value": "<point>[[336, 468], [416, 482], [504, 542], [588, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[340, 527], [437, 527], [521, 557], [597, 586]]</point>"}]}
{"id": 6714, "image": ["rtx_frames_success_52/62_robo_set#episode_9001/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_9008/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[345, 512], [420, 512], [496, 534], [597, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[303, 468], [374, 505], [412, 557], [429, 616]]</point>"}]}
{"id": 6716, "image": ["rtx_frames_success_52/62_robo_set#episode_9028/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_9031/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[336, 512], [424, 497], [492, 549], [576, 631]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[483, 37], [475, 156], [441, 334], [412, 512]]</point>"}]}
{"id": 6718, "image": ["rtx_frames_success_52/62_robo_set#episode_9041/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_9043/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[307, 482], [458, 460], [605, 482], [782, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[307, 490], [387, 497], [487, 527], [571, 579]]</point>"}]}
{"id": 6720, "image": ["rtx_frames_success_52/62_robo_set#episode_9046/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_9048/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[303, 512], [433, 512], [534, 534], [634, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[303, 475], [454, 453], [597, 453], [761, 482]]</point>"}]}
{"id": 6722, "image": ["rtx_frames_success_52/62_robo_set#episode_9058/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_9060/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[265, 468], [378, 512], [483, 512], [588, 564]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach towards the drawer."}, {"from": "gpt", "value": "<point>[[521, 468], [445, 520], [366, 572], [282, 631]]</point>"}]}
{"id": 6724, "image": ["rtx_frames_success_52/62_robo_set#episode_9063/frame_0.png", "rtx_frames_success_52/62_robo_set#episode_9064/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[277, 512], [420, 512], [525, 512], [647, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 454], [402, 541], [514, 531], [705, 511]]</point>"}]}
{"id": 6726, "image": ["rtx_frames_success_53/62_robo_set#episode_96/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_910/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[315, 449], [413, 526], [503, 572], [584, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[327, 465], [396, 551], [480, 582], [578, 582]]</point>"}]}
{"id": 6728, "image": ["rtx_frames_success_53/62_robo_set#episode_911/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_946/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[321, 475], [436, 531], [543, 541], [630, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[569, 429], [500, 495], [442, 546], [353, 603]]</point>"}]}
{"id": 6730, "image": ["rtx_frames_success_53/62_robo_set#episode_947/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_964/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[341, 465], [399, 414], [457, 363], [538, 281]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[301, 449], [413, 516], [471, 562], [549, 562]]</point>"}]}
{"id": 6732, "image": ["rtx_frames_success_53/62_robo_set#episode_968/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_987/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[275, 498], [387, 505], [523, 528], [628, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pick up the lid."}, {"from": "gpt", "value": "<point>[[491, 33], [577, 50], [685, 133], [741, 246]]</point>"}]}
{"id": 6734, "image": ["rtx_frames_success_53/62_robo_set#episode_988/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9116/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[312, 472], [399, 571], [470, 611], [560, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the banana on the plate."}, {"from": "gpt", "value": "<point>[[318, 475], [373, 434], [439, 398], [529, 347]]</point>"}]}
{"id": 6736, "image": ["rtx_frames_success_53/62_robo_set#episode_9122/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9123/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 439], [347, 546], [370, 587], [410, 643]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 454], [376, 516], [445, 557], [587, 572]]</point>"}]}
{"id": 6738, "image": ["rtx_frames_success_53/62_robo_set#episode_9134/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9137/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[318, 470], [410, 526], [526, 572], [633, 597], [763, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[309, 480], [410, 577], [506, 603], [627, 572]]</point>"}]}
{"id": 6740, "image": ["rtx_frames_success_53/62_robo_set#episode_9143/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9162/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[315, 465], [410, 536], [532, 582], [645, 608], [780, 613]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[503, 414], [431, 470], [364, 551], [312, 613]]</point>"}]}
{"id": 6742, "image": ["rtx_frames_success_53/62_robo_set#episode_9165/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9179/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[292, 465], [382, 434], [468, 393], [532, 347]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 475], [387, 521], [471, 562], [558, 592]]</point>"}]}
{"id": 6744, "image": ["rtx_frames_success_53/62_robo_set#episode_9183/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9192/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[303, 485], [393, 587], [491, 603], [613, 603]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[329, 454], [503, 454], [647, 460], [829, 460]]</point>"}]}
{"id": 6746, "image": ["rtx_frames_success_53/62_robo_set#episode_9202/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9209/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[312, 470], [402, 516], [480, 541], [581, 557]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to approach the open drawer."}, {"from": "gpt", "value": "<point>[[462, 276], [410, 378], [364, 470], [315, 551]]</point>"}]}
{"id": 6748, "image": ["rtx_frames_success_53/62_robo_set#episode_9215/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9222/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[318, 449], [399, 516], [514, 551], [604, 562]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[303, 470], [422, 541], [497, 562], [566, 582]]</point>"}]}
{"id": 6750, "image": ["rtx_frames_success_53/62_robo_set#episode_9223/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9224/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[295, 470], [353, 536], [439, 582], [503, 587], [584, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[303, 480], [396, 526], [471, 531], [552, 541]]</point>"}]}
{"id": 6752, "image": ["rtx_frames_success_53/62_robo_set#episode_9233/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9241/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[347, 460], [439, 546], [497, 577], [572, 613], [642, 648]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[301, 454], [382, 398], [457, 368], [526, 322]]</point>"}]}
{"id": 6754, "image": ["rtx_frames_success_53/62_robo_set#episode_9245/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9250/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[332, 490], [439, 577], [529, 628], [653, 654]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[298, 449], [410, 521], [529, 541], [645, 572], [801, 557]]</point>"}]}
{"id": 6756, "image": ["rtx_frames_success_53/62_robo_set#episode_9267/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9271/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana peel."}, {"from": "gpt", "value": "<point>[[298, 460], [419, 526], [517, 557], [607, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[298, 465], [384, 531], [460, 536], [561, 551]]</point>"}]}
{"id": 6758, "image": ["rtx_frames_success_53/62_robo_set#episode_9273/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9288/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 475], [436, 511], [520, 531], [595, 557], [766, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to push the drawer inwards."}, {"from": "gpt", "value": "<point>[[442, 255], [367, 368], [324, 424], [257, 511]]</point>"}]}
{"id": 6760, "image": ["rtx_frames_success_53/62_robo_set#episode_9292/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9300/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 465], [382, 531], [480, 587], [601, 608]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[324, 454], [399, 393], [520, 306]]</point>"}]}
{"id": 6762, "image": ["rtx_frames_success_53/62_robo_set#episode_9306/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9313/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[318, 475], [428, 562], [514, 582], [584, 603]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[309, 470], [425, 541], [497, 567], [558, 557]]</point>"}]}
{"id": 6764, "image": ["rtx_frames_success_53/62_robo_set#episode_9321/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9322/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[321, 465], [387, 393], [471, 327], [552, 276]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[312, 470], [390, 526], [474, 562], [552, 592], [671, 597], [803, 587]]</point>"}]}
{"id": 6766, "image": ["rtx_frames_success_53/62_robo_set#episode_9330/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9331/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[315, 460], [373, 444], [460, 419], [540, 342]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[306, 470], [405, 511], [474, 521], [566, 521]]</point>"}]}
{"id": 6768, "image": ["rtx_frames_success_53/62_robo_set#episode_9332/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9332/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[306, 480], [419, 511], [509, 531], [688, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to open the oven."}, {"from": "gpt", "value": "<point>[[699, 551], [624, 521], [523, 490], [434, 470]]</point>"}]}
{"id": 6770, "image": ["rtx_frames_success_53/62_robo_set#episode_9339/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9342/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[321, 465], [416, 511], [480, 531], [592, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[318, 505], [390, 521], [474, 536], [561, 551]]</point>"}]}
{"id": 6772, "image": ["rtx_frames_success_53/62_robo_set#episode_9351/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9351/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[318, 470], [413, 388], [520, 306], [639, 184]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[592, 306], [592, 490], [592, 613], [607, 761]]</point>"}]}
{"id": 6774, "image": ["rtx_frames_success_53/62_robo_set#episode_9352/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9355/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[318, 460], [399, 500], [488, 541], [566, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[327, 475], [410, 557], [523, 608], [621, 669]]</point>"}]}
{"id": 6776, "image": ["rtx_frames_success_53/62_robo_set#episode_9365/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9372/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[341, 470], [416, 485], [520, 490], [601, 500], [679, 500]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[341, 465], [358, 546], [382, 608], [408, 689]]</point>"}]}
{"id": 6778, "image": ["rtx_frames_success_53/62_robo_set#episode_9372/frame_15.png", "rtx_frames_success_53/62_robo_set#episode_9381/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[379, 756], [442, 638], [503, 562], [540, 531]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle."}, {"from": "gpt", "value": "<point>[[350, 444], [350, 531], [370, 587], [410, 654]]</point>"}]}
{"id": 6780, "image": ["rtx_frames_success_53/62_robo_set#episode_9383/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9414/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[327, 460], [439, 541], [506, 582], [555, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[303, 470], [393, 521], [503, 541], [604, 567]]</point>"}]}
{"id": 6782, "image": ["rtx_frames_success_53/62_robo_set#episode_9415/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9422/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the table."}, {"from": "gpt", "value": "<point>[[321, 465], [448, 516], [520, 536], [607, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[318, 511], [396, 398], [457, 296]]</point>"}]}
{"id": 6784, "image": ["rtx_frames_success_53/62_robo_set#episode_9424/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9425/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[329, 470], [408, 541], [486, 572], [618, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[329, 470], [410, 531], [468, 582], [529, 603], [590, 633], [656, 664]]</point>"}]}
{"id": 6786, "image": ["rtx_frames_success_53/62_robo_set#episode_9429/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9434/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[318, 449], [405, 536], [497, 551], [587, 562]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[329, 449], [428, 516], [503, 526], [627, 541]]</point>"}]}
{"id": 6788, "image": ["rtx_frames_success_53/62_robo_set#episode_9446/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9448/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[335, 460], [410, 526], [474, 551], [566, 582]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[315, 465], [390, 393], [451, 337], [517, 306]]</point>"}]}
{"id": 6790, "image": ["rtx_frames_success_53/62_robo_set#episode_9453/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9459/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[321, 470], [370, 557], [431, 597], [517, 592], [584, 592]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[309, 449], [399, 480], [460, 511], [543, 557]]</point>"}]}
{"id": 6792, "image": ["rtx_frames_success_53/62_robo_set#episode_9468/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9488/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[312, 475], [410, 511], [494, 541], [592, 551]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[321, 470], [382, 434], [468, 388], [535, 347]]</point>"}]}
{"id": 6794, "image": ["rtx_frames_success_53/62_robo_set#episode_9497/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9498/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[318, 460], [399, 490], [514, 516], [621, 567]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[318, 470], [405, 541], [494, 577], [587, 613]]</point>"}]}
{"id": 6796, "image": ["rtx_frames_success_53/62_robo_set#episode_9505/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9526/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[335, 470], [434, 531], [514, 551], [598, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[312, 475], [425, 526], [538, 562], [630, 582], [789, 608]]</point>"}]}
{"id": 6798, "image": ["rtx_frames_success_53/62_robo_set#episode_9564/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9565/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[318, 449], [413, 403], [535, 327]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[355, 460], [419, 516], [491, 557], [549, 597], [639, 664]]</point>"}]}
{"id": 6800, "image": ["rtx_frames_success_53/62_robo_set#episode_9568/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9576/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[312, 460], [390, 434], [468, 398], [552, 337]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[324, 449], [382, 541], [462, 582], [601, 608]]</point>"}]}
{"id": 6802, "image": ["rtx_frames_success_53/62_robo_set#episode_9577/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9583/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[327, 475], [390, 516], [445, 536], [517, 557], [607, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[327, 485], [428, 577], [566, 597], [647, 608], [780, 597]]</point>"}]}
{"id": 6804, "image": ["rtx_frames_success_53/62_robo_set#episode_9624/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9632/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[335, 460], [442, 505], [549, 546], [630, 567], [815, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer handle."}, {"from": "gpt", "value": "<point>[[448, 260], [390, 357], [318, 434], [280, 475]]</point>"}]}
{"id": 6806, "image": ["rtx_frames_success_53/62_robo_set#episode_9635/frame_15.png", "rtx_frames_success_53/62_robo_set#episode_9641/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the ketchup bottle inside the container."}, {"from": "gpt", "value": "<point>[[376, 684], [465, 659], [549, 638], [613, 638]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the drawer."}, {"from": "gpt", "value": "<point>[[399, 296], [350, 408], [309, 475]]</point>"}]}
{"id": 6808, "image": ["rtx_frames_success_53/62_robo_set#episode_9656/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9667/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana on the plate."}, {"from": "gpt", "value": "<point>[[329, 475], [390, 429], [448, 383], [535, 327]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[332, 475], [428, 511], [483, 521], [581, 546]]</point>"}]}
{"id": 6810, "image": ["rtx_frames_success_53/62_robo_set#episode_9671/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9673/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[301, 460], [399, 521], [471, 546], [549, 572], [607, 577]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[344, 449], [405, 490], [460, 521], [517, 526], [575, 557]]</point>"}]}
{"id": 6812, "image": ["rtx_frames_success_53/62_robo_set#episode_9675/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9686/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 458], [312, 548], [346, 598], [391, 615]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[276, 492], [316, 555], [374, 601], [427, 611]]</point>"}]}
{"id": 6814, "image": ["rtx_frames_success_53/62_robo_set#episode_9686/frame_15.png", "rtx_frames_success_53/62_robo_set#episode_9687/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to drag the strainer from left to right."}, {"from": "gpt", "value": "<point>[[436, 598], [491, 568], [560, 575], [615, 595], [551, 635], [476, 641], [378, 625]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the bottle on the plate."}, {"from": "gpt", "value": "<point>[[244, 475], [354, 342], [440, 292], [549, 289]]</point>"}]}
{"id": 6816, "image": ["rtx_frames_success_53/62_robo_set#episode_9691/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9691/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the bowl."}, {"from": "gpt", "value": "<point>[[301, 478], [346, 502], [384, 555], [406, 611]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to release the bowl."}, {"from": "gpt", "value": "<point>[[399, 608], [461, 641], [517, 638], [575, 611], [483, 538], [391, 525]]</point>"}]}
{"id": 6818, "image": ["rtx_frames_success_53/62_robo_set#episode_9693/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9696/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[308, 498], [372, 508], [414, 555], [457, 635]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[282, 488], [395, 515], [513, 575], [590, 641]]</point>"}]}
{"id": 6820, "image": ["rtx_frames_success_53/62_robo_set#episode_9702/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9704/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[305, 478], [365, 385], [440, 332], [519, 319]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[269, 482], [380, 452], [481, 482], [585, 585]]</point>"}]}
{"id": 6822, "image": ["rtx_frames_success_53/62_robo_set#episode_9765/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9768/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle on the plate."}, {"from": "gpt", "value": "<point>[[310, 505], [419, 458], [478, 395], [530, 296]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the ketchup bottle."}, {"from": "gpt", "value": "<point>[[275, 462], [374, 458], [491, 485], [579, 538]]</point>"}]}
{"id": 6824, "image": ["rtx_frames_success_53/62_robo_set#episode_9774/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9808/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[260, 502], [323, 581], [380, 615], [431, 601]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[250, 462], [295, 561], [354, 628], [402, 654]]</point>"}]}
{"id": 6826, "image": ["rtx_frames_success_53/62_robo_set#episode_9820/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9823/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[252, 455], [369, 465], [470, 505], [577, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[282, 495], [414, 435], [538, 336], [621, 209]]</point>"}]}
{"id": 6828, "image": ["rtx_frames_success_53/62_robo_set#episode_9823/frame_15.png", "rtx_frames_success_53/62_robo_set#episode_9825/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to pull the mug backwards."}, {"from": "gpt", "value": "<point>[[609, 272], [600, 419], [613, 585], [637, 728]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[271, 488], [389, 571], [495, 615], [581, 618]]</point>"}]}
{"id": 6830, "image": ["rtx_frames_success_53/62_robo_set#episode_9852/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9856/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[301, 478], [414, 468], [519, 508], [602, 578]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to lift the lid."}, {"from": "gpt", "value": "<point>[[500, 30], [594, 50], [688, 120], [745, 252]]</point>"}]}
{"id": 6832, "image": ["rtx_frames_success_53/62_robo_set#episode_9860/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9865/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the oven."}, {"from": "gpt", "value": "<point>[[278, 492], [416, 452], [566, 458], [711, 515]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[278, 445], [387, 452], [472, 495], [560, 565]]</point>"}]}
{"id": 6834, "image": ["rtx_frames_success_53/62_robo_set#episode_9880/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9897/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[267, 478], [457, 478], [609, 498], [820, 588]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[263, 495], [361, 575], [489, 631], [590, 631]]</point>"}]}
{"id": 6836, "image": ["rtx_frames_success_53/62_robo_set#episode_9899/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9901/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the oven."}, {"from": "gpt", "value": "<point>[[233, 475], [391, 475], [570, 518], [690, 581]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[275, 498], [303, 472], [367, 468], [401, 478]]</point>"}]}
{"id": 6838, "image": ["rtx_frames_success_53/62_robo_set#episode_9901/frame_15.png", "rtx_frames_success_53/62_robo_set#episode_9936/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[382, 561], [378, 664], [382, 751], [393, 834]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[299, 455], [397, 455], [491, 488], [575, 551]]</point>"}]}
{"id": 6840, "image": ["rtx_frames_success_53/62_robo_set#episode_9938/frame_0.png", "rtx_frames_success_53/62_robo_set#episode_9938/frame_15.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the strainer."}, {"from": "gpt", "value": "<point>[[278, 498], [331, 518], [378, 555], [414, 625]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to drag the strainer across the table."}, {"from": "gpt", "value": "<point>[[457, 588], [502, 555], [572, 558], [628, 598]]</point>"}]}
{"id": 6842, "image": ["rtx_frames_success_54/17_dlr_sara_pour_converted_externally_to_rlds#episode_37/frame_0.png", "rtx_frames_success_54/17_dlr_sara_pour_converted_externally_to_rlds#episode_58/frame_0.png"], "height": [480, 480], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the mug."}, {"from": "gpt", "value": "<point>[[176, 465], [243, 451], [292, 460], [327, 493]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the mug."}, {"from": "gpt", "value": "<point>[[14, 516], [88, 587], [187, 610], [292, 582]]</point>"}]}
{"id": 6844, "image": ["rtx_frames_success_54/17_dlr_sara_pour_converted_externally_to_rlds#episode_91/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_0/frame_0.png"], "height": [480, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the mug."}, {"from": "gpt", "value": "<point>[[222, 563], [257, 521], [292, 484], [345, 498]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cube on the shelf."}, {"from": "gpt", "value": "<point>[[440, 338], [426, 495], [461, 613], [546, 714]]</point>"}]}
{"id": 6846, "image": ["rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_3/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_9/frame_0.png"], "height": [360, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the green apple on the plate."}, {"from": "gpt", "value": "<point>[[518, 232], [493, 313], [493, 407], [525, 501]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the glass."}, {"from": "gpt", "value": "<point>[[465, 194], [447, 282], [461, 369], [489, 401]]</point>"}]}
{"id": 6848, "image": ["rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_17/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_19/frame_0.png"], "height": [360, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach towards the green apple."}, {"from": "gpt", "value": "<point>[[426, 213], [408, 300], [412, 357], [440, 401]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the apple on the plate."}, {"from": "gpt", "value": "<point>[[394, 250], [401, 338], [405, 401], [444, 438]]</point>"}]}
{"id": 6850, "image": ["rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_25/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_34/frame_0.png"], "height": [360, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the orange."}, {"from": "gpt", "value": "<point>[[440, 213], [423, 275], [426, 344], [447, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the orange on the plate."}, {"from": "gpt", "value": "<point>[[475, 200], [528, 250], [560, 363], [567, 526]]</point>"}]}
{"id": 6852, "image": ["rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_38/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_40/frame_0.png"], "height": [360, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the shelf."}, {"from": "gpt", "value": "<point>[[447, 363], [454, 501], [518, 632], [599, 682]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to hover over the plate."}, {"from": "gpt", "value": "<point>[[461, 182], [500, 269], [489, 351], [465, 457]]</point>"}]}
{"id": 6854, "image": ["rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_50/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_51/frame_0.png"], "height": [360, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the yellow apple on the table."}, {"from": "gpt", "value": "<point>[[426, 188], [415, 275], [433, 363], [468, 432]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the cube."}, {"from": "gpt", "value": "<point>[[518, 388], [489, 588], [514, 764], [588, 864]]</point>"}]}
{"id": 6856, "image": ["rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_52/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_53/frame_0.png"], "height": [360, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the orange."}, {"from": "gpt", "value": "<point>[[511, 213], [535, 300], [539, 382], [511, 426]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to move towards the green apple."}, {"from": "gpt", "value": "<point>[[437, 188], [412, 257], [412, 319], [430, 401]]</point>"}]}
{"id": 6858, "image": ["rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_56/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_70/frame_0.png"], "height": [360, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the tennis ball."}, {"from": "gpt", "value": "<point>[[447, 250], [433, 326], [437, 426], [475, 482]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the green apple."}, {"from": "gpt", "value": "<point>[[581, 275], [532, 363], [539, 457], [546, 576]]</point>"}]}
{"id": 6860, "image": ["rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_85/frame_0.png", "rtx_frames_success_54/18_dlr_edan_shared_control_converted_externally_to_rlds#episode_86/frame_0.png"], "height": [360, 360], "width": [640, 640], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move towards the red apple."}, {"from": "gpt", "value": "<point>[[419, 225], [405, 294], [401, 357], [426, 407]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[475, 188], [447, 263], [454, 369], [482, 426]]</point>"}]}
{"id": 6862, "image": ["rtx_frames_success_54/62_robo_set#episode_999/frame_0.png", "rtx_frames_success_54/62_robo_set#episode_9982/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 460], [394, 572], [461, 610], [585, 579]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[342, 498], [349, 585], [377, 622], [447, 622]]</point>"}]}
{"id": 6864, "image": ["rtx_frames_success_54/62_robo_set#episode_9982/frame_15.png", "rtx_frames_success_54/62_robo_set#episode_9983/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to move the mug to the right."}, {"from": "gpt", "value": "<point>[[479, 585], [535, 541], [602, 547], [658, 597]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[363, 460], [433, 373], [482, 299], [623, 255]]</point>"}]}
{"id": 6866, "image": ["rtx_frames_success_54/62_robo_set#episode_9985/frame_0.png", "rtx_frames_success_54/62_robo_set#episode_9994/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the ketchup bottle."}, {"from": "gpt", "value": "<point>[[363, 473], [412, 560], [489, 597], [588, 585]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the mug."}, {"from": "gpt", "value": "<point>[[352, 473], [380, 579], [433, 622], [482, 628]]</point>"}]}
{"id": 6868, "image": ["rtx_frames_success_54/62_robo_set#episode_9995/frame_0.png", "rtx_frames_success_54/62_robo_set#episode_9998/frame_0.png"], "height": [240, 240], "width": [424, 424], "conversations": [{"from": "system", "value": "You are InternVL, a robotic entity agent.\n\nYou are given one observation of the scene and a task to perform. Based on the task, You should return a set of points as the trajectory to execute the task. The point trajectory format should be \"<point>[[x1, y1], [x2, y2], [x3, y3],...]</point>\". The point coordinates are normalized to integers between 0 and 1000. Please return the point trajectory directly."}, {"from": "human", "value": "Image-1:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[356, 473], [398, 579], [461, 610], [577, 572]]</point>"}, {"from": "human", "value": "Image-2:<image>\n You need to reach for the banana."}, {"from": "gpt", "value": "<point>[[352, 479], [387, 579], [461, 603], [602, 585]]</point>"}]}