Training in progress, epoch 4, checkpoint

checkpoint-1480/config.json

@@ -0,0 +1,283 @@
+{
+  "activation_dropout": 0.0,
+  "activation_function": "silu",
+  "anchor_image_size": null,
+  "architectures": [
+    "RTDetrV2ForObjectDetection"
+  ],
+  "attention_dropout": 0.0,
+  "auxiliary_loss": true,
+  "backbone": null,
+  "backbone_config": {
+    "depths": [
+      3,
+      4,
+      23,
+      3
+    ],
+    "downsample_in_bottleneck": false,
+    "downsample_in_first_stage": false,
+    "dtype": "float32",
+    "embedding_size": 64,
+    "hidden_act": "relu",
+    "hidden_sizes": [
+      256,
+      512,
+      1024,
+      2048
+    ],
+    "layer_type": "bottleneck",
+    "model_type": "rt_detr_resnet",
+    "num_channels": 3,
+    "out_features": [
+      "stage2",
+      "stage3",
+      "stage4"
+    ],
+    "out_indices": [
+      2,
+      3,
+      4
+    ],
+    "stage_names": [
+      "stem",
+      "stage1",
+      "stage2",
+      "stage3",
+      "stage4"
+    ]
+  },
+  "backbone_kwargs": null,
+  "batch_norm_eps": 1e-05,
+  "box_noise_scale": 1.0,
+  "d_model": 256,
+  "decoder_activation_function": "relu",
+  "decoder_attention_heads": 8,
+  "decoder_ffn_dim": 1024,
+  "decoder_in_channels": [
+    384,
+    384,
+    384
+  ],
+  "decoder_layers": 6,
+  "decoder_method": "default",
+  "decoder_n_levels": 3,
+  "decoder_n_points": 4,
+  "decoder_offset_scale": 0.5,
+  "disable_custom_kernels": true,
+  "dropout": 0.0,
+  "dtype": "float32",
+  "encode_proj_layers": [
+    2
+  ],
+  "encoder_activation_function": "gelu",
+  "encoder_attention_heads": 8,
+  "encoder_ffn_dim": 2048,
+  "encoder_hidden_dim": 384,
+  "encoder_in_channels": [
+    512,
+    1024,
+    2048
+  ],
+  "encoder_layers": 1,
+  "eos_coefficient": 0.0001,
+  "eval_size": null,
+  "feat_strides": [
+    8,
+    16,
+    32
+  ],
+  "focal_loss_alpha": 0.75,
+  "focal_loss_gamma": 2.0,
+  "freeze_backbone_batch_norms": true,
+  "hidden_expansion": 1.0,
+  "id2label": {
+    "0": "person",
+    "1": "bicycle",
+    "2": "car",
+    "3": "motorcycle",
+    "4": "airplane",
+    "5": "bus",
+    "6": "train",
+    "7": "truck",
+    "8": "boat",
+    "9": "traffic light",
+    "10": "fire hydrant",
+    "11": "stop sign",
+    "12": "parking meter",
+    "13": "bench",
+    "14": "bird",
+    "15": "cat",
+    "16": "dog",
+    "17": "horse",
+    "18": "sheep",
+    "19": "cow",
+    "20": "elephant",
+    "21": "bear",
+    "22": "zebra",
+    "23": "giraffe",
+    "24": "backpack",
+    "25": "umbrella",
+    "26": "handbag",
+    "27": "tie",
+    "28": "suitcase",
+    "29": "frisbee",
+    "30": "skis",
+    "31": "snowboard",
+    "32": "sports ball",
+    "33": "kite",
+    "34": "baseball bat",
+    "35": "baseball glove",
+    "36": "skateboard",
+    "37": "surfboard",
+    "38": "tennis racket",
+    "39": "bottle",
+    "40": "wine glass",
+    "41": "cup",
+    "42": "fork",
+    "43": "knife",
+    "44": "spoon",
+    "45": "bowl",
+    "46": "banana",
+    "47": "apple",
+    "48": "sandwich",
+    "49": "orange",
+    "50": "broccoli",
+    "51": "carrot",
+    "52": "hot dog",
+    "53": "pizza",
+    "54": "donut",
+    "55": "cake",
+    "56": "chair",
+    "57": "couch",
+    "58": "potted plant",
+    "59": "bed",
+    "60": "dining table",
+    "61": "toilet",
+    "62": "tv",
+    "63": "laptop",
+    "64": "mouse",
+    "65": "remote",
+    "66": "keyboard",
+    "67": "cell phone",
+    "68": "microwave",
+    "69": "oven",
+    "70": "toaster",
+    "71": "sink",
+    "72": "refrigerator",
+    "73": "book",
+    "74": "clock",
+    "75": "vase",
+    "76": "scissors",
+    "77": "teddy bear",
+    "78": "hair drier",
+    "79": "toothbrush"
+  },
+  "initializer_bias_prior_prob": null,
+  "initializer_range": 0.01,
+  "is_encoder_decoder": true,
+  "label2id": {
+    "airplane": 4,
+    "apple": 47,
+    "backpack": 24,
+    "banana": 46,
+    "baseball bat": 34,
+    "baseball glove": 35,
+    "bear": 21,
+    "bed": 59,
+    "bench": 13,
+    "bicycle": 1,
+    "bird": 14,
+    "boat": 8,
+    "book": 73,
+    "bottle": 39,
+    "bowl": 45,
+    "broccoli": 50,
+    "bus": 5,
+    "cake": 55,
+    "car": 2,
+    "carrot": 51,
+    "cat": 15,
+    "cell phone": 67,
+    "chair": 56,
+    "clock": 74,
+    "couch": 57,
+    "cow": 19,
+    "cup": 41,
+    "dining table": 60,
+    "dog": 16,
+    "donut": 54,
+    "elephant": 20,
+    "fire hydrant": 10,
+    "fork": 42,
+    "frisbee": 29,
+    "giraffe": 23,
+    "hair drier": 78,
+    "handbag": 26,
+    "horse": 17,
+    "hot dog": 52,
+    "keyboard": 66,
+    "kite": 33,
+    "knife": 43,
+    "laptop": 63,
+    "microwave": 68,
+    "motorcycle": 3,
+    "mouse": 64,
+    "orange": 49,
+    "oven": 69,
+    "parking meter": 12,
+    "person": 0,
+    "pizza": 53,
+    "potted plant": 58,
+    "refrigerator": 72,
+    "remote": 65,
+    "sandwich": 48,
+    "scissors": 76,
+    "sheep": 18,
+    "sink": 71,
+    "skateboard": 36,
+    "skis": 30,
+    "snowboard": 31,
+    "spoon": 44,
+    "sports ball": 32,
+    "stop sign": 11,
+    "suitcase": 28,
+    "surfboard": 37,
+    "teddy bear": 77,
+    "tennis racket": 38,
+    "tie": 27,
+    "toaster": 70,
+    "toilet": 61,
+    "toothbrush": 79,
+    "traffic light": 9,
+    "train": 6,
+    "truck": 7,
+    "tv": 62,
+    "umbrella": 25,
+    "vase": 75,
+    "wine glass": 40,
+    "zebra": 22
+  },
+  "label_noise_ratio": 0.5,
+  "layer_norm_eps": 1e-05,
+  "learn_initial_query": false,
+  "matcher_alpha": 0.25,
+  "matcher_bbox_cost": 5.0,
+  "matcher_class_cost": 2.0,
+  "matcher_gamma": 2.0,
+  "matcher_giou_cost": 2.0,
+  "model_type": "rt_detr_v2",
+  "normalize_before": false,
+  "num_denoising": 100,
+  "num_feature_levels": 3,
+  "num_queries": 300,
+  "positional_encoding_temperature": 10000,
+  "transformers_version": "4.57.3",
+  "use_focal_loss": true,
+  "use_pretrained_backbone": false,
+  "use_timm_backbone": false,
+  "weight_loss_bbox": 5.0,
+  "weight_loss_giou": 2.0,
+  "weight_loss_vfl": 1.0,
+  "with_box_refine": true
+}

checkpoint-1480/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d722d5cc23fa6643cccbaedb9eaf31c7ac5d477adb6fc42937614e3e5af4d79a
+size 307332000

checkpoint-1480/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4c94fb6bdbeafa813e693739a087ab4b0427d1b908197ef8d57deda94f80c1f
+size 612853474

checkpoint-1480/preprocessor_config.json

@@ -0,0 +1,36 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": false,
+  "device": null,
+  "disable_grouping": null,
+  "do_center_crop": null,
+  "do_convert_annotations": true,
+  "do_convert_rgb": null,
+  "do_normalize": false,
+  "do_pad": false,
+  "do_rescale": true,
+  "do_resize": true,
+  "format": "coco_detection",
+  "image_mean": [
+    0.485,
+    0.456,
+    0.406
+  ],
+  "image_processor_type": "RTDetrImageProcessorFast",
+  "image_std": [
+    0.229,
+    0.224,
+    0.225
+  ],
+  "input_data_format": null,
+  "pad_size": null,
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "return_segmentation_masks": null,
+  "return_tensors": null,
+  "size": {
+    "height": 640,
+    "width": 640
+  }
+}

checkpoint-1480/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:395cd5926cbfbfef51870019046198c5d7451264b54a0b671e6d4ab24905e736
+size 14645

checkpoint-1480/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd5b83c8cb76fb67e14f86551a3cddf78dcabed31f2b294c2e03362050550a93
+size 1383

checkpoint-1480/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bdb337fa4236cb7049a57519833627e759fba48dfa112766821d6f79c55753a3
+size 1465

checkpoint-1480/trainer_state.json

@@ -0,0 +1,768 @@
+{
+  "best_global_step": 1480,
+  "best_metric": 0.5394,
+  "best_model_checkpoint": "output/checkpoint-1480",
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1480,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "eval_loss": 9.46501350402832,
+      "eval_map": 0.5295,
+      "eval_map_50": 0.6826,
+      "eval_map_75": 0.5707,
+      "eval_map_airplane": 0.7824,
+      "eval_map_apple": 0.1443,
+      "eval_map_backpack": 0.2061,
+      "eval_map_banana": 0.2278,
+      "eval_map_baseball bat": 0.5873,
+      "eval_map_baseball glove": 0.3108,
+      "eval_map_bear": 0.9337,
+      "eval_map_bed": 0.6263,
+      "eval_map_bench": 0.4226,
+      "eval_map_bicycle": 0.1765,
+      "eval_map_bird": 1.0,
+      "eval_map_boat": 0.2403,
+      "eval_map_book": 0.1317,
+      "eval_map_bottle": 0.4378,
+      "eval_map_bowl": 0.5302,
+      "eval_map_broccoli": 0.487,
+      "eval_map_bus": 0.8264,
+      "eval_map_cake": 0.6871,
+      "eval_map_car": 0.5276,
+      "eval_map_carrot": 0.3177,
+      "eval_map_cat": 0.8147,
+      "eval_map_cell phone": 0.3335,
+      "eval_map_chair": 0.3875,
+      "eval_map_clock": 0.325,
+      "eval_map_couch": 0.5228,
+      "eval_map_cow": 0.7618,
+      "eval_map_cup": 0.5623,
+      "eval_map_dining table": 0.3977,
+      "eval_map_dog": 0.8948,
+      "eval_map_donut": -1.0,
+      "eval_map_elephant": 0.4286,
+      "eval_map_fire hydrant": -1.0,
+      "eval_map_fork": 0.7265,
+      "eval_map_frisbee": 0.9337,
+      "eval_map_giraffe": 0.8705,
+      "eval_map_hair drier": -1.0,
+      "eval_map_handbag": 0.2144,
+      "eval_map_horse": 0.9376,
+      "eval_map_hot dog": 0.305,
+      "eval_map_keyboard": 0.6782,
+      "eval_map_kite": 0.4356,
+      "eval_map_knife": 0.6093,
+      "eval_map_laptop": 0.7286,
+      "eval_map_large": 0.6695,
+      "eval_map_medium": 0.4823,
+      "eval_map_microwave": 0.2535,
+      "eval_map_motorcycle": 0.4147,
+      "eval_map_mouse": 0.6485,
+      "eval_map_orange": 0.2655,
+      "eval_map_oven": 0.5475,
+      "eval_map_parking meter": 0.9,
+      "eval_map_person": 0.5963,
+      "eval_map_pizza": 0.9282,
+      "eval_map_potted plant": 0.2062,
+      "eval_map_refrigerator": 0.61,
+      "eval_map_remote": 0.5045,
+      "eval_map_sandwich": 0.5005,
+      "eval_map_scissors": 0.6634,
+      "eval_map_sheep": 0.7738,
+      "eval_map_sink": 0.531,
+      "eval_map_skateboard": 0.8327,
+      "eval_map_skis": 0.3297,
+      "eval_map_small": 0.2751,
+      "eval_map_snowboard": 0.6829,
+      "eval_map_spoon": 0.0732,
+      "eval_map_sports ball": 0.2975,
+      "eval_map_stop sign": 0.8495,
+      "eval_map_suitcase": 0.5866,
+      "eval_map_surfboard": 0.3841,
+      "eval_map_teddy bear": 0.2667,
+      "eval_map_tennis racket": 0.6098,
+      "eval_map_tie": 0.285,
+      "eval_map_toaster": -1.0,
+      "eval_map_toilet": 0.4233,
+      "eval_map_toothbrush": 0.0027,
+      "eval_map_traffic light": 0.3928,
+      "eval_map_train": 0.6929,
+      "eval_map_truck": 0.49,
+      "eval_map_tv": 0.8589,
+      "eval_map_umbrella": 0.5607,
+      "eval_map_vase": 0.5855,
+      "eval_map_wine glass": 0.6441,
+      "eval_map_zebra": 0.5759,
+      "eval_mar_1": 0.4481,
+      "eval_mar_10": 0.6711,
+      "eval_mar_100": 0.7283,
+      "eval_mar_100_airplane": 0.8571,
+      "eval_mar_100_apple": 0.5571,
+      "eval_mar_100_backpack": 0.3667,
+      "eval_mar_100_banana": 0.7182,
+      "eval_mar_100_baseball bat": 0.8375,
+      "eval_mar_100_baseball glove": 0.4812,
+      "eval_mar_100_bear": 0.9333,
+      "eval_mar_100_bed": 0.95,
+      "eval_mar_100_bench": 0.5875,
+      "eval_mar_100_bicycle": 0.5,
+      "eval_mar_100_bird": 1.0,
+      "eval_mar_100_boat": 0.5327,
+      "eval_mar_100_book": 0.4441,
+      "eval_mar_100_bottle": 0.7074,
+      "eval_mar_100_bowl": 0.852,
+      "eval_mar_100_broccoli": 0.7286,
+      "eval_mar_100_bus": 0.9167,
+      "eval_mar_100_cake": 0.85,
+      "eval_mar_100_car": 0.6557,
+      "eval_mar_100_carrot": 0.8,
+      "eval_mar_100_cat": 0.9077,
+      "eval_mar_100_cell phone": 0.5211,
+      "eval_mar_100_chair": 0.6292,
+      "eval_mar_100_clock": 0.6412,
+      "eval_mar_100_couch": 0.8727,
+      "eval_mar_100_cow": 0.8417,
+      "eval_mar_100_cup": 0.8057,
+      "eval_mar_100_dining table": 0.6344,
+      "eval_mar_100_dog": 0.9273,
+      "eval_mar_100_donut": -1.0,
+      "eval_mar_100_elephant": 0.7778,
+      "eval_mar_100_fire hydrant": -1.0,
+      "eval_mar_100_fork": 0.8143,
+      "eval_mar_100_frisbee": 0.9333,
+      "eval_mar_100_giraffe": 0.9222,
+      "eval_mar_100_hair drier": -1.0,
+      "eval_mar_100_handbag": 0.548,
+      "eval_mar_100_horse": 0.94,
+      "eval_mar_100_hot dog": 0.4833,
+      "eval_mar_100_keyboard": 0.875,
+      "eval_mar_100_kite": 0.75,
+      "eval_mar_100_knife": 0.9,
+      "eval_mar_100_laptop": 0.8444,
+      "eval_mar_100_microwave": 0.7667,
+      "eval_mar_100_motorcycle": 0.6357,
+      "eval_mar_100_mouse": 0.7,
+      "eval_mar_100_orange": 0.6211,
+      "eval_mar_100_oven": 0.8667,
+      "eval_mar_100_parking meter": 0.9,
+      "eval_mar_100_person": 0.7068,
+      "eval_mar_100_pizza": 0.94,
+      "eval_mar_100_potted plant": 0.475,
+      "eval_mar_100_refrigerator": 0.8286,
+      "eval_mar_100_remote": 0.7143,
+      "eval_mar_100_sandwich": 0.7364,
+      "eval_mar_100_scissors": 0.6667,
+      "eval_mar_100_sheep": 0.7857,
+      "eval_mar_100_sink": 0.62,
+      "eval_mar_100_skateboard": 0.8333,
+      "eval_mar_100_skis": 0.5875,
+      "eval_mar_100_snowboard": 0.7857,
+      "eval_mar_100_spoon": 0.62,
+      "eval_mar_100_sports ball": 0.4136,
+      "eval_mar_100_stop sign": 0.8875,
+      "eval_mar_100_suitcase": 0.7375,
+      "eval_mar_100_surfboard": 0.6727,
+      "eval_mar_100_teddy bear": 0.8,
+      "eval_mar_100_tennis racket": 0.7667,
+      "eval_mar_100_tie": 0.42,
+      "eval_mar_100_toaster": -1.0,
+      "eval_mar_100_toilet": 0.6235,
+      "eval_mar_100_toothbrush": 0.3,
+      "eval_mar_100_traffic light": 0.5125,
+      "eval_mar_100_train": 0.8667,
+      "eval_mar_100_truck": 0.7357,
+      "eval_mar_100_tv": 0.9077,
+      "eval_mar_100_umbrella": 0.8059,
+      "eval_mar_100_vase": 0.7944,
+      "eval_mar_100_wine glass": 0.6875,
+      "eval_mar_100_zebra": 0.7818,
+      "eval_mar_large": 0.8283,
+      "eval_mar_medium": 0.6894,
+      "eval_mar_small": 0.4893,
+      "eval_runtime": 25.2611,
+      "eval_samples_per_second": 9.897,
+      "eval_steps_per_second": 1.267,
+      "step": 370
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 31.571361541748047,
+      "learning_rate": 4.3580645161290324e-05,
+      "loss": 15.7786,
+      "step": 500
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 9.811163902282715,
+      "eval_map": 0.5137,
+      "eval_map_50": 0.6598,
+      "eval_map_75": 0.5479,
+      "eval_map_airplane": 0.8008,
+      "eval_map_apple": 0.1239,
+      "eval_map_backpack": 0.1423,
+      "eval_map_banana": 0.2135,
+      "eval_map_baseball bat": 0.616,
+      "eval_map_baseball glove": 0.3174,
+      "eval_map_bear": 0.9554,
+      "eval_map_bed": 0.5406,
+      "eval_map_bench": 0.3767,
+      "eval_map_bicycle": 0.2086,
+      "eval_map_bird": 1.0,
+      "eval_map_boat": 0.3062,
+      "eval_map_book": 0.1202,
+      "eval_map_bottle": 0.4315,
+      "eval_map_bowl": 0.5122,
+      "eval_map_broccoli": 0.5114,
+      "eval_map_bus": 0.815,
+      "eval_map_cake": 0.7341,
+      "eval_map_car": 0.4924,
+      "eval_map_carrot": 0.5676,
+      "eval_map_cat": 0.7966,
+      "eval_map_cell phone": 0.2773,
+      "eval_map_chair": 0.3967,
+      "eval_map_clock": 0.4162,
+      "eval_map_couch": 0.4688,
+      "eval_map_cow": 0.7638,
+      "eval_map_cup": 0.5324,
+      "eval_map_dining table": 0.3746,
+      "eval_map_dog": 0.8275,
+      "eval_map_donut": -1.0,
+      "eval_map_elephant": 0.3519,
+      "eval_map_fire hydrant": -1.0,
+      "eval_map_fork": 0.6943,
+      "eval_map_frisbee": 0.9663,
+      "eval_map_giraffe": 0.8393,
+      "eval_map_hair drier": -1.0,
+      "eval_map_handbag": 0.1768,
+      "eval_map_horse": 0.9145,
+      "eval_map_hot dog": 0.376,
+      "eval_map_keyboard": 0.7277,
+      "eval_map_kite": 0.3979,
+      "eval_map_knife": 0.6698,
+      "eval_map_laptop": 0.6304,
+      "eval_map_large": 0.6598,
+      "eval_map_medium": 0.4607,
+      "eval_map_microwave": 0.6257,
+      "eval_map_motorcycle": 0.4431,
+      "eval_map_mouse": 0.6453,
+      "eval_map_orange": 0.2497,
+      "eval_map_oven": 0.3727,
+      "eval_map_parking meter": 0.45,
+      "eval_map_person": 0.5881,
+      "eval_map_pizza": 0.8485,
+      "eval_map_potted plant": 0.1926,
+      "eval_map_refrigerator": 0.7337,
+      "eval_map_remote": 0.4485,
+      "eval_map_sandwich": 0.516,
+      "eval_map_scissors": 0.5,
+      "eval_map_sheep": 0.7663,
+      "eval_map_sink": 0.5839,
+      "eval_map_skateboard": 0.8699,
+      "eval_map_skis": 0.3641,
+      "eval_map_small": 0.2697,
+      "eval_map_snowboard": 0.3515,
+      "eval_map_spoon": 0.1486,
+      "eval_map_sports ball": 0.2841,
+      "eval_map_stop sign": 0.8457,
+      "eval_map_suitcase": 0.5284,
+      "eval_map_surfboard": 0.5131,
+      "eval_map_teddy bear": 0.16,
+      "eval_map_tennis racket": 0.6334,
+      "eval_map_tie": 0.2763,
+      "eval_map_toaster": -1.0,
+      "eval_map_toilet": 0.3091,
+      "eval_map_toothbrush": 0.0,
+      "eval_map_traffic light": 0.3713,
+      "eval_map_train": 0.6606,
+      "eval_map_truck": 0.4001,
+      "eval_map_tv": 0.7874,
+      "eval_map_umbrella": 0.4815,
+      "eval_map_vase": 0.433,
+      "eval_map_wine glass": 0.6178,
+      "eval_map_zebra": 0.6551,
+      "eval_mar_1": 0.4396,
+      "eval_mar_10": 0.6637,
+      "eval_mar_100": 0.723,
+      "eval_mar_100_airplane": 0.9143,
+      "eval_mar_100_apple": 0.5643,
+      "eval_mar_100_backpack": 0.3833,
+      "eval_mar_100_banana": 0.7091,
+      "eval_mar_100_baseball bat": 0.825,
+      "eval_mar_100_baseball glove": 0.4812,
+      "eval_mar_100_bear": 0.9667,
+      "eval_mar_100_bed": 0.7833,
+      "eval_mar_100_bench": 0.775,
+      "eval_mar_100_bicycle": 0.51,
+      "eval_mar_100_bird": 1.0,
+      "eval_mar_100_boat": 0.5429,
+      "eval_mar_100_book": 0.4075,
+      "eval_mar_100_bottle": 0.6519,
+      "eval_mar_100_bowl": 0.84,
+      "eval_mar_100_broccoli": 0.7071,
+      "eval_mar_100_bus": 0.925,
+      "eval_mar_100_cake": 0.825,
+      "eval_mar_100_car": 0.6426,
+      "eval_mar_100_carrot": 0.85,
+      "eval_mar_100_cat": 0.9,
+      "eval_mar_100_cell phone": 0.4737,
+      "eval_mar_100_chair": 0.65,
+      "eval_mar_100_clock": 0.6059,
+      "eval_mar_100_couch": 0.8909,
+      "eval_mar_100_cow": 0.85,
+      "eval_mar_100_cup": 0.7743,
+      "eval_mar_100_dining table": 0.6375,
+      "eval_mar_100_dog": 0.8727,
+      "eval_mar_100_donut": -1.0,
+      "eval_mar_100_elephant": 0.7222,
+      "eval_mar_100_fire hydrant": -1.0,
+      "eval_mar_100_fork": 0.7071,
+      "eval_mar_100_frisbee": 0.9667,
+      "eval_mar_100_giraffe": 0.9333,
+      "eval_mar_100_hair drier": -1.0,
+      "eval_mar_100_handbag": 0.448,
+      "eval_mar_100_horse": 0.94,
+      "eval_mar_100_hot dog": 0.5778,
+      "eval_mar_100_keyboard": 0.875,
+      "eval_mar_100_kite": 0.75,
+      "eval_mar_100_knife": 0.8778,
+      "eval_mar_100_laptop": 0.8,
+      "eval_mar_100_microwave": 0.9,
+      "eval_mar_100_motorcycle": 0.6214,
+      "eval_mar_100_mouse": 0.6833,
+      "eval_mar_100_orange": 0.7053,
+      "eval_mar_100_oven": 0.7583,
+      "eval_mar_100_parking meter": 0.9,
+      "eval_mar_100_person": 0.7,
+      "eval_mar_100_pizza": 0.9,
+      "eval_mar_100_potted plant": 0.44,
+      "eval_mar_100_refrigerator": 0.8857,
+      "eval_mar_100_remote": 0.7048,
+      "eval_mar_100_sandwich": 0.7909,
+      "eval_mar_100_scissors": 0.6667,
+      "eval_mar_100_sheep": 0.7714,
+      "eval_mar_100_sink": 0.64,
+      "eval_mar_100_skateboard": 0.9,
+      "eval_mar_100_skis": 0.5562,
+      "eval_mar_100_snowboard": 0.6857,
+      "eval_mar_100_spoon": 0.64,
+      "eval_mar_100_sports ball": 0.4227,
+      "eval_mar_100_stop sign": 0.875,
+      "eval_mar_100_suitcase": 0.7563,
+      "eval_mar_100_surfboard": 0.7091,
+      "eval_mar_100_teddy bear": 0.8,
+      "eval_mar_100_tennis racket": 0.7444,
+      "eval_mar_100_tie": 0.46,
+      "eval_mar_100_toaster": -1.0,
+      "eval_mar_100_toilet": 0.7,
+      "eval_mar_100_toothbrush": 0.0,
+      "eval_mar_100_traffic light": 0.5042,
+      "eval_mar_100_train": 0.8333,
+      "eval_mar_100_truck": 0.7429,
+      "eval_mar_100_tv": 0.9231,
+      "eval_mar_100_umbrella": 0.7412,
+      "eval_mar_100_vase": 0.7278,
+      "eval_mar_100_wine glass": 0.8,
+      "eval_mar_100_zebra": 0.8045,
+      "eval_mar_large": 0.8132,
+      "eval_mar_medium": 0.7122,
+      "eval_mar_small": 0.4903,
+      "eval_runtime": 23.7192,
+      "eval_samples_per_second": 10.54,
+      "eval_steps_per_second": 1.349,
+      "step": 740
+    },
+    {
+      "epoch": 2.7027027027027026,
+      "grad_norm": 44.71751403808594,
+      "learning_rate": 2.745161290322581e-05,
+      "loss": 15.2267,
+      "step": 1000
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 9.518135070800781,
+      "eval_map": 0.5272,
+      "eval_map_50": 0.6782,
+      "eval_map_75": 0.5667,
+      "eval_map_airplane": 0.8202,
+      "eval_map_apple": 0.1313,
+      "eval_map_backpack": 0.1826,
+      "eval_map_banana": 0.199,
+      "eval_map_baseball bat": 0.6363,
+      "eval_map_baseball glove": 0.2784,
+      "eval_map_bear": 0.9346,
+      "eval_map_bed": 0.4713,
+      "eval_map_bench": 0.3367,
+      "eval_map_bicycle": 0.2675,
+      "eval_map_bird": 1.0,
+      "eval_map_boat": 0.2705,
+      "eval_map_book": 0.1111,
+      "eval_map_bottle": 0.4555,
+      "eval_map_bowl": 0.4361,
+      "eval_map_broccoli": 0.5568,
+      "eval_map_bus": 0.8327,
+      "eval_map_cake": 0.7817,
+      "eval_map_car": 0.5146,
+      "eval_map_carrot": 0.503,
+      "eval_map_cat": 0.8129,
+      "eval_map_cell phone": 0.3142,
+      "eval_map_chair": 0.4112,
+      "eval_map_clock": 0.4465,
+      "eval_map_couch": 0.5598,
+      "eval_map_cow": 0.7585,
+      "eval_map_cup": 0.5997,
+      "eval_map_dining table": 0.426,
+      "eval_map_dog": 0.8235,
+      "eval_map_donut": -1.0,
+      "eval_map_elephant": 0.5243,
+      "eval_map_fire hydrant": -1.0,
+      "eval_map_fork": 0.7199,
+      "eval_map_frisbee": 0.9554,
+      "eval_map_giraffe": 0.8826,
+      "eval_map_hair drier": -1.0,
+      "eval_map_handbag": 0.2344,
+      "eval_map_horse": 0.9148,
+      "eval_map_hot dog": 0.3306,
+      "eval_map_keyboard": 0.6499,
+      "eval_map_kite": 0.4659,
+      "eval_map_knife": 0.6277,
+      "eval_map_laptop": 0.7954,
+      "eval_map_large": 0.6746,
+      "eval_map_medium": 0.4743,
+      "eval_map_microwave": 0.699,
+      "eval_map_motorcycle": 0.403,
+      "eval_map_mouse": 0.7102,
+      "eval_map_orange": 0.3368,
+      "eval_map_oven": 0.5545,
+      "eval_map_parking meter": 0.1667,
+      "eval_map_person": 0.5821,
+      "eval_map_pizza": 0.9063,
+      "eval_map_potted plant": 0.176,
+      "eval_map_refrigerator": 0.6574,
+      "eval_map_remote": 0.5149,
+      "eval_map_sandwich": 0.4585,
+      "eval_map_scissors": 0.4168,
+      "eval_map_sheep": 0.7807,
+      "eval_map_sink": 0.5914,
+      "eval_map_skateboard": 0.7041,
+      "eval_map_skis": 0.3572,
+      "eval_map_small": 0.2848,
+      "eval_map_snowboard": 0.53,
+      "eval_map_spoon": 0.1853,
+      "eval_map_sports ball": 0.3592,
+      "eval_map_stop sign": 0.8428,
+      "eval_map_suitcase": 0.5124,
+      "eval_map_surfboard": 0.4682,
+      "eval_map_teddy bear": 0.1367,
+      "eval_map_tennis racket": 0.5722,
+      "eval_map_tie": 0.1997,
+      "eval_map_toaster": -1.0,
+      "eval_map_toilet": 0.4154,
+      "eval_map_toothbrush": 0.0,
+      "eval_map_traffic light": 0.3679,
+      "eval_map_train": 0.7368,
+      "eval_map_truck": 0.5293,
+      "eval_map_tv": 0.8263,
+      "eval_map_umbrella": 0.5598,
+      "eval_map_vase": 0.5349,
+      "eval_map_wine glass": 0.66,
+      "eval_map_zebra": 0.6425,
+      "eval_mar_1": 0.4334,
+      "eval_mar_10": 0.6705,
+      "eval_mar_100": 0.728,
+      "eval_mar_100_airplane": 0.9429,
+      "eval_mar_100_apple": 0.55,
+      "eval_mar_100_backpack": 0.4333,
+      "eval_mar_100_banana": 0.7227,
+      "eval_mar_100_baseball bat": 0.8625,
+      "eval_mar_100_baseball glove": 0.4313,
+      "eval_mar_100_bear": 0.9667,
+      "eval_mar_100_bed": 0.9,
+      "eval_mar_100_bench": 0.675,
+      "eval_mar_100_bicycle": 0.46,
+      "eval_mar_100_bird": 1.0,
+      "eval_mar_100_boat": 0.5184,
+      "eval_mar_100_book": 0.4269,
+      "eval_mar_100_bottle": 0.6444,
+      "eval_mar_100_bowl": 0.824,
+      "eval_mar_100_broccoli": 0.7286,
+      "eval_mar_100_bus": 0.9167,
+      "eval_mar_100_cake": 0.825,
+      "eval_mar_100_car": 0.6516,
+      "eval_mar_100_carrot": 0.8,
+      "eval_mar_100_cat": 0.8923,
+      "eval_mar_100_cell phone": 0.4895,
+      "eval_mar_100_chair": 0.6573,
+      "eval_mar_100_clock": 0.6,
+      "eval_mar_100_couch": 0.8818,
+      "eval_mar_100_cow": 0.8542,
+      "eval_mar_100_cup": 0.7829,
+      "eval_mar_100_dining table": 0.7125,
+      "eval_mar_100_dog": 0.9091,
+      "eval_mar_100_donut": -1.0,
+      "eval_mar_100_elephant": 0.8,
+      "eval_mar_100_fire hydrant": -1.0,
+      "eval_mar_100_fork": 0.7857,
+      "eval_mar_100_frisbee": 0.9667,
+      "eval_mar_100_giraffe": 0.9222,
+      "eval_mar_100_hair drier": -1.0,
+      "eval_mar_100_handbag": 0.512,
+      "eval_mar_100_horse": 0.96,
+      "eval_mar_100_hot dog": 0.5056,
+      "eval_mar_100_keyboard": 0.85,
+      "eval_mar_100_kite": 0.7444,
+      "eval_mar_100_knife": 0.8556,
+      "eval_mar_100_laptop": 0.8444,
+      "eval_mar_100_microwave": 0.8333,
+      "eval_mar_100_motorcycle": 0.5786,
+      "eval_mar_100_mouse": 0.75,
+      "eval_mar_100_orange": 0.6316,
+      "eval_mar_100_oven": 0.875,
+      "eval_mar_100_parking meter": 1.0,
+      "eval_mar_100_person": 0.7002,
+      "eval_mar_100_pizza": 0.93,
+      "eval_mar_100_potted plant": 0.52,
+      "eval_mar_100_refrigerator": 0.8429,
+      "eval_mar_100_remote": 0.7143,
+      "eval_mar_100_sandwich": 0.8182,
+      "eval_mar_100_scissors": 0.6667,
+      "eval_mar_100_sheep": 0.8,
+      "eval_mar_100_sink": 0.64,
+      "eval_mar_100_skateboard": 0.9,
+      "eval_mar_100_skis": 0.5813,
+      "eval_mar_100_snowboard": 0.7143,
+      "eval_mar_100_spoon": 0.58,
+      "eval_mar_100_sports ball": 0.4727,
+      "eval_mar_100_stop sign": 0.85,
+      "eval_mar_100_suitcase": 0.75,
+      "eval_mar_100_surfboard": 0.7273,
+      "eval_mar_100_teddy bear": 0.8,
+      "eval_mar_100_tennis racket": 0.6667,
+      "eval_mar_100_tie": 0.3,
+      "eval_mar_100_toaster": -1.0,
+      "eval_mar_100_toilet": 0.7,
+      "eval_mar_100_toothbrush": 0.0,
+      "eval_mar_100_traffic light": 0.5229,
+      "eval_mar_100_train": 0.8667,
+      "eval_mar_100_truck": 0.7643,
+      "eval_mar_100_tv": 0.9154,
+      "eval_mar_100_umbrella": 0.8059,
+      "eval_mar_100_vase": 0.7556,
+      "eval_mar_100_wine glass": 0.7375,
+      "eval_mar_100_zebra": 0.8091,
+      "eval_mar_large": 0.8302,
+      "eval_mar_medium": 0.6968,
+      "eval_mar_small": 0.4981,
+      "eval_runtime": 22.5643,
+      "eval_samples_per_second": 11.079,
+      "eval_steps_per_second": 1.418,
+      "step": 1110
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 9.485584259033203,
+      "eval_map": 0.5394,
+      "eval_map_50": 0.6966,
+      "eval_map_75": 0.5711,
+      "eval_map_airplane": 0.832,
+      "eval_map_apple": 0.1781,
+      "eval_map_backpack": 0.2231,
+      "eval_map_banana": 0.2332,
+      "eval_map_baseball bat": 0.5927,
+      "eval_map_baseball glove": 0.3282,
+      "eval_map_bear": 0.9345,
+      "eval_map_bed": 0.4515,
+      "eval_map_bench": 0.4156,
+      "eval_map_bicycle": 0.2869,
+      "eval_map_bird": 1.0,
+      "eval_map_boat": 0.3023,
+      "eval_map_book": 0.1171,
+      "eval_map_bottle": 0.4458,
+      "eval_map_bowl": 0.491,
+      "eval_map_broccoli": 0.5539,
+      "eval_map_bus": 0.8937,
+      "eval_map_cake": 0.7341,
+      "eval_map_car": 0.5224,
+      "eval_map_carrot": 0.6525,
+      "eval_map_cat": 0.8174,
+      "eval_map_cell phone": 0.2528,
+      "eval_map_chair": 0.4212,
+      "eval_map_clock": 0.4157,
+      "eval_map_couch": 0.488,
+      "eval_map_cow": 0.7192,
+      "eval_map_cup": 0.5973,
+      "eval_map_dining table": 0.3833,
+      "eval_map_dog": 0.9042,
+      "eval_map_donut": -1.0,
+      "eval_map_elephant": 0.4684,
+      "eval_map_fire hydrant": -1.0,
+      "eval_map_fork": 0.7317,
+      "eval_map_frisbee": 0.9554,
+      "eval_map_giraffe": 0.8757,
+      "eval_map_hair drier": -1.0,
+      "eval_map_handbag": 0.2454,
+      "eval_map_horse": 0.8901,
+      "eval_map_hot dog": 0.3568,
+      "eval_map_keyboard": 0.7607,
+      "eval_map_kite": 0.487,
+      "eval_map_knife": 0.6544,
+      "eval_map_laptop": 0.7827,
+      "eval_map_large": 0.6885,
+      "eval_map_medium": 0.4944,
+      "eval_map_microwave": 0.5245,
+      "eval_map_motorcycle": 0.435,
+      "eval_map_mouse": 0.7686,
+      "eval_map_orange": 0.3405,
+      "eval_map_oven": 0.5762,
+      "eval_map_parking meter": 0.3,
+      "eval_map_person": 0.5779,
+      "eval_map_pizza": 0.9147,
+      "eval_map_potted plant": 0.1858,
+      "eval_map_refrigerator": 0.6893,
+      "eval_map_remote": 0.5286,
+      "eval_map_sandwich": 0.4309,
+      "eval_map_scissors": 0.6634,
+      "eval_map_sheep": 0.7839,
+      "eval_map_sink": 0.5976,
+      "eval_map_skateboard": 0.8705,
+      "eval_map_skis": 0.3541,
+      "eval_map_small": 0.2738,
+      "eval_map_snowboard": 0.4877,
+      "eval_map_spoon": 0.0935,
+      "eval_map_sports ball": 0.3512,
+      "eval_map_stop sign": 0.8572,
+      "eval_map_suitcase": 0.5985,
+      "eval_map_surfboard": 0.3872,
+      "eval_map_teddy bear": 0.1333,
+      "eval_map_tennis racket": 0.6002,
+      "eval_map_tie": 0.3936,
+      "eval_map_toaster": -1.0,
+      "eval_map_toilet": 0.443,
+      "eval_map_toothbrush": 0.002,
+      "eval_map_traffic light": 0.3554,
+      "eval_map_train": 0.7535,
+      "eval_map_truck": 0.4417,
+      "eval_map_tv": 0.8287,
+      "eval_map_umbrella": 0.5582,
+      "eval_map_vase": 0.5086,
+      "eval_map_wine glass": 0.636,
+      "eval_map_zebra": 0.6268,
+      "eval_mar_1": 0.4664,
+      "eval_mar_10": 0.6739,
+      "eval_mar_100": 0.7308,
+      "eval_mar_100_airplane": 0.8857,
+      "eval_mar_100_apple": 0.5643,
+      "eval_mar_100_backpack": 0.4556,
+      "eval_mar_100_banana": 0.7636,
+      "eval_mar_100_baseball bat": 0.85,
+      "eval_mar_100_baseball glove": 0.4812,
+      "eval_mar_100_bear": 0.9667,
+      "eval_mar_100_bed": 0.8833,
+      "eval_mar_100_bench": 0.675,
+      "eval_mar_100_bicycle": 0.47,
+      "eval_mar_100_bird": 1.0,
+      "eval_mar_100_boat": 0.5755,
+      "eval_mar_100_book": 0.4398,
+      "eval_mar_100_bottle": 0.6444,
+      "eval_mar_100_bowl": 0.828,
+      "eval_mar_100_broccoli": 0.65,
+      "eval_mar_100_bus": 0.9333,
+      "eval_mar_100_cake": 0.825,
+      "eval_mar_100_car": 0.659,
+      "eval_mar_100_carrot": 0.85,
+      "eval_mar_100_cat": 0.8846,
+      "eval_mar_100_cell phone": 0.4737,
+      "eval_mar_100_chair": 0.624,
+      "eval_mar_100_clock": 0.5765,
+      "eval_mar_100_couch": 0.8909,
+      "eval_mar_100_cow": 0.85,
+      "eval_mar_100_cup": 0.8114,
+      "eval_mar_100_dining table": 0.6187,
+      "eval_mar_100_dog": 0.9182,
+      "eval_mar_100_donut": -1.0,
+      "eval_mar_100_elephant": 0.7667,
+      "eval_mar_100_fire hydrant": -1.0,
+      "eval_mar_100_fork": 0.7857,
+      "eval_mar_100_frisbee": 0.9667,
+      "eval_mar_100_giraffe": 0.9444,
+      "eval_mar_100_hair drier": -1.0,
+      "eval_mar_100_handbag": 0.524,
+      "eval_mar_100_horse": 0.94,
+      "eval_mar_100_hot dog": 0.5556,
+      "eval_mar_100_keyboard": 0.875,
+      "eval_mar_100_kite": 0.7167,
+      "eval_mar_100_knife": 0.8778,
+      "eval_mar_100_laptop": 0.8556,
+      "eval_mar_100_microwave": 0.7333,
+      "eval_mar_100_motorcycle": 0.6143,
+      "eval_mar_100_mouse": 0.8,
+      "eval_mar_100_orange": 0.6421,
+      "eval_mar_100_oven": 0.8667,
+      "eval_mar_100_parking meter": 0.9,
+      "eval_mar_100_person": 0.7096,
+      "eval_mar_100_pizza": 0.95,
+      "eval_mar_100_potted plant": 0.485,
+      "eval_mar_100_refrigerator": 0.8143,
+      "eval_mar_100_remote": 0.7429,
+      "eval_mar_100_sandwich": 0.7182,
+      "eval_mar_100_scissors": 0.6667,
+      "eval_mar_100_sheep": 0.8429,
+      "eval_mar_100_sink": 0.69,
+      "eval_mar_100_skateboard": 0.9333,
+      "eval_mar_100_skis": 0.6187,
+      "eval_mar_100_snowboard": 0.7143,
+      "eval_mar_100_spoon": 0.44,
+      "eval_mar_100_sports ball": 0.4909,
+      "eval_mar_100_stop sign": 0.8625,
+      "eval_mar_100_suitcase": 0.75,
+      "eval_mar_100_surfboard": 0.6545,
+      "eval_mar_100_teddy bear": 0.8,
+      "eval_mar_100_tennis racket": 0.7,
+      "eval_mar_100_tie": 0.44,
+      "eval_mar_100_toaster": -1.0,
+      "eval_mar_100_toilet": 0.7059,
+      "eval_mar_100_toothbrush": 0.3,
+      "eval_mar_100_traffic light": 0.5042,
+      "eval_mar_100_train": 0.8889,
+      "eval_mar_100_truck": 0.775,
+      "eval_mar_100_tv": 0.9077,
+      "eval_mar_100_umbrella": 0.7824,
+      "eval_mar_100_vase": 0.7333,
+      "eval_mar_100_wine glass": 0.75,
+      "eval_mar_100_zebra": 0.7591,
+      "eval_mar_large": 0.8268,
+      "eval_mar_medium": 0.6963,
+      "eval_mar_small": 0.504,
+      "eval_runtime": 21.9961,
+      "eval_samples_per_second": 11.366,
+      "eval_steps_per_second": 1.455,
+      "step": 1480
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 1850,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3348635365317018e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1480/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aad8ecf7619f7fa0357b193770195af05fc352390eedd48e2bf6b0db12bd25aa
+size 5841