diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,97630 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 9760, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "gumbel_tau": 1.0, + "image_masks_loss": 9.375, + "llm_learning_rate": 1.7064846416382254e-08, + "loss": 10.4927, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.7064846416382253e-06, + "step": 1 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9998, + "image_masks_loss": 9.1797, + "llm_learning_rate": 3.412969283276451e-08, + "loss": 10.4797, + "output_text_mask_loss": 0.0984, + "predictor_learning_rate": 3.4129692832764506e-06, + "step": 2 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9995, + "image_masks_loss": 8.3496, + "llm_learning_rate": 5.119453924914676e-08, + "loss": 9.6042, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 5.119453924914675e-06, + "step": 3 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9993, + "image_masks_loss": 8.6914, + "llm_learning_rate": 6.825938566552902e-08, + "loss": 10.0235, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 6.825938566552901e-06, + "step": 4 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9991, + "image_masks_loss": 8.4961, + "llm_learning_rate": 8.532423208191127e-08, + "loss": 9.7061, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 8.532423208191126e-06, + "step": 5 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9988, + "image_masks_loss": 8.9844, + "llm_learning_rate": 1.0238907849829352e-07, + "loss": 10.2919, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 1.023890784982935e-05, + "step": 6 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9986, + "image_masks_loss": 9.7656, + "llm_learning_rate": 1.1945392491467578e-07, + "loss": 10.4853, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.1945392491467578e-05, + "step": 7 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9983, + "image_masks_loss": 8.8867, + "llm_learning_rate": 1.3651877133105803e-07, + "loss": 10.1011, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 1.3651877133105803e-05, + "step": 8 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9981, + "image_masks_loss": 8.252, + "llm_learning_rate": 1.535836177474403e-07, + "loss": 9.3879, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.5358361774744027e-05, + "step": 9 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9979, + "image_masks_loss": 8.1543, + "llm_learning_rate": 1.7064846416382255e-07, + "loss": 9.652, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 1.7064846416382253e-05, + "step": 10 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9976, + "image_masks_loss": 8.3496, + "llm_learning_rate": 1.877133105802048e-07, + "loss": 9.2976, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 1.877133105802048e-05, + "step": 11 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9974, + "image_masks_loss": 7.959, + "llm_learning_rate": 2.0477815699658704e-07, + "loss": 9.0414, + "output_text_mask_loss": 0.0629, + "predictor_learning_rate": 2.04778156996587e-05, + "step": 12 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9972, + "image_masks_loss": 8.1055, + "llm_learning_rate": 2.218430034129693e-07, + "loss": 9.2871, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.218430034129693e-05, + "step": 13 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9969, + "image_masks_loss": 6.7383, + "llm_learning_rate": 2.3890784982935155e-07, + "loss": 8.0306, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 2.3890784982935157e-05, + "step": 14 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9967, + "image_masks_loss": 7.0312, + "llm_learning_rate": 2.559726962457338e-07, + "loss": 8.1173, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 2.559726962457338e-05, + "step": 15 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9965, + "image_masks_loss": 6.9824, + "llm_learning_rate": 2.7303754266211607e-07, + "loss": 8.2247, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 2.7303754266211605e-05, + "step": 16 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9962, + "image_masks_loss": 6.5918, + "llm_learning_rate": 2.901023890784983e-07, + "loss": 7.8378, + "output_text_mask_loss": 0.0614, + "predictor_learning_rate": 2.901023890784983e-05, + "step": 17 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.996, + "image_masks_loss": 5.9082, + "llm_learning_rate": 3.071672354948806e-07, + "loss": 7.2037, + "output_text_mask_loss": 0.0916, + "predictor_learning_rate": 3.0716723549488054e-05, + "step": 18 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9958, + "image_masks_loss": 5.542, + "llm_learning_rate": 3.242320819112628e-07, + "loss": 6.8942, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 3.2423208191126276e-05, + "step": 19 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9955, + "image_masks_loss": 5.2734, + "llm_learning_rate": 3.412969283276451e-07, + "loss": 6.5651, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 3.4129692832764505e-05, + "step": 20 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9953, + "image_masks_loss": 4.2725, + "llm_learning_rate": 3.583617747440273e-07, + "loss": 5.473, + "output_text_mask_loss": 0.1503, + "predictor_learning_rate": 3.583617747440273e-05, + "step": 21 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9951, + "image_masks_loss": 4.3945, + "llm_learning_rate": 3.754266211604096e-07, + "loss": 5.5969, + "output_text_mask_loss": 0.2563, + "predictor_learning_rate": 3.754266211604096e-05, + "step": 22 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9948, + "image_masks_loss": 3.8818, + "llm_learning_rate": 3.9249146757679187e-07, + "loss": 5.2452, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 3.924914675767918e-05, + "step": 23 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9946, + "image_masks_loss": 2.3315, + "llm_learning_rate": 4.0955631399317407e-07, + "loss": 3.371, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 4.09556313993174e-05, + "step": 24 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9944, + "image_masks_loss": 2.4658, + "llm_learning_rate": 4.266211604095564e-07, + "loss": 3.5179, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.266211604095564e-05, + "step": 25 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9941, + "image_masks_loss": 1.5991, + "llm_learning_rate": 4.436860068259386e-07, + "loss": 2.9296, + "output_text_mask_loss": 0.0452, + "predictor_learning_rate": 4.436860068259386e-05, + "step": 26 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9939, + "image_masks_loss": 0.9949, + "llm_learning_rate": 4.6075085324232084e-07, + "loss": 2.1732, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 4.6075085324232084e-05, + "step": 27 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9937, + "image_masks_loss": 0.5463, + "llm_learning_rate": 4.778156996587031e-07, + "loss": 1.8103, + "output_text_mask_loss": 0.1404, + "predictor_learning_rate": 4.778156996587031e-05, + "step": 28 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9934, + "image_masks_loss": 0.2258, + "llm_learning_rate": 4.948805460750854e-07, + "loss": 1.4283, + "output_text_mask_loss": 0.0832, + "predictor_learning_rate": 4.9488054607508536e-05, + "step": 29 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9932, + "image_masks_loss": 0.1038, + "llm_learning_rate": 5.119453924914676e-07, + "loss": 1.414, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.119453924914676e-05, + "step": 30 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9929, + "image_masks_loss": 0.0345, + "llm_learning_rate": 5.290102389078499e-07, + "loss": 1.109, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.290102389078499e-05, + "step": 31 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9927, + "image_masks_loss": 0.1999, + "llm_learning_rate": 5.460750853242321e-07, + "loss": 1.4674, + "output_text_mask_loss": 0.1068, + "predictor_learning_rate": 5.460750853242321e-05, + "step": 32 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9925, + "image_masks_loss": 0.415, + "llm_learning_rate": 5.631399317406143e-07, + "loss": 1.5806, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 5.631399317406143e-05, + "step": 33 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9922, + "image_masks_loss": 0.3235, + "llm_learning_rate": 5.802047781569966e-07, + "loss": 1.6224, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 5.802047781569966e-05, + "step": 34 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.992, + "image_masks_loss": 0.531, + "llm_learning_rate": 5.972696245733788e-07, + "loss": 1.8502, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 5.9726962457337885e-05, + "step": 35 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9918, + "image_masks_loss": 0.5188, + "llm_learning_rate": 6.143344709897612e-07, + "loss": 1.4884, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 6.143344709897611e-05, + "step": 36 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9915, + "image_masks_loss": 0.589, + "llm_learning_rate": 6.313993174061434e-07, + "loss": 1.914, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 6.313993174061433e-05, + "step": 37 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9913, + "image_masks_loss": 0.4242, + "llm_learning_rate": 6.484641638225256e-07, + "loss": 1.6582, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 6.484641638225255e-05, + "step": 38 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9911, + "image_masks_loss": 0.2838, + "llm_learning_rate": 6.65529010238908e-07, + "loss": 1.5518, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 6.655290102389079e-05, + "step": 39 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9908, + "image_masks_loss": 0.1503, + "llm_learning_rate": 6.825938566552902e-07, + "loss": 1.3173, + "output_text_mask_loss": 0.1717, + "predictor_learning_rate": 6.825938566552901e-05, + "step": 40 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9906, + "image_masks_loss": 0.0399, + "llm_learning_rate": 6.996587030716724e-07, + "loss": 1.3421, + "output_text_mask_loss": 0.1495, + "predictor_learning_rate": 6.996587030716723e-05, + "step": 41 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9904, + "image_masks_loss": 0.0244, + "llm_learning_rate": 7.167235494880546e-07, + "loss": 1.0825, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 7.167235494880546e-05, + "step": 42 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9901, + "image_masks_loss": 0.0313, + "llm_learning_rate": 7.337883959044369e-07, + "loss": 1.3682, + "output_text_mask_loss": 0.174, + "predictor_learning_rate": 7.337883959044368e-05, + "step": 43 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9899, + "image_masks_loss": 0.0835, + "llm_learning_rate": 7.508532423208192e-07, + "loss": 1.2334, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 7.508532423208191e-05, + "step": 44 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9897, + "image_masks_loss": 0.1244, + "llm_learning_rate": 7.679180887372014e-07, + "loss": 1.3364, + "output_text_mask_loss": 0.1213, + "predictor_learning_rate": 7.679180887372014e-05, + "step": 45 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9894, + "image_masks_loss": 0.2106, + "llm_learning_rate": 7.849829351535837e-07, + "loss": 1.4084, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 7.849829351535836e-05, + "step": 46 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.9892, + "image_masks_loss": 0.3799, + "llm_learning_rate": 8.020477815699659e-07, + "loss": 1.6601, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 8.020477815699658e-05, + "step": 47 + }, + { + "epoch": 0.0, + "gumbel_tau": 0.989, + "image_masks_loss": 0.3967, + "llm_learning_rate": 8.191126279863481e-07, + "loss": 1.6461, + "output_text_mask_loss": 0.1907, + "predictor_learning_rate": 8.19112627986348e-05, + "step": 48 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9887, + "image_masks_loss": 0.3403, + "llm_learning_rate": 8.361774744027303e-07, + "loss": 1.6222, + "output_text_mask_loss": 0.3571, + "predictor_learning_rate": 8.361774744027303e-05, + "step": 49 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9885, + "image_masks_loss": 0.293, + "llm_learning_rate": 8.532423208191128e-07, + "loss": 1.5358, + "output_text_mask_loss": 0.1602, + "predictor_learning_rate": 8.532423208191128e-05, + "step": 50 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9883, + "image_masks_loss": 0.2975, + "llm_learning_rate": 8.70307167235495e-07, + "loss": 1.4384, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 8.70307167235495e-05, + "step": 51 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.988, + "image_masks_loss": 0.1099, + "llm_learning_rate": 8.873720136518772e-07, + "loss": 1.4917, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 8.873720136518772e-05, + "step": 52 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9878, + "image_masks_loss": 0.1518, + "llm_learning_rate": 9.044368600682595e-07, + "loss": 1.2001, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 9.044368600682594e-05, + "step": 53 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9876, + "image_masks_loss": 0.0572, + "llm_learning_rate": 9.215017064846417e-07, + "loss": 1.435, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.215017064846417e-05, + "step": 54 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9873, + "image_masks_loss": 0.0448, + "llm_learning_rate": 9.385665529010239e-07, + "loss": 1.2901, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.385665529010239e-05, + "step": 55 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9871, + "image_masks_loss": 0.0607, + "llm_learning_rate": 9.556313993174062e-07, + "loss": 1.1596, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 9.556313993174063e-05, + "step": 56 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9869, + "image_masks_loss": 0.0326, + "llm_learning_rate": 9.726962457337885e-07, + "loss": 1.2636, + "output_text_mask_loss": 0.2472, + "predictor_learning_rate": 9.726962457337885e-05, + "step": 57 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9866, + "image_masks_loss": 0.0341, + "llm_learning_rate": 9.897610921501708e-07, + "loss": 1.3986, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 9.897610921501707e-05, + "step": 58 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9864, + "image_masks_loss": 0.0759, + "llm_learning_rate": 1.006825938566553e-06, + "loss": 1.3569, + "output_text_mask_loss": 0.4211, + "predictor_learning_rate": 0.0001006825938566553, + "step": 59 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9862, + "image_masks_loss": 0.177, + "llm_learning_rate": 1.0238907849829352e-06, + "loss": 1.4202, + "output_text_mask_loss": 0.4089, + "predictor_learning_rate": 0.00010238907849829352, + "step": 60 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9859, + "image_masks_loss": 0.0263, + "llm_learning_rate": 1.0409556313993173e-06, + "loss": 1.007, + "output_text_mask_loss": 0.1099, + "predictor_learning_rate": 0.00010409556313993174, + "step": 61 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9857, + "image_masks_loss": 0.0969, + "llm_learning_rate": 1.0580204778156999e-06, + "loss": 1.1982, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010580204778156998, + "step": 62 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9855, + "image_masks_loss": 0.074, + "llm_learning_rate": 1.075085324232082e-06, + "loss": 1.2425, + "output_text_mask_loss": 0.4578, + "predictor_learning_rate": 0.0001075085324232082, + "step": 63 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9852, + "image_masks_loss": 0.0858, + "llm_learning_rate": 1.0921501706484643e-06, + "loss": 1.3013, + "output_text_mask_loss": 1.0498, + "predictor_learning_rate": 0.00010921501706484642, + "step": 64 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.985, + "image_masks_loss": 0.0694, + "llm_learning_rate": 1.1092150170648466e-06, + "loss": 1.6468, + "output_text_mask_loss": 0.5066, + "predictor_learning_rate": 0.00011092150170648464, + "step": 65 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9848, + "image_masks_loss": 0.0313, + "llm_learning_rate": 1.1262798634812287e-06, + "loss": 1.1948, + "output_text_mask_loss": 0.1862, + "predictor_learning_rate": 0.00011262798634812287, + "step": 66 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9845, + "image_masks_loss": 0.0196, + "llm_learning_rate": 1.143344709897611e-06, + "loss": 1.2201, + "output_text_mask_loss": 0.3204, + "predictor_learning_rate": 0.00011433447098976109, + "step": 67 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9843, + "image_masks_loss": 0.1053, + "llm_learning_rate": 1.1604095563139933e-06, + "loss": 1.8991, + "output_text_mask_loss": 0.1076, + "predictor_learning_rate": 0.00011604095563139932, + "step": 68 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9841, + "image_masks_loss": 0.0984, + "llm_learning_rate": 1.1774744027303756e-06, + "loss": 1.6352, + "output_text_mask_loss": 0.3693, + "predictor_learning_rate": 0.00011774744027303755, + "step": 69 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9839, + "image_masks_loss": 0.1335, + "llm_learning_rate": 1.1945392491467577e-06, + "loss": 1.1763, + "output_text_mask_loss": 0.8484, + "predictor_learning_rate": 0.00011945392491467577, + "step": 70 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9836, + "image_masks_loss": 0.1915, + "llm_learning_rate": 1.21160409556314e-06, + "loss": 1.1943, + "output_text_mask_loss": 0.1305, + "predictor_learning_rate": 0.00012116040955631399, + "step": 71 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9834, + "image_masks_loss": 0.1564, + "llm_learning_rate": 1.2286689419795223e-06, + "loss": 1.228, + "output_text_mask_loss": 0.2579, + "predictor_learning_rate": 0.00012286689419795221, + "step": 72 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9832, + "image_masks_loss": 0.0511, + "llm_learning_rate": 1.2457337883959044e-06, + "loss": 1.3548, + "output_text_mask_loss": 0.7935, + "predictor_learning_rate": 0.00012457337883959045, + "step": 73 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9829, + "image_masks_loss": 0.053, + "llm_learning_rate": 1.2627986348122867e-06, + "loss": 1.4312, + "output_text_mask_loss": 0.8179, + "predictor_learning_rate": 0.00012627986348122866, + "step": 74 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9827, + "image_masks_loss": 0.0232, + "llm_learning_rate": 1.279863481228669e-06, + "loss": 1.6028, + "output_text_mask_loss": 0.2609, + "predictor_learning_rate": 0.0001279863481228669, + "step": 75 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9825, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.2969283276450511e-06, + "loss": 1.2457, + "output_text_mask_loss": 0.322, + "predictor_learning_rate": 0.0001296928327645051, + "step": 76 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9822, + "image_masks_loss": 0.0565, + "llm_learning_rate": 1.3139931740614337e-06, + "loss": 1.4503, + "output_text_mask_loss": 0.4822, + "predictor_learning_rate": 0.00013139931740614337, + "step": 77 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.982, + "image_masks_loss": 0.0332, + "llm_learning_rate": 1.331058020477816e-06, + "loss": 1.6198, + "output_text_mask_loss": 0.7111, + "predictor_learning_rate": 0.00013310580204778158, + "step": 78 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9818, + "image_masks_loss": 0.0504, + "llm_learning_rate": 1.348122866894198e-06, + "loss": 1.4199, + "output_text_mask_loss": 1.1108, + "predictor_learning_rate": 0.0001348122866894198, + "step": 79 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9815, + "image_masks_loss": 0.032, + "llm_learning_rate": 1.3651877133105804e-06, + "loss": 2.066, + "output_text_mask_loss": 1.2695, + "predictor_learning_rate": 0.00013651877133105802, + "step": 80 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9813, + "image_masks_loss": 0.0357, + "llm_learning_rate": 1.3822525597269627e-06, + "loss": 1.9931, + "output_text_mask_loss": 2.1606, + "predictor_learning_rate": 0.00013822525597269626, + "step": 81 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9811, + "image_masks_loss": 0.0406, + "llm_learning_rate": 1.3993174061433448e-06, + "loss": 1.614, + "output_text_mask_loss": 1.0254, + "predictor_learning_rate": 0.00013993174061433447, + "step": 82 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9808, + "image_masks_loss": 0.0828, + "llm_learning_rate": 1.416382252559727e-06, + "loss": 2.0747, + "output_text_mask_loss": 0.9033, + "predictor_learning_rate": 0.0001416382252559727, + "step": 83 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9806, + "image_masks_loss": 0.0145, + "llm_learning_rate": 1.4334470989761092e-06, + "loss": 1.9274, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001433447098976109, + "step": 84 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9804, + "image_masks_loss": 0.1251, + "llm_learning_rate": 1.4505119453924915e-06, + "loss": 2.3284, + "output_text_mask_loss": 1.1963, + "predictor_learning_rate": 0.00014505119453924915, + "step": 85 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9801, + "image_masks_loss": 0.0353, + "llm_learning_rate": 1.4675767918088738e-06, + "loss": 1.7805, + "output_text_mask_loss": 1.7334, + "predictor_learning_rate": 0.00014675767918088736, + "step": 86 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9799, + "image_masks_loss": 0.037, + "llm_learning_rate": 1.484641638225256e-06, + "loss": 2.01, + "output_text_mask_loss": 0.7202, + "predictor_learning_rate": 0.0001484641638225256, + "step": 87 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9797, + "image_masks_loss": 0.0584, + "llm_learning_rate": 1.5017064846416384e-06, + "loss": 1.4137, + "output_text_mask_loss": 0.5524, + "predictor_learning_rate": 0.00015017064846416383, + "step": 88 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9795, + "image_masks_loss": 0.0402, + "llm_learning_rate": 1.5187713310580207e-06, + "loss": 1.7328, + "output_text_mask_loss": 0.4395, + "predictor_learning_rate": 0.00015187713310580207, + "step": 89 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9792, + "image_masks_loss": 0.0263, + "llm_learning_rate": 1.5358361774744028e-06, + "loss": 1.7785, + "output_text_mask_loss": 1.77, + "predictor_learning_rate": 0.00015358361774744027, + "step": 90 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.979, + "image_masks_loss": 0.0439, + "llm_learning_rate": 1.5529010238907852e-06, + "loss": 1.559, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001552901023890785, + "step": 91 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9788, + "image_masks_loss": 0.0782, + "llm_learning_rate": 1.5699658703071675e-06, + "loss": 1.8771, + "output_text_mask_loss": 0.1602, + "predictor_learning_rate": 0.00015699658703071672, + "step": 92 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9785, + "image_masks_loss": 0.0462, + "llm_learning_rate": 1.5870307167235496e-06, + "loss": 1.7002, + "output_text_mask_loss": 1.178, + "predictor_learning_rate": 0.00015870307167235496, + "step": 93 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9783, + "image_masks_loss": 0.0584, + "llm_learning_rate": 1.6040955631399319e-06, + "loss": 1.8069, + "output_text_mask_loss": 0.5798, + "predictor_learning_rate": 0.00016040955631399316, + "step": 94 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9781, + "image_masks_loss": 0.0496, + "llm_learning_rate": 1.6211604095563142e-06, + "loss": 1.6974, + "output_text_mask_loss": 0.3357, + "predictor_learning_rate": 0.0001621160409556314, + "step": 95 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9778, + "image_masks_loss": 0.0211, + "llm_learning_rate": 1.6382252559726963e-06, + "loss": 1.9288, + "output_text_mask_loss": 0.9155, + "predictor_learning_rate": 0.0001638225255972696, + "step": 96 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9776, + "image_masks_loss": 0.0261, + "llm_learning_rate": 1.6552901023890786e-06, + "loss": 1.4767, + "output_text_mask_loss": 0.9888, + "predictor_learning_rate": 0.00016552901023890785, + "step": 97 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9774, + "image_masks_loss": 0.05, + "llm_learning_rate": 1.6723549488054607e-06, + "loss": 1.6855, + "output_text_mask_loss": 0.9216, + "predictor_learning_rate": 0.00016723549488054606, + "step": 98 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9771, + "image_masks_loss": 0.0523, + "llm_learning_rate": 1.689419795221843e-06, + "loss": 1.6608, + "output_text_mask_loss": 0.3052, + "predictor_learning_rate": 0.0001689419795221843, + "step": 99 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9769, + "image_masks_loss": 0.0584, + "llm_learning_rate": 1.7064846416382255e-06, + "loss": 1.5641, + "output_text_mask_loss": 0.264, + "predictor_learning_rate": 0.00017064846416382255, + "step": 100 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9767, + "image_masks_loss": 0.0338, + "llm_learning_rate": 1.7235494880546078e-06, + "loss": 1.3973, + "output_text_mask_loss": 0.4395, + "predictor_learning_rate": 0.00017235494880546076, + "step": 101 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9765, + "image_masks_loss": 0.1099, + "llm_learning_rate": 1.74061433447099e-06, + "loss": 1.5314, + "output_text_mask_loss": 1.3611, + "predictor_learning_rate": 0.000174061433447099, + "step": 102 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9762, + "image_masks_loss": 0.0418, + "llm_learning_rate": 1.7576791808873722e-06, + "loss": 1.359, + "output_text_mask_loss": 0.7019, + "predictor_learning_rate": 0.0001757679180887372, + "step": 103 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.976, + "image_masks_loss": 0.106, + "llm_learning_rate": 1.7747440273037543e-06, + "loss": 1.4042, + "output_text_mask_loss": 0.6287, + "predictor_learning_rate": 0.00017747440273037544, + "step": 104 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9758, + "image_masks_loss": 0.0254, + "llm_learning_rate": 1.7918088737201367e-06, + "loss": 1.3158, + "output_text_mask_loss": 0.4028, + "predictor_learning_rate": 0.00017918088737201365, + "step": 105 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9755, + "image_masks_loss": 0.0353, + "llm_learning_rate": 1.808873720136519e-06, + "loss": 1.0601, + "output_text_mask_loss": 0.0839, + "predictor_learning_rate": 0.0001808873720136519, + "step": 106 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9753, + "image_masks_loss": 0.0187, + "llm_learning_rate": 1.825938566552901e-06, + "loss": 1.1416, + "output_text_mask_loss": 0.5981, + "predictor_learning_rate": 0.0001825938566552901, + "step": 107 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9751, + "image_masks_loss": 0.0523, + "llm_learning_rate": 1.8430034129692834e-06, + "loss": 1.1433, + "output_text_mask_loss": 0.1389, + "predictor_learning_rate": 0.00018430034129692833, + "step": 108 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9748, + "image_masks_loss": 0.1343, + "llm_learning_rate": 1.8600682593856657e-06, + "loss": 1.452, + "output_text_mask_loss": 0.3281, + "predictor_learning_rate": 0.00018600682593856654, + "step": 109 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9746, + "image_masks_loss": 0.0326, + "llm_learning_rate": 1.8771331058020478e-06, + "loss": 1.2239, + "output_text_mask_loss": 0.3204, + "predictor_learning_rate": 0.00018771331058020478, + "step": 110 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9744, + "image_masks_loss": 0.0237, + "llm_learning_rate": 1.89419795221843e-06, + "loss": 1.2525, + "output_text_mask_loss": 0.1953, + "predictor_learning_rate": 0.000189419795221843, + "step": 111 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9742, + "image_masks_loss": 0.0465, + "llm_learning_rate": 1.9112627986348124e-06, + "loss": 1.6726, + "output_text_mask_loss": 1.0498, + "predictor_learning_rate": 0.00019112627986348125, + "step": 112 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9739, + "image_masks_loss": 0.0483, + "llm_learning_rate": 1.9283276450511947e-06, + "loss": 1.3543, + "output_text_mask_loss": 0.2899, + "predictor_learning_rate": 0.00019283276450511946, + "step": 113 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9737, + "image_masks_loss": 0.042, + "llm_learning_rate": 1.945392491467577e-06, + "loss": 1.6207, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001945392491467577, + "step": 114 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9735, + "image_masks_loss": 0.0938, + "llm_learning_rate": 1.9624573378839593e-06, + "loss": 2.0347, + "output_text_mask_loss": 0.2472, + "predictor_learning_rate": 0.0001962457337883959, + "step": 115 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9732, + "image_masks_loss": 0.0467, + "llm_learning_rate": 1.9795221843003416e-06, + "loss": 1.7988, + "output_text_mask_loss": 0.8789, + "predictor_learning_rate": 0.00019795221843003414, + "step": 116 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.973, + "image_masks_loss": 0.0355, + "llm_learning_rate": 1.9965870307167235e-06, + "loss": 1.8511, + "output_text_mask_loss": 1.0132, + "predictor_learning_rate": 0.00019965870307167235, + "step": 117 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9728, + "image_masks_loss": 0.0334, + "llm_learning_rate": 2.013651877133106e-06, + "loss": 1.6861, + "output_text_mask_loss": 0.8057, + "predictor_learning_rate": 0.0002013651877133106, + "step": 118 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9725, + "image_masks_loss": 0.0843, + "llm_learning_rate": 2.030716723549488e-06, + "loss": 2.4694, + "output_text_mask_loss": 1.2451, + "predictor_learning_rate": 0.0002030716723549488, + "step": 119 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9723, + "image_masks_loss": 0.0755, + "llm_learning_rate": 2.0477815699658705e-06, + "loss": 2.9211, + "output_text_mask_loss": 1.6357, + "predictor_learning_rate": 0.00020477815699658703, + "step": 120 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9721, + "image_masks_loss": 0.05, + "llm_learning_rate": 2.0648464163822528e-06, + "loss": 2.1128, + "output_text_mask_loss": 1.5442, + "predictor_learning_rate": 0.00020648464163822524, + "step": 121 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9719, + "image_masks_loss": 0.0629, + "llm_learning_rate": 2.0819112627986347e-06, + "loss": 1.9704, + "output_text_mask_loss": 0.8789, + "predictor_learning_rate": 0.00020819112627986348, + "step": 122 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9716, + "image_masks_loss": 0.0782, + "llm_learning_rate": 2.098976109215017e-06, + "loss": 2.2882, + "output_text_mask_loss": 0.4761, + "predictor_learning_rate": 0.0002098976109215017, + "step": 123 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9714, + "image_masks_loss": 0.0271, + "llm_learning_rate": 2.1160409556313997e-06, + "loss": 2.6233, + "output_text_mask_loss": 2.4536, + "predictor_learning_rate": 0.00021160409556313995, + "step": 124 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9712, + "image_masks_loss": 0.0675, + "llm_learning_rate": 2.133105802047782e-06, + "loss": 1.0958, + "output_text_mask_loss": 0.4761, + "predictor_learning_rate": 0.00021331058020477816, + "step": 125 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9709, + "image_masks_loss": 0.0429, + "llm_learning_rate": 2.150170648464164e-06, + "loss": 2.0029, + "output_text_mask_loss": 0.5371, + "predictor_learning_rate": 0.0002150170648464164, + "step": 126 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9707, + "image_masks_loss": 0.0477, + "llm_learning_rate": 2.1672354948805462e-06, + "loss": 1.3788, + "output_text_mask_loss": 1.2024, + "predictor_learning_rate": 0.0002167235494880546, + "step": 127 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9705, + "image_masks_loss": 0.0427, + "llm_learning_rate": 2.1843003412969285e-06, + "loss": 1.4163, + "output_text_mask_loss": 0.2747, + "predictor_learning_rate": 0.00021843003412969284, + "step": 128 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9703, + "image_masks_loss": 0.0212, + "llm_learning_rate": 2.201365187713311e-06, + "loss": 1.5372, + "output_text_mask_loss": 1.0193, + "predictor_learning_rate": 0.00022013651877133105, + "step": 129 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.97, + "image_masks_loss": 0.0343, + "llm_learning_rate": 2.218430034129693e-06, + "loss": 1.9073, + "output_text_mask_loss": 1.1597, + "predictor_learning_rate": 0.00022184300341296929, + "step": 130 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9698, + "image_masks_loss": 0.0473, + "llm_learning_rate": 2.235494880546075e-06, + "loss": 1.1556, + "output_text_mask_loss": 0.2472, + "predictor_learning_rate": 0.00022354948805460752, + "step": 131 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9696, + "image_masks_loss": 0.0437, + "llm_learning_rate": 2.2525597269624573e-06, + "loss": 1.3231, + "output_text_mask_loss": 0.6805, + "predictor_learning_rate": 0.00022525597269624573, + "step": 132 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9693, + "image_masks_loss": 0.045, + "llm_learning_rate": 2.2696245733788397e-06, + "loss": 1.2525, + "output_text_mask_loss": 0.1648, + "predictor_learning_rate": 0.00022696245733788397, + "step": 133 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9691, + "image_masks_loss": 0.0675, + "llm_learning_rate": 2.286689419795222e-06, + "loss": 0.9701, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022866894197952218, + "step": 134 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9689, + "image_masks_loss": 0.0843, + "llm_learning_rate": 2.3037542662116043e-06, + "loss": 1.0154, + "output_text_mask_loss": 0.106, + "predictor_learning_rate": 0.00023037542662116044, + "step": 135 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9687, + "image_masks_loss": 0.0763, + "llm_learning_rate": 2.3208191126279866e-06, + "loss": 1.0776, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00023208191126279865, + "step": 136 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9684, + "image_masks_loss": 0.0675, + "llm_learning_rate": 2.337883959044369e-06, + "loss": 1.1668, + "output_text_mask_loss": 0.0084, + "predictor_learning_rate": 0.00023378839590443688, + "step": 137 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9682, + "image_masks_loss": 0.018, + "llm_learning_rate": 2.354948805460751e-06, + "loss": 1.2392, + "output_text_mask_loss": 0.5463, + "predictor_learning_rate": 0.0002354948805460751, + "step": 138 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.968, + "image_masks_loss": 0.0303, + "llm_learning_rate": 2.3720136518771335e-06, + "loss": 1.1046, + "output_text_mask_loss": 0.0893, + "predictor_learning_rate": 0.00023720136518771333, + "step": 139 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9677, + "image_masks_loss": 0.0452, + "llm_learning_rate": 2.3890784982935154e-06, + "loss": 1.1573, + "output_text_mask_loss": 0.0858, + "predictor_learning_rate": 0.00023890784982935154, + "step": 140 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9675, + "image_masks_loss": 0.0648, + "llm_learning_rate": 2.4061433447098977e-06, + "loss": 1.0259, + "output_text_mask_loss": 0.1602, + "predictor_learning_rate": 0.00024061433447098977, + "step": 141 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9673, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.42320819112628e-06, + "loss": 1.0321, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00024232081911262798, + "step": 142 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9671, + "image_masks_loss": 0.066, + "llm_learning_rate": 2.4402730375426623e-06, + "loss": 1.2075, + "output_text_mask_loss": 0.2289, + "predictor_learning_rate": 0.00024402730375426622, + "step": 143 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9668, + "image_masks_loss": 0.0702, + "llm_learning_rate": 2.4573378839590446e-06, + "loss": 1.0752, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.00024573378839590443, + "step": 144 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9666, + "image_masks_loss": 0.0414, + "llm_learning_rate": 2.4744027303754265e-06, + "loss": 1.6189, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00024744027303754264, + "step": 145 + }, + { + "epoch": 0.01, + "gumbel_tau": 0.9664, + "image_masks_loss": 0.0222, + "llm_learning_rate": 2.491467576791809e-06, + "loss": 1.2671, + "output_text_mask_loss": 0.3571, + "predictor_learning_rate": 0.0002491467576791809, + "step": 146 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9661, + "image_masks_loss": 0.053, + "llm_learning_rate": 2.5085324232081916e-06, + "loss": 1.3658, + "output_text_mask_loss": 0.5005, + "predictor_learning_rate": 0.0002508532423208191, + "step": 147 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9659, + "image_masks_loss": 0.0307, + "llm_learning_rate": 2.5255972696245735e-06, + "loss": 1.364, + "output_text_mask_loss": 1.0071, + "predictor_learning_rate": 0.0002525597269624573, + "step": 148 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9657, + "image_masks_loss": 0.0203, + "llm_learning_rate": 2.5426621160409558e-06, + "loss": 1.2633, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.0002542662116040956, + "step": 149 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9655, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.559726962457338e-06, + "loss": 1.3092, + "output_text_mask_loss": 0.5981, + "predictor_learning_rate": 0.0002559726962457338, + "step": 150 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9652, + "image_masks_loss": 0.0387, + "llm_learning_rate": 2.5767918088737204e-06, + "loss": 1.0592, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.000257679180887372, + "step": 151 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.965, + "image_masks_loss": 0.0357, + "llm_learning_rate": 2.5938566552901023e-06, + "loss": 1.0869, + "output_text_mask_loss": 0.3098, + "predictor_learning_rate": 0.0002593856655290102, + "step": 152 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9648, + "image_masks_loss": 0.0372, + "llm_learning_rate": 2.610921501706485e-06, + "loss": 1.1765, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00026109215017064847, + "step": 153 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9645, + "image_masks_loss": 0.0732, + "llm_learning_rate": 2.6279863481228673e-06, + "loss": 1.1275, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.00026279863481228674, + "step": 154 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9643, + "image_masks_loss": 0.0572, + "llm_learning_rate": 2.645051194539249e-06, + "loss": 0.8769, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0002645051194539249, + "step": 155 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9641, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.662116040955632e-06, + "loss": 0.7999, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00026621160409556315, + "step": 156 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9639, + "image_masks_loss": 0.0813, + "llm_learning_rate": 2.679180887372014e-06, + "loss": 1.2546, + "output_text_mask_loss": 0.0973, + "predictor_learning_rate": 0.00026791808873720136, + "step": 157 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9636, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.696245733788396e-06, + "loss": 1.0249, + "output_text_mask_loss": 0.1541, + "predictor_learning_rate": 0.0002696245733788396, + "step": 158 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9634, + "image_masks_loss": 0.0206, + "llm_learning_rate": 2.713310580204778e-06, + "loss": 1.0851, + "output_text_mask_loss": 0.1244, + "predictor_learning_rate": 0.0002713310580204778, + "step": 159 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9632, + "image_masks_loss": 0.028, + "llm_learning_rate": 2.7303754266211608e-06, + "loss": 1.1269, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.00027303754266211604, + "step": 160 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.963, + "image_masks_loss": 0.0256, + "llm_learning_rate": 2.7474402730375426e-06, + "loss": 0.9415, + "output_text_mask_loss": 0.2472, + "predictor_learning_rate": 0.00027474402730375425, + "step": 161 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9627, + "image_masks_loss": 0.0626, + "llm_learning_rate": 2.7645051194539254e-06, + "loss": 1.0479, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0002764505119453925, + "step": 162 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9625, + "image_masks_loss": 0.0454, + "llm_learning_rate": 2.7815699658703073e-06, + "loss": 1.2268, + "output_text_mask_loss": 0.1762, + "predictor_learning_rate": 0.00027815699658703067, + "step": 163 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9623, + "image_masks_loss": 0.0404, + "llm_learning_rate": 2.7986348122866896e-06, + "loss": 1.1014, + "output_text_mask_loss": 0.6958, + "predictor_learning_rate": 0.00027986348122866893, + "step": 164 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.962, + "image_masks_loss": 0.036, + "llm_learning_rate": 2.8156996587030723e-06, + "loss": 1.2246, + "output_text_mask_loss": 0.4517, + "predictor_learning_rate": 0.0002815699658703072, + "step": 165 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9618, + "image_masks_loss": 0.1007, + "llm_learning_rate": 2.832764505119454e-06, + "loss": 0.9837, + "output_text_mask_loss": 0.209, + "predictor_learning_rate": 0.0002832764505119454, + "step": 166 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9616, + "image_masks_loss": 0.0116, + "llm_learning_rate": 2.8498293515358365e-06, + "loss": 0.9578, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00028498293515358367, + "step": 167 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9614, + "image_masks_loss": 0.0343, + "llm_learning_rate": 2.8668941979522184e-06, + "loss": 1.051, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0002866894197952218, + "step": 168 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9611, + "image_masks_loss": 0.0828, + "llm_learning_rate": 2.883959044368601e-06, + "loss": 1.0464, + "output_text_mask_loss": 0.1099, + "predictor_learning_rate": 0.0002883959044368601, + "step": 169 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9609, + "image_masks_loss": 0.1648, + "llm_learning_rate": 2.901023890784983e-06, + "loss": 1.1888, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002901023890784983, + "step": 170 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9607, + "image_masks_loss": 0.0345, + "llm_learning_rate": 2.9180887372013657e-06, + "loss": 0.9621, + "output_text_mask_loss": 0.1099, + "predictor_learning_rate": 0.00029180887372013656, + "step": 171 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9605, + "image_masks_loss": 0.0471, + "llm_learning_rate": 2.9351535836177476e-06, + "loss": 1.0936, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.0002935153583617747, + "step": 172 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9602, + "image_masks_loss": 0.0771, + "llm_learning_rate": 2.95221843003413e-06, + "loss": 1.0369, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.000295221843003413, + "step": 173 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.96, + "image_masks_loss": 0.1022, + "llm_learning_rate": 2.969283276450512e-06, + "loss": 0.9353, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.0002969283276450512, + "step": 174 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9598, + "image_masks_loss": 0.0607, + "llm_learning_rate": 2.9863481228668946e-06, + "loss": 0.8633, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00029863481228668945, + "step": 175 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9596, + "image_masks_loss": 0.0324, + "llm_learning_rate": 3.003412969283277e-06, + "loss": 0.944, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00030034129692832766, + "step": 176 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9593, + "image_masks_loss": 0.0683, + "llm_learning_rate": 3.0204778156996588e-06, + "loss": 1.0827, + "output_text_mask_loss": 0.1968, + "predictor_learning_rate": 0.00030204778156996587, + "step": 177 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9591, + "image_masks_loss": 0.1175, + "llm_learning_rate": 3.0375426621160415e-06, + "loss": 0.9507, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.00030375426621160413, + "step": 178 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9589, + "image_masks_loss": 0.0324, + "llm_learning_rate": 3.0546075085324234e-06, + "loss": 1.1019, + "output_text_mask_loss": 0.1038, + "predictor_learning_rate": 0.00030546075085324234, + "step": 179 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9586, + "image_masks_loss": 0.0507, + "llm_learning_rate": 3.0716723549488057e-06, + "loss": 0.8756, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030716723549488055, + "step": 180 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9584, + "image_masks_loss": 0.0275, + "llm_learning_rate": 3.088737201365188e-06, + "loss": 0.8336, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00030887372013651876, + "step": 181 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9582, + "image_masks_loss": 0.0576, + "llm_learning_rate": 3.1058020477815703e-06, + "loss": 1.1216, + "output_text_mask_loss": 0.119, + "predictor_learning_rate": 0.000310580204778157, + "step": 182 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.958, + "image_masks_loss": 0.087, + "llm_learning_rate": 3.122866894197952e-06, + "loss": 0.8907, + "output_text_mask_loss": 0.164, + "predictor_learning_rate": 0.00031228668941979523, + "step": 183 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9577, + "image_masks_loss": 0.0328, + "llm_learning_rate": 3.139931740614335e-06, + "loss": 0.9657, + "output_text_mask_loss": 0.1495, + "predictor_learning_rate": 0.00031399317406143344, + "step": 184 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9575, + "image_masks_loss": 0.0319, + "llm_learning_rate": 3.156996587030717e-06, + "loss": 1.1309, + "output_text_mask_loss": 0.0633, + "predictor_learning_rate": 0.00031569965870307165, + "step": 185 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9573, + "image_masks_loss": 0.0383, + "llm_learning_rate": 3.174061433447099e-06, + "loss": 0.9624, + "output_text_mask_loss": 0.145, + "predictor_learning_rate": 0.0003174061433447099, + "step": 186 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9571, + "image_masks_loss": 0.05, + "llm_learning_rate": 3.191126279863481e-06, + "loss": 0.9577, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003191126279863481, + "step": 187 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9568, + "image_masks_loss": 0.071, + "llm_learning_rate": 3.2081911262798638e-06, + "loss": 1.0396, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 0.00032081911262798633, + "step": 188 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9566, + "image_masks_loss": 0.0507, + "llm_learning_rate": 3.225255972696246e-06, + "loss": 0.9046, + "output_text_mask_loss": 0.1984, + "predictor_learning_rate": 0.0003225255972696246, + "step": 189 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9564, + "image_masks_loss": 0.023, + "llm_learning_rate": 3.2423208191126284e-06, + "loss": 0.9534, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0003242320819112628, + "step": 190 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9562, + "image_masks_loss": 0.0374, + "llm_learning_rate": 3.2593856655290107e-06, + "loss": 1.0294, + "output_text_mask_loss": 0.0828, + "predictor_learning_rate": 0.00032593856655290107, + "step": 191 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9559, + "image_masks_loss": 0.0313, + "llm_learning_rate": 3.2764505119453926e-06, + "loss": 0.9395, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003276450511945392, + "step": 192 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9557, + "image_masks_loss": 0.0637, + "llm_learning_rate": 3.2935153583617753e-06, + "loss": 1.0068, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.0003293515358361775, + "step": 193 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9555, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.310580204778157e-06, + "loss": 0.9319, + "output_text_mask_loss": 0.1747, + "predictor_learning_rate": 0.0003310580204778157, + "step": 194 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9553, + "image_masks_loss": 0.0721, + "llm_learning_rate": 3.3276450511945395e-06, + "loss": 0.9497, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00033276450511945396, + "step": 195 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.955, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.3447098976109214e-06, + "loss": 1.0612, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0003344709897610921, + "step": 196 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9548, + "image_masks_loss": 0.1007, + "llm_learning_rate": 3.361774744027304e-06, + "loss": 1.0582, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003361774744027304, + "step": 197 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9546, + "image_masks_loss": 0.0299, + "llm_learning_rate": 3.378839590443686e-06, + "loss": 0.8801, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003378839590443686, + "step": 198 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9544, + "image_masks_loss": 0.014, + "llm_learning_rate": 3.3959044368600687e-06, + "loss": 1.0591, + "output_text_mask_loss": 0.2502, + "predictor_learning_rate": 0.00033959044368600685, + "step": 199 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9541, + "image_masks_loss": 0.0326, + "llm_learning_rate": 3.412969283276451e-06, + "loss": 0.9768, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003412969283276451, + "step": 200 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9539, + "image_masks_loss": 0.0315, + "llm_learning_rate": 3.430034129692833e-06, + "loss": 1.1212, + "output_text_mask_loss": 0.2884, + "predictor_learning_rate": 0.00034300341296928326, + "step": 201 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9537, + "image_masks_loss": 0.0322, + "llm_learning_rate": 3.4470989761092157e-06, + "loss": 0.9482, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.0003447098976109215, + "step": 202 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9535, + "image_masks_loss": 0.0269, + "llm_learning_rate": 3.4641638225255976e-06, + "loss": 0.8352, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.00034641638225255974, + "step": 203 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9532, + "image_masks_loss": 0.0315, + "llm_learning_rate": 3.48122866894198e-06, + "loss": 0.8881, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000348122866894198, + "step": 204 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.953, + "image_masks_loss": 0.1091, + "llm_learning_rate": 3.4982935153583618e-06, + "loss": 0.9645, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00034982935153583615, + "step": 205 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9528, + "image_masks_loss": 0.0278, + "llm_learning_rate": 3.5153583617747445e-06, + "loss": 0.9791, + "output_text_mask_loss": 0.3815, + "predictor_learning_rate": 0.0003515358361774744, + "step": 206 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9526, + "image_masks_loss": 0.033, + "llm_learning_rate": 3.5324232081911264e-06, + "loss": 1.0559, + "output_text_mask_loss": 0.1038, + "predictor_learning_rate": 0.0003532423208191126, + "step": 207 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9523, + "image_masks_loss": 0.0462, + "llm_learning_rate": 3.5494880546075087e-06, + "loss": 0.8559, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003549488054607509, + "step": 208 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9521, + "image_masks_loss": 0.033, + "llm_learning_rate": 3.566552901023891e-06, + "loss": 0.8854, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.00035665529010238904, + "step": 209 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9519, + "image_masks_loss": 0.0793, + "llm_learning_rate": 3.5836177474402733e-06, + "loss": 0.9323, + "output_text_mask_loss": 0.0916, + "predictor_learning_rate": 0.0003583617747440273, + "step": 210 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9517, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.600682593856656e-06, + "loss": 0.9161, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.00036006825938566557, + "step": 211 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9514, + "image_masks_loss": 0.0462, + "llm_learning_rate": 3.617747440273038e-06, + "loss": 0.9211, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.0003617747440273038, + "step": 212 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9512, + "image_masks_loss": 0.0252, + "llm_learning_rate": 3.6348122866894202e-06, + "loss": 0.9872, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.000363481228668942, + "step": 213 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.951, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.651877133105802e-06, + "loss": 0.8372, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0003651877133105802, + "step": 214 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9508, + "image_masks_loss": 0.0465, + "llm_learning_rate": 3.668941979522185e-06, + "loss": 0.9219, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00036689419795221846, + "step": 215 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9505, + "image_masks_loss": 0.0534, + "llm_learning_rate": 3.6860068259385667e-06, + "loss": 0.8352, + "output_text_mask_loss": 0.2075, + "predictor_learning_rate": 0.00036860068259385667, + "step": 216 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9503, + "image_masks_loss": 0.0381, + "llm_learning_rate": 3.703071672354949e-06, + "loss": 0.7811, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0003703071672354949, + "step": 217 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9501, + "image_masks_loss": 0.0483, + "llm_learning_rate": 3.7201365187713314e-06, + "loss": 0.8824, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003720136518771331, + "step": 218 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9499, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.7372013651877137e-06, + "loss": 0.9776, + "output_text_mask_loss": 0.1785, + "predictor_learning_rate": 0.00037372013651877135, + "step": 219 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9496, + "image_masks_loss": 0.0378, + "llm_learning_rate": 3.7542662116040956e-06, + "loss": 0.9586, + "output_text_mask_loss": 0.132, + "predictor_learning_rate": 0.00037542662116040956, + "step": 220 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9494, + "image_masks_loss": 0.0368, + "llm_learning_rate": 3.7713310580204783e-06, + "loss": 0.8727, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00037713310580204777, + "step": 221 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9492, + "image_masks_loss": 0.018, + "llm_learning_rate": 3.78839590443686e-06, + "loss": 0.8275, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000378839590443686, + "step": 222 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.949, + "image_masks_loss": 0.0456, + "llm_learning_rate": 3.8054607508532425e-06, + "loss": 0.9244, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00038054607508532424, + "step": 223 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9487, + "image_masks_loss": 0.0347, + "llm_learning_rate": 3.822525597269625e-06, + "loss": 0.8746, + "output_text_mask_loss": 0.0923, + "predictor_learning_rate": 0.0003822525597269625, + "step": 224 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9485, + "image_masks_loss": 0.0641, + "llm_learning_rate": 3.839590443686007e-06, + "loss": 0.8642, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038395904436860066, + "step": 225 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9483, + "image_masks_loss": 0.1038, + "llm_learning_rate": 3.8566552901023894e-06, + "loss": 1.0432, + "output_text_mask_loss": 0.1816, + "predictor_learning_rate": 0.0003856655290102389, + "step": 226 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9481, + "image_masks_loss": 0.0148, + "llm_learning_rate": 3.873720136518771e-06, + "loss": 0.9725, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 0.00038737201365187713, + "step": 227 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9479, + "image_masks_loss": 0.0366, + "llm_learning_rate": 3.890784982935154e-06, + "loss": 0.8957, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.0003890784982935154, + "step": 228 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9476, + "image_masks_loss": 0.0771, + "llm_learning_rate": 3.907849829351536e-06, + "loss": 0.9074, + "output_text_mask_loss": 0.1419, + "predictor_learning_rate": 0.0003907849829351536, + "step": 229 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9474, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.924914675767919e-06, + "loss": 0.9179, + "output_text_mask_loss": 0.0338, + "predictor_learning_rate": 0.0003924914675767918, + "step": 230 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9472, + "image_masks_loss": 0.0275, + "llm_learning_rate": 3.9419795221843006e-06, + "loss": 0.883, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.00039419795221843, + "step": 231 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.947, + "image_masks_loss": 0.0648, + "llm_learning_rate": 3.959044368600683e-06, + "loss": 0.9598, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0003959044368600683, + "step": 232 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9467, + "image_masks_loss": 0.0374, + "llm_learning_rate": 3.976109215017065e-06, + "loss": 0.856, + "output_text_mask_loss": 0.1305, + "predictor_learning_rate": 0.0003976109215017065, + "step": 233 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9465, + "image_masks_loss": 0.0519, + "llm_learning_rate": 3.993174061433447e-06, + "loss": 0.8509, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0003993174061433447, + "step": 234 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9463, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.01023890784983e-06, + "loss": 0.7986, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00040102389078498297, + "step": 235 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9461, + "image_masks_loss": 0.0782, + "llm_learning_rate": 4.027303754266212e-06, + "loss": 0.8143, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0004027303754266212, + "step": 236 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9458, + "image_masks_loss": 0.0195, + "llm_learning_rate": 4.044368600682594e-06, + "loss": 0.8543, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00040443686006825944, + "step": 237 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9456, + "image_masks_loss": 0.0439, + "llm_learning_rate": 4.061433447098976e-06, + "loss": 0.899, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0004061433447098976, + "step": 238 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9454, + "image_masks_loss": 0.0839, + "llm_learning_rate": 4.078498293515359e-06, + "loss": 0.8814, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.00040784982935153586, + "step": 239 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9452, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.095563139931741e-06, + "loss": 0.8872, + "output_text_mask_loss": 0.1205, + "predictor_learning_rate": 0.00040955631399317407, + "step": 240 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.945, + "image_masks_loss": 0.0813, + "llm_learning_rate": 4.112627986348124e-06, + "loss": 0.9527, + "output_text_mask_loss": 0.1068, + "predictor_learning_rate": 0.00041126279863481233, + "step": 241 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9447, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.1296928327645055e-06, + "loss": 0.8074, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0004129692832764505, + "step": 242 + }, + { + "epoch": 0.02, + "gumbel_tau": 0.9445, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.1467576791808874e-06, + "loss": 0.998, + "output_text_mask_loss": 0.1793, + "predictor_learning_rate": 0.00041467576791808875, + "step": 243 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9443, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.163822525597269e-06, + "loss": 0.8935, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00041638225255972696, + "step": 244 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9441, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.180887372013652e-06, + "loss": 0.8686, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.0004180887372013652, + "step": 245 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9438, + "image_masks_loss": 0.042, + "llm_learning_rate": 4.197952218430034e-06, + "loss": 0.8741, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004197952218430034, + "step": 246 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9436, + "image_masks_loss": 0.0763, + "llm_learning_rate": 4.215017064846417e-06, + "loss": 0.9365, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00042150170648464164, + "step": 247 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9434, + "image_masks_loss": 0.0351, + "llm_learning_rate": 4.232081911262799e-06, + "loss": 0.9097, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004232081911262799, + "step": 248 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9432, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.249146757679181e-06, + "loss": 0.8887, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004249146757679181, + "step": 249 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9429, + "image_masks_loss": 0.0229, + "llm_learning_rate": 4.266211604095564e-06, + "loss": 0.9336, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 0.0004266211604095563, + "step": 250 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9427, + "image_masks_loss": 0.0595, + "llm_learning_rate": 4.283276450511946e-06, + "loss": 0.8548, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00042832764505119453, + "step": 251 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9425, + "image_masks_loss": 0.1007, + "llm_learning_rate": 4.300341296928328e-06, + "loss": 1.0235, + "output_text_mask_loss": 0.2258, + "predictor_learning_rate": 0.0004300341296928328, + "step": 252 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9423, + "image_masks_loss": 0.1007, + "llm_learning_rate": 4.31740614334471e-06, + "loss": 1.0307, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000431740614334471, + "step": 253 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9421, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.3344709897610924e-06, + "loss": 0.9835, + "output_text_mask_loss": 0.3113, + "predictor_learning_rate": 0.0004334470989761092, + "step": 254 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9418, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.351535836177474e-06, + "loss": 1.1469, + "output_text_mask_loss": 0.0885, + "predictor_learning_rate": 0.0004351535836177474, + "step": 255 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9416, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.368600682593857e-06, + "loss": 0.9815, + "output_text_mask_loss": 0.1602, + "predictor_learning_rate": 0.0004368600682593857, + "step": 256 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9414, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.385665529010239e-06, + "loss": 0.9434, + "output_text_mask_loss": 0.2991, + "predictor_learning_rate": 0.0004385665529010239, + "step": 257 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9412, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.402730375426622e-06, + "loss": 0.9817, + "output_text_mask_loss": 0.1495, + "predictor_learning_rate": 0.0004402730375426621, + "step": 258 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9409, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.419795221843004e-06, + "loss": 0.8758, + "output_text_mask_loss": 0.0694, + "predictor_learning_rate": 0.00044197952218430036, + "step": 259 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9407, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.436860068259386e-06, + "loss": 1.0058, + "output_text_mask_loss": 0.1244, + "predictor_learning_rate": 0.00044368600682593857, + "step": 260 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9405, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.453924914675768e-06, + "loss": 0.8361, + "output_text_mask_loss": 0.0927, + "predictor_learning_rate": 0.00044539249146757683, + "step": 261 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9403, + "image_masks_loss": 0.0629, + "llm_learning_rate": 4.47098976109215e-06, + "loss": 1.0199, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.00044709897610921504, + "step": 262 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9401, + "image_masks_loss": 0.1472, + "llm_learning_rate": 4.488054607508533e-06, + "loss": 0.9094, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00044880546075085325, + "step": 263 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9398, + "image_masks_loss": 0.0553, + "llm_learning_rate": 4.505119453924915e-06, + "loss": 0.8607, + "output_text_mask_loss": 0.1083, + "predictor_learning_rate": 0.00045051194539249146, + "step": 264 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9396, + "image_masks_loss": 0.0908, + "llm_learning_rate": 4.522184300341297e-06, + "loss": 1.0556, + "output_text_mask_loss": 0.1717, + "predictor_learning_rate": 0.0004522184300341297, + "step": 265 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9394, + "image_masks_loss": 0.0877, + "llm_learning_rate": 4.539249146757679e-06, + "loss": 1.2271, + "output_text_mask_loss": 0.4486, + "predictor_learning_rate": 0.00045392491467576793, + "step": 266 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9392, + "image_masks_loss": 0.0549, + "llm_learning_rate": 4.556313993174062e-06, + "loss": 1.0062, + "output_text_mask_loss": 0.1564, + "predictor_learning_rate": 0.00045563139931740614, + "step": 267 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.939, + "image_masks_loss": 0.0385, + "llm_learning_rate": 4.573378839590444e-06, + "loss": 0.95, + "output_text_mask_loss": 0.2838, + "predictor_learning_rate": 0.00045733788395904435, + "step": 268 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9387, + "image_masks_loss": 0.046, + "llm_learning_rate": 4.590443686006827e-06, + "loss": 1.0818, + "output_text_mask_loss": 0.47, + "predictor_learning_rate": 0.0004590443686006826, + "step": 269 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9385, + "image_masks_loss": 0.0835, + "llm_learning_rate": 4.6075085324232085e-06, + "loss": 0.8902, + "output_text_mask_loss": 0.1892, + "predictor_learning_rate": 0.0004607508532423209, + "step": 270 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9383, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.6245733788395904e-06, + "loss": 0.8734, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00046245733788395903, + "step": 271 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9381, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.641638225255973e-06, + "loss": 0.8601, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0004641638225255973, + "step": 272 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9378, + "image_masks_loss": 0.0054, + "llm_learning_rate": 4.658703071672355e-06, + "loss": 0.9113, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004658703071672355, + "step": 273 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9376, + "image_masks_loss": 0.0549, + "llm_learning_rate": 4.675767918088738e-06, + "loss": 0.9157, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00046757679180887377, + "step": 274 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9374, + "image_masks_loss": 0.0793, + "llm_learning_rate": 4.69283276450512e-06, + "loss": 0.8396, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 0.0004692832764505119, + "step": 275 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9372, + "image_masks_loss": 0.0341, + "llm_learning_rate": 4.709897610921502e-06, + "loss": 0.84, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004709897610921502, + "step": 276 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.937, + "image_masks_loss": 0.0385, + "llm_learning_rate": 4.726962457337884e-06, + "loss": 0.7764, + "output_text_mask_loss": 0.1282, + "predictor_learning_rate": 0.0004726962457337884, + "step": 277 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9367, + "image_masks_loss": 0.079, + "llm_learning_rate": 4.744027303754267e-06, + "loss": 0.8857, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.00047440273037542666, + "step": 278 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9365, + "image_masks_loss": 0.058, + "llm_learning_rate": 4.761092150170649e-06, + "loss": 0.8904, + "output_text_mask_loss": 0.2792, + "predictor_learning_rate": 0.0004761092150170648, + "step": 279 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9363, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.778156996587031e-06, + "loss": 1.009, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004778156996587031, + "step": 280 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9361, + "image_masks_loss": 0.0204, + "llm_learning_rate": 4.795221843003413e-06, + "loss": 1.1617, + "output_text_mask_loss": 0.2411, + "predictor_learning_rate": 0.0004795221843003413, + "step": 281 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9359, + "image_masks_loss": 0.0452, + "llm_learning_rate": 4.812286689419795e-06, + "loss": 1.0645, + "output_text_mask_loss": 0.4303, + "predictor_learning_rate": 0.00048122866894197955, + "step": 282 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9356, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.829351535836178e-06, + "loss": 0.9589, + "output_text_mask_loss": 0.1587, + "predictor_learning_rate": 0.00048293515358361776, + "step": 283 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9354, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.84641638225256e-06, + "loss": 0.7861, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048464163822525597, + "step": 284 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9352, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.863481228668943e-06, + "loss": 0.7513, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048634812286689423, + "step": 285 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.935, + "image_masks_loss": 0.0483, + "llm_learning_rate": 4.880546075085325e-06, + "loss": 0.8173, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00048805460750853244, + "step": 286 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9348, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.897610921501707e-06, + "loss": 0.8429, + "output_text_mask_loss": 0.238, + "predictor_learning_rate": 0.0004897610921501706, + "step": 287 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9345, + "image_masks_loss": 0.0626, + "llm_learning_rate": 4.914675767918089e-06, + "loss": 0.8542, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.0004914675767918089, + "step": 288 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9343, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.931740614334471e-06, + "loss": 0.898, + "output_text_mask_loss": 0.1625, + "predictor_learning_rate": 0.0004931740614334471, + "step": 289 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9341, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.948805460750853e-06, + "loss": 0.9843, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.0004948805460750853, + "step": 290 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9339, + "image_masks_loss": 0.0345, + "llm_learning_rate": 4.965870307167236e-06, + "loss": 0.8433, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0004965870307167236, + "step": 291 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9337, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.982935153583618e-06, + "loss": 0.9095, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004982935153583618, + "step": 292 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9334, + "image_masks_loss": 0.0284, + "llm_learning_rate": 5e-06, + "loss": 0.8123, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0005, + "step": 293 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9332, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.999999862347216e-06, + "loss": 0.7864, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999999862347216, + "step": 294 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.933, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.999999449388878e-06, + "loss": 0.8175, + "output_text_mask_loss": 0.1717, + "predictor_learning_rate": 0.0004999999449388878, + "step": 295 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9328, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.9999987611250335e-06, + "loss": 0.8707, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.0004999998761125033, + "step": 296 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9326, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.999997797555756e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004999997797555755, + "step": 297 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9323, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.999996558681152e-06, + "loss": 0.9039, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999996558681151, + "step": 298 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9321, + "image_masks_loss": 0.045, + "llm_learning_rate": 4.999995044501359e-06, + "loss": 0.7929, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004999995044501358, + "step": 299 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9319, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.999993255016543e-06, + "loss": 0.8932, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004999993255016543, + "step": 300 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9317, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.999991190226901e-06, + "loss": 0.8906, + "output_text_mask_loss": 0.1434, + "predictor_learning_rate": 0.00049999911902269, + "step": 301 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9315, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.99998885013266e-06, + "loss": 0.962, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.000499998885013266, + "step": 302 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9312, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.99998623473408e-06, + "loss": 0.9058, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000499998623473408, + "step": 303 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.931, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.999983344031446e-06, + "loss": 0.7237, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999983344031446, + "step": 304 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9308, + "image_masks_loss": 0.0105, + "llm_learning_rate": 4.999980178025078e-06, + "loss": 1.0101, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004999980178025078, + "step": 305 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9306, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.999976736715325e-06, + "loss": 0.9071, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999976736715325, + "step": 306 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9304, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.999973020102564e-06, + "loss": 0.8728, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0004999973020102564, + "step": 307 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9301, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.999969028187207e-06, + "loss": 0.864, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0004999969028187207, + "step": 308 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9299, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.999964760969691e-06, + "loss": 0.8931, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004999964760969691, + "step": 309 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9297, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.9999602184504884e-06, + "loss": 0.8917, + "output_text_mask_loss": 0.0087, + "predictor_learning_rate": 0.0004999960218450488, + "step": 310 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9295, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.999955400630096e-06, + "loss": 0.8354, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 0.0004999955400630097, + "step": 311 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9293, + "image_masks_loss": 0.0507, + "llm_learning_rate": 4.999950307509048e-06, + "loss": 0.7159, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.0004999950307509048, + "step": 312 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.929, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.999944939087903e-06, + "loss": 0.8964, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999944939087903, + "step": 313 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9288, + "image_masks_loss": 0.0402, + "llm_learning_rate": 4.999939295367254e-06, + "loss": 0.8996, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004999939295367254, + "step": 314 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9286, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.999933376347721e-06, + "loss": 0.9399, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.000499993337634772, + "step": 315 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9284, + "image_masks_loss": 0.0383, + "llm_learning_rate": 4.999927182029955e-06, + "loss": 0.8665, + "output_text_mask_loss": 0.0629, + "predictor_learning_rate": 0.0004999927182029955, + "step": 316 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9282, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.999920712414641e-06, + "loss": 0.7372, + "output_text_mask_loss": 0.0553, + "predictor_learning_rate": 0.0004999920712414641, + "step": 317 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9279, + "image_masks_loss": 0.0198, + "llm_learning_rate": 4.9999139675024885e-06, + "loss": 0.8607, + "output_text_mask_loss": 0.1534, + "predictor_learning_rate": 0.0004999913967502488, + "step": 318 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9277, + "image_masks_loss": 0.0085, + "llm_learning_rate": 4.9999069472942415e-06, + "loss": 0.8246, + "output_text_mask_loss": 0.1953, + "predictor_learning_rate": 0.0004999906947294241, + "step": 319 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9275, + "image_masks_loss": 0.0173, + "llm_learning_rate": 4.999899651790675e-06, + "loss": 0.8511, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004999899651790674, + "step": 320 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9273, + "image_masks_loss": 0.0368, + "llm_learning_rate": 4.999892080992589e-06, + "loss": 0.9007, + "output_text_mask_loss": 0.025, + "predictor_learning_rate": 0.0004999892080992588, + "step": 321 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9271, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.999884234900818e-06, + "loss": 0.8497, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004999884234900818, + "step": 322 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9268, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.999876113516229e-06, + "loss": 0.9316, + "output_text_mask_loss": 0.1518, + "predictor_learning_rate": 0.0004999876113516229, + "step": 323 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9266, + "image_masks_loss": 0.0155, + "llm_learning_rate": 4.9998677168397124e-06, + "loss": 0.7027, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999867716839712, + "step": 324 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9264, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.999859044872195e-06, + "loss": 0.79, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999859044872195, + "step": 325 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9262, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.9998500976146316e-06, + "loss": 0.9394, + "output_text_mask_loss": 0.0717, + "predictor_learning_rate": 0.0004999850097614631, + "step": 326 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.926, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.9998408750680075e-06, + "loss": 0.8599, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004999840875068007, + "step": 327 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9258, + "image_masks_loss": 0.0572, + "llm_learning_rate": 4.999831377233337e-06, + "loss": 0.8828, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999831377233337, + "step": 328 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9255, + "image_masks_loss": 0.0397, + "llm_learning_rate": 4.999821604111668e-06, + "loss": 0.8645, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0004999821604111667, + "step": 329 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9253, + "image_masks_loss": 0.0402, + "llm_learning_rate": 4.9998115557040754e-06, + "loss": 0.7591, + "output_text_mask_loss": 0.0393, + "predictor_learning_rate": 0.0004999811555704075, + "step": 330 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9251, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.999801232011666e-06, + "loss": 0.8883, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004999801232011666, + "step": 331 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9249, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.999790633035577e-06, + "loss": 0.9099, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004999790633035577, + "step": 332 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9247, + "image_masks_loss": 0.0047, + "llm_learning_rate": 4.999779758776976e-06, + "loss": 0.8214, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.0004999779758776975, + "step": 333 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9244, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.99976860923706e-06, + "loss": 0.814, + "output_text_mask_loss": 0.1495, + "predictor_learning_rate": 0.000499976860923706, + "step": 334 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9242, + "image_masks_loss": 0.0376, + "llm_learning_rate": 4.999757184417056e-06, + "loss": 0.8577, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004999757184417055, + "step": 335 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.924, + "image_masks_loss": 0.0353, + "llm_learning_rate": 4.999745484318223e-06, + "loss": 0.801, + "output_text_mask_loss": 0.1091, + "predictor_learning_rate": 0.0004999745484318222, + "step": 336 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9238, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.9997335089418496e-06, + "loss": 0.7833, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004999733508941849, + "step": 337 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9236, + "image_masks_loss": 0.0307, + "llm_learning_rate": 4.999721258289254e-06, + "loss": 0.7735, + "output_text_mask_loss": 0.0648, + "predictor_learning_rate": 0.0004999721258289254, + "step": 338 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9234, + "image_masks_loss": 0.0192, + "llm_learning_rate": 4.999708732361785e-06, + "loss": 0.8779, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999708732361785, + "step": 339 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9231, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.999695931160824e-06, + "loss": 0.9111, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004999695931160824, + "step": 340 + }, + { + "epoch": 0.03, + "gumbel_tau": 0.9229, + "image_masks_loss": 0.0303, + "llm_learning_rate": 4.999682854687778e-06, + "loss": 0.8551, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.0004999682854687778, + "step": 341 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9227, + "image_masks_loss": 0.0353, + "llm_learning_rate": 4.999669502944089e-06, + "loss": 0.908, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.0004999669502944088, + "step": 342 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9225, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.999655875931225e-06, + "loss": 0.6894, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999655875931225, + "step": 343 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9223, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.99964197365069e-06, + "loss": 0.8282, + "output_text_mask_loss": 0.2167, + "predictor_learning_rate": 0.0004999641973650689, + "step": 344 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.922, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.999627796104012e-06, + "loss": 0.7943, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004999627796104013, + "step": 345 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9218, + "image_masks_loss": 0.0698, + "llm_learning_rate": 4.999613343292754e-06, + "loss": 0.9849, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0004999613343292754, + "step": 346 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9216, + "image_masks_loss": 0.0343, + "llm_learning_rate": 4.9995986152185075e-06, + "loss": 0.8854, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.0004999598615218507, + "step": 347 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9214, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.999583611882892e-06, + "loss": 0.8579, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.0004999583611882892, + "step": 348 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9212, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.999568333287563e-06, + "loss": 0.7768, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004999568333287563, + "step": 349 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.921, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.999552779434201e-06, + "loss": 0.7951, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0004999552779434201, + "step": 350 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9207, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.99953695032452e-06, + "loss": 0.8058, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000499953695032452, + "step": 351 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9205, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.999520845960262e-06, + "loss": 0.8554, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0004999520845960262, + "step": 352 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9203, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.999504466343202e-06, + "loss": 0.7921, + "output_text_mask_loss": 0.0885, + "predictor_learning_rate": 0.0004999504466343201, + "step": 353 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9201, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.999487811475141e-06, + "loss": 0.908, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.000499948781147514, + "step": 354 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9199, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.999470881357915e-06, + "loss": 0.8269, + "output_text_mask_loss": 0.0475, + "predictor_learning_rate": 0.0004999470881357916, + "step": 355 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9197, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.99945367599339e-06, + "loss": 0.9074, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.000499945367599339, + "step": 356 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9194, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.999436195383457e-06, + "loss": 0.936, + "output_text_mask_loss": 0.1312, + "predictor_learning_rate": 0.0004999436195383456, + "step": 357 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9192, + "image_masks_loss": 0.0238, + "llm_learning_rate": 4.999418439530044e-06, + "loss": 0.7935, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004999418439530043, + "step": 358 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.919, + "image_masks_loss": 0.0389, + "llm_learning_rate": 4.9994004084351045e-06, + "loss": 0.9073, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004999400408435104, + "step": 359 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9188, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.999382102100625e-06, + "loss": 0.7435, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004999382102100625, + "step": 360 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9186, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.999363520528622e-06, + "loss": 0.8273, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004999363520528621, + "step": 361 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9184, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.999344663721141e-06, + "loss": 0.906, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.000499934466372114, + "step": 362 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9181, + "image_masks_loss": 0.0111, + "llm_learning_rate": 4.9993255316802565e-06, + "loss": 0.8343, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004999325531680256, + "step": 363 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9179, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.99930612440808e-06, + "loss": 0.7641, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004999306124408079, + "step": 364 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9177, + "image_masks_loss": 0.0538, + "llm_learning_rate": 4.9992864419067445e-06, + "loss": 0.8156, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999286441906744, + "step": 365 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9175, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.9992664841784196e-06, + "loss": 0.7456, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999266484178419, + "step": 366 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9173, + "image_masks_loss": 0.0805, + "llm_learning_rate": 4.999246251225303e-06, + "loss": 0.856, + "output_text_mask_loss": 0.1831, + "predictor_learning_rate": 0.0004999246251225302, + "step": 367 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9171, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.999225743049622e-06, + "loss": 0.7465, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 0.0004999225743049622, + "step": 368 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9168, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.999204959653635e-06, + "loss": 0.8088, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999204959653634, + "step": 369 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9166, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.999183901039631e-06, + "loss": 0.7836, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004999183901039632, + "step": 370 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9164, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.99916256720993e-06, + "loss": 0.8242, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000499916256720993, + "step": 371 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9162, + "image_masks_loss": 0.0189, + "llm_learning_rate": 4.999140958166879e-06, + "loss": 0.9184, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004999140958166879, + "step": 372 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.916, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.99911907391286e-06, + "loss": 0.8167, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.000499911907391286, + "step": 373 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9158, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.9990969144502824e-06, + "loss": 0.8355, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 0.0004999096914450282, + "step": 374 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9155, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.999074479781586e-06, + "loss": 0.7632, + "output_text_mask_loss": 0.0406, + "predictor_learning_rate": 0.0004999074479781585, + "step": 375 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9153, + "image_masks_loss": 0.0075, + "llm_learning_rate": 4.999051769909241e-06, + "loss": 0.9124, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0004999051769909241, + "step": 376 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9151, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.999028784835749e-06, + "loss": 0.9003, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.0004999028784835748, + "step": 377 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9149, + "image_masks_loss": 0.0307, + "llm_learning_rate": 4.99900552456364e-06, + "loss": 0.8423, + "output_text_mask_loss": 0.1297, + "predictor_learning_rate": 0.000499900552456364, + "step": 378 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9147, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.998981989095477e-06, + "loss": 0.7539, + "output_text_mask_loss": 0.0877, + "predictor_learning_rate": 0.0004998981989095477, + "step": 379 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9145, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.998958178433851e-06, + "loss": 0.8778, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.000499895817843385, + "step": 380 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9143, + "image_masks_loss": 0.0706, + "llm_learning_rate": 4.998934092581384e-06, + "loss": 0.8819, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0004998934092581384, + "step": 381 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.914, + "image_masks_loss": 0.0511, + "llm_learning_rate": 4.998909731540728e-06, + "loss": 0.7686, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0004998909731540729, + "step": 382 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9138, + "image_masks_loss": 0.0416, + "llm_learning_rate": 4.998885095314568e-06, + "loss": 0.8191, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004998885095314567, + "step": 383 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9136, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.9988601839056135e-06, + "loss": 0.9105, + "output_text_mask_loss": 0.0954, + "predictor_learning_rate": 0.0004998860183905614, + "step": 384 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9134, + "image_masks_loss": 0.0481, + "llm_learning_rate": 4.998834997316611e-06, + "loss": 0.8114, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.000499883499731661, + "step": 385 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9132, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.9988095355503316e-06, + "loss": 0.7924, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004998809535550332, + "step": 386 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.913, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.998783798609581e-06, + "loss": 0.8128, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.000499878379860958, + "step": 387 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9127, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.998757786497192e-06, + "loss": 0.9912, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.0004998757786497191, + "step": 388 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9125, + "image_masks_loss": 0.042, + "llm_learning_rate": 4.99873149921603e-06, + "loss": 1.1995, + "output_text_mask_loss": 0.3876, + "predictor_learning_rate": 0.000499873149921603, + "step": 389 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9123, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.99870493676899e-06, + "loss": 1.215, + "output_text_mask_loss": 0.1854, + "predictor_learning_rate": 0.000499870493676899, + "step": 390 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9121, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.998678099158997e-06, + "loss": 0.9877, + "output_text_mask_loss": 0.3296, + "predictor_learning_rate": 0.0004998678099158996, + "step": 391 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9119, + "image_masks_loss": 0.061, + "llm_learning_rate": 4.998650986389005e-06, + "loss": 0.8004, + "output_text_mask_loss": 0.1282, + "predictor_learning_rate": 0.0004998650986389005, + "step": 392 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9117, + "image_masks_loss": 0.0595, + "llm_learning_rate": 4.9986235984620016e-06, + "loss": 0.8807, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004998623598462002, + "step": 393 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9115, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.998595935381001e-06, + "loss": 0.8469, + "output_text_mask_loss": 0.0942, + "predictor_learning_rate": 0.0004998595935381001, + "step": 394 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9112, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.998567997149052e-06, + "loss": 0.8371, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004998567997149052, + "step": 395 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.911, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.998539783769229e-06, + "loss": 1.0029, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000499853978376923, + "step": 396 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9108, + "image_masks_loss": 0.0591, + "llm_learning_rate": 4.9985112952446406e-06, + "loss": 1.0179, + "output_text_mask_loss": 0.2075, + "predictor_learning_rate": 0.000499851129524464, + "step": 397 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9106, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.998482531578423e-06, + "loss": 1.1628, + "output_text_mask_loss": 0.1793, + "predictor_learning_rate": 0.0004998482531578422, + "step": 398 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9104, + "image_masks_loss": 0.0429, + "llm_learning_rate": 4.9984534927737425e-06, + "loss": 1.2669, + "output_text_mask_loss": 0.8179, + "predictor_learning_rate": 0.0004998453492773742, + "step": 399 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9102, + "image_masks_loss": 0.0319, + "llm_learning_rate": 4.9984241788337994e-06, + "loss": 1.1858, + "output_text_mask_loss": 0.3052, + "predictor_learning_rate": 0.0004998424178833799, + "step": 400 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9099, + "image_masks_loss": 0.0228, + "llm_learning_rate": 4.998394589761821e-06, + "loss": 1.1818, + "output_text_mask_loss": 0.3937, + "predictor_learning_rate": 0.000499839458976182, + "step": 401 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9097, + "image_masks_loss": 0.0234, + "llm_learning_rate": 4.998364725561065e-06, + "loss": 0.878, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004998364725561065, + "step": 402 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9095, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.9983345862348195e-06, + "loss": 0.792, + "output_text_mask_loss": 0.1411, + "predictor_learning_rate": 0.000499833458623482, + "step": 403 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9093, + "image_masks_loss": 0.0431, + "llm_learning_rate": 4.998304171786406e-06, + "loss": 0.8618, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004998304171786405, + "step": 404 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9091, + "image_masks_loss": 0.0139, + "llm_learning_rate": 4.998273482219172e-06, + "loss": 0.854, + "output_text_mask_loss": 0.0119, + "predictor_learning_rate": 0.0004998273482219171, + "step": 405 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9089, + "image_masks_loss": 0.0355, + "llm_learning_rate": 4.998242517536497e-06, + "loss": 0.9303, + "output_text_mask_loss": 0.1022, + "predictor_learning_rate": 0.0004998242517536497, + "step": 406 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9087, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.998211277741792e-06, + "loss": 0.8426, + "output_text_mask_loss": 0.0942, + "predictor_learning_rate": 0.000499821127774179, + "step": 407 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9084, + "image_masks_loss": 0.0343, + "llm_learning_rate": 4.998179762838496e-06, + "loss": 0.7287, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004998179762838495, + "step": 408 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9082, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.99814797283008e-06, + "loss": 0.7631, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000499814797283008, + "step": 409 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.908, + "image_masks_loss": 0.0219, + "llm_learning_rate": 4.998115907720045e-06, + "loss": 0.845, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004998115907720044, + "step": 410 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9078, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.998083567511922e-06, + "loss": 0.9346, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004998083567511922, + "step": 411 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9076, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.998050952209273e-06, + "loss": 0.7939, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004998050952209273, + "step": 412 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9074, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.998018061815687e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 0.0004998018061815687, + "step": 413 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9072, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.99798489633479e-06, + "loss": 0.8074, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004997984896334789, + "step": 414 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9069, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.997951455770231e-06, + "loss": 0.8602, + "output_text_mask_loss": 0.1419, + "predictor_learning_rate": 0.000499795145577023, + "step": 415 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9067, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.997917740125694e-06, + "loss": 0.8683, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004997917740125694, + "step": 416 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9065, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.997883749404891e-06, + "loss": 0.863, + "output_text_mask_loss": 0.0778, + "predictor_learning_rate": 0.0004997883749404892, + "step": 417 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9063, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.997849483611566e-06, + "loss": 0.9017, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004997849483611566, + "step": 418 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9061, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.997814942749493e-06, + "loss": 0.8212, + "output_text_mask_loss": 0.3387, + "predictor_learning_rate": 0.0004997814942749492, + "step": 419 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9059, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.997780126822475e-06, + "loss": 0.946, + "output_text_mask_loss": 0.2975, + "predictor_learning_rate": 0.0004997780126822474, + "step": 420 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9057, + "image_masks_loss": 0.0355, + "llm_learning_rate": 4.997745035834344e-06, + "loss": 0.8377, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0004997745035834344, + "step": 421 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9055, + "image_masks_loss": 0.0446, + "llm_learning_rate": 4.997709669788968e-06, + "loss": 0.7924, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004997709669788968, + "step": 422 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9052, + "image_masks_loss": 0.0395, + "llm_learning_rate": 4.9976740286902385e-06, + "loss": 0.8698, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.0004997674028690239, + "step": 423 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.905, + "image_masks_loss": 0.0059, + "llm_learning_rate": 4.997638112542083e-06, + "loss": 0.8597, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004997638112542082, + "step": 424 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9048, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.997601921348454e-06, + "loss": 0.8212, + "output_text_mask_loss": 0.116, + "predictor_learning_rate": 0.0004997601921348454, + "step": 425 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9046, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.997565455113338e-06, + "loss": 0.8727, + "output_text_mask_loss": 0.0481, + "predictor_learning_rate": 0.0004997565455113338, + "step": 426 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9044, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.997528713840752e-06, + "loss": 0.8893, + "output_text_mask_loss": 0.1053, + "predictor_learning_rate": 0.0004997528713840751, + "step": 427 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9042, + "image_masks_loss": 0.0058, + "llm_learning_rate": 4.99749169753474e-06, + "loss": 0.8938, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.000499749169753474, + "step": 428 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.904, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.9974544061993805e-06, + "loss": 0.9037, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.000499745440619938, + "step": 429 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9037, + "image_masks_loss": 0.0717, + "llm_learning_rate": 4.997416839838778e-06, + "loss": 0.7443, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.0004997416839838777, + "step": 430 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9035, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.997378998457071e-06, + "loss": 0.9455, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.000499737899845707, + "step": 431 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9033, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.997340882058425e-06, + "loss": 0.8482, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004997340882058425, + "step": 432 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9031, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.997302490647039e-06, + "loss": 0.8956, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0004997302490647039, + "step": 433 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9029, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.997263824227141e-06, + "loss": 0.863, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004997263824227141, + "step": 434 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9027, + "image_masks_loss": 0.0683, + "llm_learning_rate": 4.997224882802987e-06, + "loss": 1.0046, + "output_text_mask_loss": 0.3357, + "predictor_learning_rate": 0.0004997224882802987, + "step": 435 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9025, + "image_masks_loss": 0.0108, + "llm_learning_rate": 4.997185666378867e-06, + "loss": 0.8675, + "output_text_mask_loss": 0.0626, + "predictor_learning_rate": 0.0004997185666378867, + "step": 436 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9023, + "image_masks_loss": 0.042, + "llm_learning_rate": 4.9971461749591e-06, + "loss": 0.859, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004997146174959099, + "step": 437 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.902, + "image_masks_loss": 0.0591, + "llm_learning_rate": 4.997106408548032e-06, + "loss": 0.9703, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0004997106408548033, + "step": 438 + }, + { + "epoch": 0.04, + "gumbel_tau": 0.9018, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.9970663671500464e-06, + "loss": 0.8304, + "output_text_mask_loss": 0.1427, + "predictor_learning_rate": 0.0004997066367150047, + "step": 439 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.9016, + "image_masks_loss": 0.0341, + "llm_learning_rate": 4.997026050769549e-06, + "loss": 0.9341, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004997026050769549, + "step": 440 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.9014, + "image_masks_loss": 0.0349, + "llm_learning_rate": 4.996985459410982e-06, + "loss": 0.8122, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004996985459410981, + "step": 441 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.9012, + "image_masks_loss": 0.0614, + "llm_learning_rate": 4.996944593078814e-06, + "loss": 0.8893, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004996944593078814, + "step": 442 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.901, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.996903451777545e-06, + "loss": 0.7428, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.0004996903451777544, + "step": 443 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.9008, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.996862035511707e-06, + "loss": 0.9257, + "output_text_mask_loss": 0.1083, + "predictor_learning_rate": 0.0004996862035511707, + "step": 444 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.9006, + "image_masks_loss": 0.05, + "llm_learning_rate": 4.9968203442858595e-06, + "loss": 0.8302, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.000499682034428586, + "step": 445 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.9003, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.996778378104595e-06, + "loss": 0.743, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.0004996778378104595, + "step": 446 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.9001, + "image_masks_loss": 0.0496, + "llm_learning_rate": 4.996736136972534e-06, + "loss": 0.7406, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004996736136972533, + "step": 447 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8999, + "image_masks_loss": 0.0359, + "llm_learning_rate": 4.996693620894329e-06, + "loss": 0.8146, + "output_text_mask_loss": 0.0935, + "predictor_learning_rate": 0.0004996693620894328, + "step": 448 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8997, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.99665082987466e-06, + "loss": 0.7452, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004996650829874659, + "step": 449 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8995, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.9966077639182405e-06, + "loss": 0.8395, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0004996607763918241, + "step": 450 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8993, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.996564423029814e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0004996564423029814, + "step": 451 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8991, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.996520807214152e-06, + "loss": 0.8217, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004996520807214152, + "step": 452 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8989, + "image_masks_loss": 0.0523, + "llm_learning_rate": 4.996476916476057e-06, + "loss": 0.7949, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0004996476916476057, + "step": 453 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8986, + "image_masks_loss": 0.0721, + "llm_learning_rate": 4.996432750820365e-06, + "loss": 0.8756, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.0004996432750820365, + "step": 454 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8984, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.996388310251937e-06, + "loss": 0.76, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004996388310251936, + "step": 455 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8982, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.996343594775668e-06, + "loss": 0.7143, + "output_text_mask_loss": 0.1114, + "predictor_learning_rate": 0.0004996343594775667, + "step": 456 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.898, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.996298604396481e-06, + "loss": 0.8919, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.000499629860439648, + "step": 457 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8978, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.996253339119333e-06, + "loss": 0.8022, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.0004996253339119332, + "step": 458 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8976, + "image_masks_loss": 0.0614, + "llm_learning_rate": 4.996207798949206e-06, + "loss": 0.748, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004996207798949206, + "step": 459 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8974, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.996161983891116e-06, + "loss": 0.8239, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004996161983891116, + "step": 460 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8972, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.996115893950109e-06, + "loss": 0.8528, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004996115893950108, + "step": 461 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8969, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.99606952913126e-06, + "loss": 0.9182, + "output_text_mask_loss": 0.0969, + "predictor_learning_rate": 0.0004996069529131259, + "step": 462 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8967, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.996022889439674e-06, + "loss": 0.8261, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004996022889439674, + "step": 463 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8965, + "image_masks_loss": 0.0319, + "llm_learning_rate": 4.995975974880488e-06, + "loss": 0.9738, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004995975974880488, + "step": 464 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8963, + "image_masks_loss": 0.0546, + "llm_learning_rate": 4.995928785458868e-06, + "loss": 0.8601, + "output_text_mask_loss": 0.1831, + "predictor_learning_rate": 0.0004995928785458867, + "step": 465 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8961, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.995881321180011e-06, + "loss": 0.7991, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.000499588132118001, + "step": 466 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8959, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.995833582049143e-06, + "loss": 0.896, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004995833582049142, + "step": 467 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8957, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.995785568071522e-06, + "loss": 0.9268, + "output_text_mask_loss": 0.0847, + "predictor_learning_rate": 0.0004995785568071521, + "step": 468 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8955, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.995737279252435e-06, + "loss": 0.7723, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004995737279252434, + "step": 469 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8953, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.9956887155972e-06, + "loss": 0.9, + "output_text_mask_loss": 0.1045, + "predictor_learning_rate": 0.0004995688715597199, + "step": 470 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.895, + "image_masks_loss": 0.0479, + "llm_learning_rate": 4.995639877111164e-06, + "loss": 0.8341, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004995639877111164, + "step": 471 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8948, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.995590763799707e-06, + "loss": 0.8564, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0004995590763799707, + "step": 472 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8946, + "image_masks_loss": 0.05, + "llm_learning_rate": 4.995541375668236e-06, + "loss": 0.8782, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.0004995541375668235, + "step": 473 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8944, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.99549171272219e-06, + "loss": 0.8456, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000499549171272219, + "step": 474 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8942, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.995441774967038e-06, + "loss": 0.8121, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004995441774967037, + "step": 475 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.894, + "image_masks_loss": 0.05, + "llm_learning_rate": 4.995391562408279e-06, + "loss": 0.732, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004995391562408279, + "step": 476 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8938, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.995341075051444e-06, + "loss": 0.8189, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.0004995341075051444, + "step": 477 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8936, + "image_masks_loss": 0.0387, + "llm_learning_rate": 4.995290312902092e-06, + "loss": 0.8401, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.0004995290312902091, + "step": 478 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8934, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.995239275965811e-06, + "loss": 0.8491, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000499523927596581, + "step": 479 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8931, + "image_masks_loss": 0.1175, + "llm_learning_rate": 4.995187964248224e-06, + "loss": 0.9773, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.0004995187964248224, + "step": 480 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8929, + "image_masks_loss": 0.0347, + "llm_learning_rate": 4.99513637775498e-06, + "loss": 0.906, + "output_text_mask_loss": 0.0744, + "predictor_learning_rate": 0.0004995136377754979, + "step": 481 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8927, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.99508451649176e-06, + "loss": 0.8441, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.000499508451649176, + "step": 482 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8925, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.995032380464276e-06, + "loss": 0.8526, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.0004995032380464276, + "step": 483 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8923, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.994979969678269e-06, + "loss": 0.8167, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.000499497996967827, + "step": 484 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8921, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.994927284139511e-06, + "loss": 0.9307, + "output_text_mask_loss": 0.1434, + "predictor_learning_rate": 0.000499492728413951, + "step": 485 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8919, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.994874323853802e-06, + "loss": 0.8253, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004994874323853802, + "step": 486 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8917, + "image_masks_loss": 0.0391, + "llm_learning_rate": 4.994821088826976e-06, + "loss": 0.7826, + "output_text_mask_loss": 0.1083, + "predictor_learning_rate": 0.0004994821088826977, + "step": 487 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8915, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.994767579064895e-06, + "loss": 0.9418, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004994767579064894, + "step": 488 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8913, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.994713794573451e-06, + "loss": 0.8053, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000499471379457345, + "step": 489 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.891, + "image_masks_loss": 0.0067, + "llm_learning_rate": 4.994659735358567e-06, + "loss": 0.8953, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004994659735358567, + "step": 490 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8908, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.9946054014261965e-06, + "loss": 0.8135, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0004994605401426196, + "step": 491 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8906, + "image_masks_loss": 0.0515, + "llm_learning_rate": 4.994550792782322e-06, + "loss": 0.7509, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.0004994550792782322, + "step": 492 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8904, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.994495909432959e-06, + "loss": 0.7996, + "output_text_mask_loss": 0.0938, + "predictor_learning_rate": 0.0004994495909432958, + "step": 493 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8902, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.994440751384149e-06, + "loss": 0.7926, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.0004994440751384148, + "step": 494 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.89, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.994385318641968e-06, + "loss": 0.9132, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004994385318641968, + "step": 495 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8898, + "image_masks_loss": 0.0223, + "llm_learning_rate": 4.9943296112125194e-06, + "loss": 0.7197, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004994329611212519, + "step": 496 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8896, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.994273629101939e-06, + "loss": 0.8335, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004994273629101939, + "step": 497 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8894, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.99421737231639e-06, + "loss": 0.8688, + "output_text_mask_loss": 0.1755, + "predictor_learning_rate": 0.000499421737231639, + "step": 498 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8892, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.994160840862068e-06, + "loss": 0.7343, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004994160840862068, + "step": 499 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8889, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.9941040347451995e-06, + "loss": 0.8898, + "output_text_mask_loss": 0.0858, + "predictor_learning_rate": 0.0004994104034745199, + "step": 500 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8887, + "image_masks_loss": 0.0402, + "llm_learning_rate": 4.994046953972038e-06, + "loss": 0.8379, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0004994046953972037, + "step": 501 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8885, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.993989598548872e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.0383, + "predictor_learning_rate": 0.0004993989598548872, + "step": 502 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8883, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.993931968482016e-06, + "loss": 0.785, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0004993931968482015, + "step": 503 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8881, + "image_masks_loss": 0.0751, + "llm_learning_rate": 4.993874063777816e-06, + "loss": 0.8759, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004993874063777816, + "step": 504 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8879, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.99381588444265e-06, + "loss": 0.8498, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0004993815884442649, + "step": 505 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8877, + "image_masks_loss": 0.0393, + "llm_learning_rate": 4.9937574304829234e-06, + "loss": 0.8549, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0004993757430482923, + "step": 506 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8875, + "image_masks_loss": 0.0877, + "llm_learning_rate": 4.993698701905075e-06, + "loss": 0.8481, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004993698701905074, + "step": 507 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8873, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.993639698715569e-06, + "loss": 0.8164, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.0004993639698715568, + "step": 508 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8871, + "image_masks_loss": 0.0801, + "llm_learning_rate": 4.9935804209209066e-06, + "loss": 0.8121, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004993580420920906, + "step": 509 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8868, + "image_masks_loss": 0.0523, + "llm_learning_rate": 4.993520868527614e-06, + "loss": 0.815, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004993520868527613, + "step": 510 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8866, + "image_masks_loss": 0.0511, + "llm_learning_rate": 4.993461041542248e-06, + "loss": 0.7546, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004993461041542248, + "step": 511 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8864, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.9934009399714e-06, + "loss": 0.8688, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.00049934009399714, + "step": 512 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8862, + "image_masks_loss": 0.0957, + "llm_learning_rate": 4.993340563821685e-06, + "loss": 0.8074, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004993340563821684, + "step": 513 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.886, + "image_masks_loss": 0.087, + "llm_learning_rate": 4.993279913099755e-06, + "loss": 0.9126, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004993279913099754, + "step": 514 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8858, + "image_masks_loss": 0.0215, + "llm_learning_rate": 4.993218987812286e-06, + "loss": 0.9533, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0004993218987812285, + "step": 515 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8856, + "image_masks_loss": 0.0393, + "llm_learning_rate": 4.993157787965989e-06, + "loss": 0.8763, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.0004993157787965989, + "step": 516 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8854, + "image_masks_loss": 0.1045, + "llm_learning_rate": 4.993096313567603e-06, + "loss": 0.867, + "output_text_mask_loss": 0.0992, + "predictor_learning_rate": 0.0004993096313567602, + "step": 517 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8852, + "image_masks_loss": 0.0402, + "llm_learning_rate": 4.993034564623898e-06, + "loss": 0.8381, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004993034564623898, + "step": 518 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.885, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.9929725411416744e-06, + "loss": 0.7604, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004992972541141674, + "step": 519 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8848, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.992910243127761e-06, + "loss": 0.8121, + "output_text_mask_loss": 0.1083, + "predictor_learning_rate": 0.0004992910243127761, + "step": 520 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8845, + "image_masks_loss": 0.1381, + "llm_learning_rate": 4.9928476705890184e-06, + "loss": 1.1425, + "output_text_mask_loss": 0.119, + "predictor_learning_rate": 0.0004992847670589018, + "step": 521 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8843, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.992784823532339e-06, + "loss": 1.3797, + "output_text_mask_loss": 0.4395, + "predictor_learning_rate": 0.0004992784823532338, + "step": 522 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8841, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.992721701964641e-06, + "loss": 1.1055, + "output_text_mask_loss": 0.4578, + "predictor_learning_rate": 0.0004992721701964641, + "step": 523 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8839, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.992658305892878e-06, + "loss": 1.8041, + "output_text_mask_loss": 2.1973, + "predictor_learning_rate": 0.0004992658305892877, + "step": 524 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8837, + "image_masks_loss": 0.0486, + "llm_learning_rate": 4.9925946353240295e-06, + "loss": 2.3226, + "output_text_mask_loss": 2.2583, + "predictor_learning_rate": 0.0004992594635324029, + "step": 525 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8835, + "image_masks_loss": 0.0383, + "llm_learning_rate": 4.992530690265107e-06, + "loss": 1.7845, + "output_text_mask_loss": 2.3804, + "predictor_learning_rate": 0.0004992530690265107, + "step": 526 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8833, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.992466470723154e-06, + "loss": 1.7782, + "output_text_mask_loss": 0.7812, + "predictor_learning_rate": 0.0004992466470723154, + "step": 527 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8831, + "image_masks_loss": 0.0492, + "llm_learning_rate": 4.9924019767052416e-06, + "loss": 2.1135, + "output_text_mask_loss": 0.6195, + "predictor_learning_rate": 0.0004992401976705241, + "step": 528 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8829, + "image_masks_loss": 0.0671, + "llm_learning_rate": 4.992337208218472e-06, + "loss": 2.2303, + "output_text_mask_loss": 2.5879, + "predictor_learning_rate": 0.0004992337208218472, + "step": 529 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8827, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.992272165269977e-06, + "loss": 2.2591, + "output_text_mask_loss": 0.4028, + "predictor_learning_rate": 0.0004992272165269977, + "step": 530 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8825, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.992206847866921e-06, + "loss": 1.4675, + "output_text_mask_loss": 0.2747, + "predictor_learning_rate": 0.0004992206847866921, + "step": 531 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8823, + "image_masks_loss": 0.0215, + "llm_learning_rate": 4.992141256016494e-06, + "loss": 1.2956, + "output_text_mask_loss": 0.6165, + "predictor_learning_rate": 0.0004992141256016494, + "step": 532 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.882, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.992075389725922e-06, + "loss": 1.0866, + "output_text_mask_loss": 0.264, + "predictor_learning_rate": 0.0004992075389725922, + "step": 533 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8818, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.992009249002456e-06, + "loss": 1.0492, + "output_text_mask_loss": 0.3113, + "predictor_learning_rate": 0.0004992009249002456, + "step": 534 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8816, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.991942833853383e-06, + "loss": 0.6835, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004991942833853382, + "step": 535 + }, + { + "epoch": 0.05, + "gumbel_tau": 0.8814, + "image_masks_loss": 0.0341, + "llm_learning_rate": 4.991876144286013e-06, + "loss": 0.7909, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0004991876144286012, + "step": 536 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8812, + "image_masks_loss": 0.045, + "llm_learning_rate": 4.9918091803076916e-06, + "loss": 0.8748, + "output_text_mask_loss": 0.069, + "predictor_learning_rate": 0.0004991809180307691, + "step": 537 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.881, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.991741941925793e-06, + "loss": 0.9941, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004991741941925792, + "step": 538 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8808, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.991674429147721e-06, + "loss": 1.0253, + "output_text_mask_loss": 0.19, + "predictor_learning_rate": 0.0004991674429147721, + "step": 539 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8806, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.991606641980912e-06, + "loss": 1.0113, + "output_text_mask_loss": 0.1694, + "predictor_learning_rate": 0.0004991606641980912, + "step": 540 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8804, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.99153858043283e-06, + "loss": 0.679, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004991538580432829, + "step": 541 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8802, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.991470244510969e-06, + "loss": 0.9802, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004991470244510969, + "step": 542 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.88, + "image_masks_loss": 0.0549, + "llm_learning_rate": 4.991401634222855e-06, + "loss": 0.8949, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0004991401634222855, + "step": 543 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8798, + "image_masks_loss": 0.0534, + "llm_learning_rate": 4.991332749576044e-06, + "loss": 0.8833, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004991332749576044, + "step": 544 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8796, + "image_masks_loss": 0.0439, + "llm_learning_rate": 4.991263590578121e-06, + "loss": 0.8497, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0004991263590578121, + "step": 545 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8793, + "image_masks_loss": 0.0702, + "llm_learning_rate": 4.991194157236704e-06, + "loss": 0.9433, + "output_text_mask_loss": 0.0152, + "predictor_learning_rate": 0.0004991194157236703, + "step": 546 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8791, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.991124449559436e-06, + "loss": 0.8893, + "output_text_mask_loss": 0.066, + "predictor_learning_rate": 0.0004991124449559435, + "step": 547 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8789, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.991054467553995e-06, + "loss": 0.7129, + "output_text_mask_loss": 0.1114, + "predictor_learning_rate": 0.0004991054467553995, + "step": 548 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8787, + "image_masks_loss": 0.0243, + "llm_learning_rate": 4.990984211228087e-06, + "loss": 0.7975, + "output_text_mask_loss": 0.106, + "predictor_learning_rate": 0.0004990984211228087, + "step": 549 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8785, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.99091368058945e-06, + "loss": 0.8987, + "output_text_mask_loss": 0.0923, + "predictor_learning_rate": 0.000499091368058945, + "step": 550 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8783, + "image_masks_loss": 0.0483, + "llm_learning_rate": 4.99084287564585e-06, + "loss": 0.834, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.000499084287564585, + "step": 551 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8781, + "image_masks_loss": 0.0029, + "llm_learning_rate": 4.9907717964050845e-06, + "loss": 0.8924, + "output_text_mask_loss": 0.1831, + "predictor_learning_rate": 0.0004990771796405084, + "step": 552 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8779, + "image_masks_loss": 0.0182, + "llm_learning_rate": 4.990700442874981e-06, + "loss": 1.0774, + "output_text_mask_loss": 0.2625, + "predictor_learning_rate": 0.000499070044287498, + "step": 553 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8777, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.990628815063396e-06, + "loss": 0.9317, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.0004990628815063396, + "step": 554 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8775, + "image_masks_loss": 0.0347, + "llm_learning_rate": 4.9905569129782185e-06, + "loss": 1.0035, + "output_text_mask_loss": 0.2487, + "predictor_learning_rate": 0.0004990556912978218, + "step": 555 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8773, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.990484736627366e-06, + "loss": 0.9649, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.0004990484736627366, + "step": 556 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8771, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.9904122860187885e-06, + "loss": 0.7811, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004990412286018788, + "step": 557 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8769, + "image_masks_loss": 0.0242, + "llm_learning_rate": 4.990339561160461e-06, + "loss": 0.9114, + "output_text_mask_loss": 0.1198, + "predictor_learning_rate": 0.0004990339561160461, + "step": 558 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8767, + "image_masks_loss": 0.0189, + "llm_learning_rate": 4.990266562060395e-06, + "loss": 0.8579, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004990266562060394, + "step": 559 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8764, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.990193288726629e-06, + "loss": 0.8904, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004990193288726629, + "step": 560 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8762, + "image_masks_loss": 0.0378, + "llm_learning_rate": 4.99011974116723e-06, + "loss": 0.9644, + "output_text_mask_loss": 0.119, + "predictor_learning_rate": 0.000499011974116723, + "step": 561 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.876, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.990045919390299e-06, + "loss": 0.8718, + "output_text_mask_loss": 0.1266, + "predictor_learning_rate": 0.00049900459193903, + "step": 562 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8758, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.989971823403966e-06, + "loss": 0.9169, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 0.0004989971823403965, + "step": 563 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8756, + "image_masks_loss": 0.0336, + "llm_learning_rate": 4.989897453216388e-06, + "loss": 0.8, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004989897453216389, + "step": 564 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8754, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.989822808835758e-06, + "loss": 0.7949, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.0004989822808835758, + "step": 565 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8752, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.989747890270293e-06, + "loss": 0.8378, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.0004989747890270293, + "step": 566 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.875, + "image_masks_loss": 0.045, + "llm_learning_rate": 4.9896726975282455e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004989672697528246, + "step": 567 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8748, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.989597230617894e-06, + "loss": 0.9649, + "output_text_mask_loss": 0.0196, + "predictor_learning_rate": 0.0004989597230617895, + "step": 568 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8746, + "image_masks_loss": 0.0626, + "llm_learning_rate": 4.9895214895475515e-06, + "loss": 0.8688, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.0004989521489547552, + "step": 569 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8744, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.9894454743255576e-06, + "loss": 0.8803, + "output_text_mask_loss": 0.0862, + "predictor_learning_rate": 0.0004989445474325556, + "step": 570 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8742, + "image_masks_loss": 0.0629, + "llm_learning_rate": 4.989369184960281e-06, + "loss": 0.9166, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.0004989369184960281, + "step": 571 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.874, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.989292621460127e-06, + "loss": 0.8411, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004989292621460126, + "step": 572 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8738, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.989215783833523e-06, + "loss": 0.9005, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.0004989215783833523, + "step": 573 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8736, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.989138672088933e-06, + "loss": 0.9159, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004989138672088933, + "step": 574 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8734, + "image_masks_loss": 0.0359, + "llm_learning_rate": 4.989061286234849e-06, + "loss": 0.8137, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.0004989061286234849, + "step": 575 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8731, + "image_masks_loss": 0.0496, + "llm_learning_rate": 4.988983626279792e-06, + "loss": 0.8139, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004988983626279791, + "step": 576 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8729, + "image_masks_loss": 0.036, + "llm_learning_rate": 4.988905692232313e-06, + "loss": 0.916, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.0004988905692232313, + "step": 577 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8727, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.988827484100996e-06, + "loss": 0.9721, + "output_text_mask_loss": 0.1068, + "predictor_learning_rate": 0.0004988827484100995, + "step": 578 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8725, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.988749001894452e-06, + "loss": 0.8103, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004988749001894452, + "step": 579 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8723, + "image_masks_loss": 0.0343, + "llm_learning_rate": 4.9886702456213255e-06, + "loss": 0.8505, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004988670245621326, + "step": 580 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8721, + "image_masks_loss": 0.0467, + "llm_learning_rate": 4.9885912152902875e-06, + "loss": 0.8699, + "output_text_mask_loss": 0.119, + "predictor_learning_rate": 0.0004988591215290287, + "step": 581 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8719, + "image_masks_loss": 0.0572, + "llm_learning_rate": 4.988511910910042e-06, + "loss": 0.9041, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004988511910910042, + "step": 582 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8717, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.988432332489323e-06, + "loss": 0.8706, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004988432332489322, + "step": 583 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8715, + "image_masks_loss": 0.0526, + "llm_learning_rate": 4.988352480036891e-06, + "loss": 0.7975, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004988352480036891, + "step": 584 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8713, + "image_masks_loss": 0.0641, + "llm_learning_rate": 4.9882723535615415e-06, + "loss": 0.738, + "output_text_mask_loss": 0.0969, + "predictor_learning_rate": 0.0004988272353561541, + "step": 585 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8711, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.988191953072099e-06, + "loss": 0.9265, + "output_text_mask_loss": 0.1678, + "predictor_learning_rate": 0.0004988191953072099, + "step": 586 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8709, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.988111278577417e-06, + "loss": 0.9041, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004988111278577416, + "step": 587 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8707, + "image_masks_loss": 0.0664, + "llm_learning_rate": 4.988030330086377e-06, + "loss": 0.8522, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.0004988030330086377, + "step": 588 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8705, + "image_masks_loss": 0.0412, + "llm_learning_rate": 4.9879491076078966e-06, + "loss": 0.8031, + "output_text_mask_loss": 0.0923, + "predictor_learning_rate": 0.0004987949107607897, + "step": 589 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8703, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.987867611150918e-06, + "loss": 0.816, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.0004987867611150918, + "step": 590 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8701, + "image_masks_loss": 0.0443, + "llm_learning_rate": 4.9877858407244175e-06, + "loss": 0.8146, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.0004987785840724417, + "step": 591 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8699, + "image_masks_loss": 0.0414, + "llm_learning_rate": 4.987703796337397e-06, + "loss": 0.901, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.0004987703796337398, + "step": 592 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8696, + "image_masks_loss": 0.0736, + "llm_learning_rate": 4.987621477998895e-06, + "loss": 0.8412, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004987621477998894, + "step": 593 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8694, + "image_masks_loss": 0.0343, + "llm_learning_rate": 4.987538885717974e-06, + "loss": 0.8518, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0004987538885717974, + "step": 594 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8692, + "image_masks_loss": 0.174, + "llm_learning_rate": 4.987456019503731e-06, + "loss": 0.8985, + "output_text_mask_loss": 0.0908, + "predictor_learning_rate": 0.0004987456019503731, + "step": 595 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.869, + "image_masks_loss": 0.1381, + "llm_learning_rate": 4.98737287936529e-06, + "loss": 0.978, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.000498737287936529, + "step": 596 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8688, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.987289465311808e-06, + "loss": 0.8013, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004987289465311807, + "step": 597 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8686, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.987205777352469e-06, + "loss": 0.7583, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004987205777352468, + "step": 598 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8684, + "image_masks_loss": 0.0427, + "llm_learning_rate": 4.987121815496489e-06, + "loss": 0.8099, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004987121815496489, + "step": 599 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8682, + "image_masks_loss": 0.0633, + "llm_learning_rate": 4.987037579753116e-06, + "loss": 0.7071, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004987037579753116, + "step": 600 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.868, + "image_masks_loss": 0.0433, + "llm_learning_rate": 4.986953070131625e-06, + "loss": 0.8427, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004986953070131625, + "step": 601 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8678, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.986868286641322e-06, + "loss": 0.9268, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0004986868286641321, + "step": 602 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8676, + "image_masks_loss": 0.0496, + "llm_learning_rate": 4.986783229291544e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.1312, + "predictor_learning_rate": 0.0004986783229291543, + "step": 603 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8674, + "image_masks_loss": 0.0347, + "llm_learning_rate": 4.986697898091658e-06, + "loss": 0.8166, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004986697898091657, + "step": 604 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8672, + "image_masks_loss": 0.0881, + "llm_learning_rate": 4.986612293051061e-06, + "loss": 0.9111, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.000498661229305106, + "step": 605 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.867, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.9865264141791796e-06, + "loss": 0.7419, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004986526414179178, + "step": 606 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8668, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.98644026148547e-06, + "loss": 0.8371, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004986440261485469, + "step": 607 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8666, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.98635383497942e-06, + "loss": 0.9429, + "output_text_mask_loss": 0.1099, + "predictor_learning_rate": 0.000498635383497942, + "step": 608 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8664, + "image_masks_loss": 0.0404, + "llm_learning_rate": 4.986267134670549e-06, + "loss": 0.8454, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004986267134670548, + "step": 609 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8662, + "image_masks_loss": 0.0381, + "llm_learning_rate": 4.986180160568402e-06, + "loss": 0.8906, + "output_text_mask_loss": 0.0916, + "predictor_learning_rate": 0.0004986180160568403, + "step": 610 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.866, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.986092912682559e-06, + "loss": 0.8199, + "output_text_mask_loss": 0.1076, + "predictor_learning_rate": 0.0004986092912682558, + "step": 611 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8658, + "image_masks_loss": 0.074, + "llm_learning_rate": 4.986005391022627e-06, + "loss": 0.9006, + "output_text_mask_loss": 0.1701, + "predictor_learning_rate": 0.0004986005391022626, + "step": 612 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8656, + "image_masks_loss": 0.1038, + "llm_learning_rate": 4.985917595598243e-06, + "loss": 0.8277, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004985917595598243, + "step": 613 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8654, + "image_masks_loss": 0.053, + "llm_learning_rate": 4.985829526419077e-06, + "loss": 0.8724, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004985829526419077, + "step": 614 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8651, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.985741183494825e-06, + "loss": 0.8088, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004985741183494826, + "step": 615 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8649, + "image_masks_loss": 0.0422, + "llm_learning_rate": 4.985652566835219e-06, + "loss": 0.9159, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0004985652566835218, + "step": 616 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8647, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.985563676450015e-06, + "loss": 0.7422, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004985563676450014, + "step": 617 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8645, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.985474512349003e-06, + "loss": 0.8995, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004985474512349003, + "step": 618 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8643, + "image_masks_loss": 0.0427, + "llm_learning_rate": 4.985385074542001e-06, + "loss": 0.8788, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004985385074542001, + "step": 619 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8641, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.985295363038859e-06, + "loss": 0.8863, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004985295363038859, + "step": 620 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8639, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.985205377849456e-06, + "loss": 0.7796, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0004985205377849455, + "step": 621 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8637, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.9851151189837e-06, + "loss": 0.853, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0004985115118983699, + "step": 622 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8635, + "image_masks_loss": 0.0446, + "llm_learning_rate": 4.985024586451533e-06, + "loss": 0.8473, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0004985024586451533, + "step": 623 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8633, + "image_masks_loss": 0.0441, + "llm_learning_rate": 4.984933780262923e-06, + "loss": 0.7903, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004984933780262923, + "step": 624 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8631, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.98484270042787e-06, + "loss": 0.8028, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000498484270042787, + "step": 625 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8629, + "image_masks_loss": 0.0492, + "llm_learning_rate": 4.984751346956404e-06, + "loss": 0.9066, + "output_text_mask_loss": 0.1579, + "predictor_learning_rate": 0.0004984751346956403, + "step": 626 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8627, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.984659719858586e-06, + "loss": 0.6869, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.0004984659719858586, + "step": 627 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8625, + "image_masks_loss": 0.0412, + "llm_learning_rate": 4.984567819144504e-06, + "loss": 0.9286, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004984567819144503, + "step": 628 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8623, + "image_masks_loss": 0.0084, + "llm_learning_rate": 4.98447564482428e-06, + "loss": 0.8614, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000498447564482428, + "step": 629 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8621, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.984383196908064e-06, + "loss": 0.9707, + "output_text_mask_loss": 0.1099, + "predictor_learning_rate": 0.0004984383196908064, + "step": 630 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8619, + "image_masks_loss": 0.0204, + "llm_learning_rate": 4.984290475406037e-06, + "loss": 0.8409, + "output_text_mask_loss": 0.0759, + "predictor_learning_rate": 0.0004984290475406037, + "step": 631 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8617, + "image_masks_loss": 0.0152, + "llm_learning_rate": 4.984197480328409e-06, + "loss": 0.8373, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004984197480328408, + "step": 632 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8615, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.984104211685421e-06, + "loss": 0.8505, + "output_text_mask_loss": 0.087, + "predictor_learning_rate": 0.0004984104211685421, + "step": 633 + }, + { + "epoch": 0.06, + "gumbel_tau": 0.8613, + "image_masks_loss": 0.0341, + "llm_learning_rate": 4.984010669487344e-06, + "loss": 0.8392, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004984010669487344, + "step": 634 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8611, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.98391685374448e-06, + "loss": 0.7651, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.000498391685374448, + "step": 635 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8609, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.983822764467159e-06, + "loss": 0.8246, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0004983822764467159, + "step": 636 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8607, + "image_masks_loss": 0.0492, + "llm_learning_rate": 4.983728401665743e-06, + "loss": 0.8266, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004983728401665743, + "step": 637 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8605, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.983633765350622e-06, + "loss": 0.7148, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.0004983633765350622, + "step": 638 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8603, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.9835388555322194e-06, + "loss": 0.7729, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004983538855532219, + "step": 639 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8601, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.983443672220986e-06, + "loss": 0.889, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004983443672220986, + "step": 640 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8599, + "image_masks_loss": 0.0368, + "llm_learning_rate": 4.983348215427404e-06, + "loss": 0.7893, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.0004983348215427403, + "step": 641 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8597, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.983252485161985e-06, + "loss": 0.8578, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.0004983252485161984, + "step": 642 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8595, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.983156481435271e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.0004983156481435271, + "step": 643 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8592, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.983060204257834e-06, + "loss": 0.8747, + "output_text_mask_loss": 0.079, + "predictor_learning_rate": 0.0004983060204257834, + "step": 644 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.859, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.982963653640277e-06, + "loss": 0.7743, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004982963653640276, + "step": 645 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8588, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.982866829593231e-06, + "loss": 0.8906, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004982866829593231, + "step": 646 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8586, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.982769732127361e-06, + "loss": 0.786, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.0004982769732127361, + "step": 647 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8584, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.982672361253357e-06, + "loss": 0.8656, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004982672361253357, + "step": 648 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8582, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.982574716981942e-06, + "loss": 0.7749, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0004982574716981942, + "step": 649 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.858, + "image_masks_loss": 0.0211, + "llm_learning_rate": 4.98247679932387e-06, + "loss": 0.8475, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.0004982476799323869, + "step": 650 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8578, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.9823786082899235e-06, + "loss": 0.8767, + "output_text_mask_loss": 0.0471, + "predictor_learning_rate": 0.0004982378608289924, + "step": 651 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8576, + "image_masks_loss": 0.1022, + "llm_learning_rate": 4.982280143890915e-06, + "loss": 0.8252, + "output_text_mask_loss": 0.0207, + "predictor_learning_rate": 0.0004982280143890915, + "step": 652 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8574, + "image_masks_loss": 0.0546, + "llm_learning_rate": 4.982181406137689e-06, + "loss": 0.8959, + "output_text_mask_loss": 0.0946, + "predictor_learning_rate": 0.0004982181406137688, + "step": 653 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8572, + "image_masks_loss": 0.0431, + "llm_learning_rate": 4.9820823950411166e-06, + "loss": 0.8044, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004982082395041116, + "step": 654 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.857, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.981983110612103e-06, + "loss": 0.8164, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.0004981983110612103, + "step": 655 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8568, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.98188355286158e-06, + "loss": 0.8998, + "output_text_mask_loss": 0.1465, + "predictor_learning_rate": 0.000498188355286158, + "step": 656 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8566, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.981783721800513e-06, + "loss": 0.7299, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0004981783721800513, + "step": 657 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8564, + "image_masks_loss": 0.0307, + "llm_learning_rate": 4.981683617439893e-06, + "loss": 0.8503, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.0004981683617439894, + "step": 658 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8562, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.981583239790747e-06, + "loss": 0.7887, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0004981583239790746, + "step": 659 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.856, + "image_masks_loss": 0.0063, + "llm_learning_rate": 4.9814825888641265e-06, + "loss": 0.7919, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.0004981482588864126, + "step": 660 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8558, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.981381664671116e-06, + "loss": 0.8192, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004981381664671115, + "step": 661 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8556, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.98128046722283e-06, + "loss": 0.7794, + "output_text_mask_loss": 0.1015, + "predictor_learning_rate": 0.000498128046722283, + "step": 662 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8554, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.9811789965304115e-06, + "loss": 0.8087, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004981178996530412, + "step": 663 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8552, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.981077252605035e-06, + "loss": 0.7688, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0004981077252605035, + "step": 664 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.855, + "image_masks_loss": 0.0071, + "llm_learning_rate": 4.9809752354579055e-06, + "loss": 0.7547, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.0004980975235457906, + "step": 665 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8548, + "image_masks_loss": 0.0057, + "llm_learning_rate": 4.980872945100257e-06, + "loss": 0.747, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004980872945100257, + "step": 666 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8546, + "image_masks_loss": 0.0433, + "llm_learning_rate": 4.980770381543354e-06, + "loss": 0.916, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004980770381543354, + "step": 667 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8544, + "image_masks_loss": 0.0193, + "llm_learning_rate": 4.980667544798491e-06, + "loss": 0.7805, + "output_text_mask_loss": 0.0736, + "predictor_learning_rate": 0.0004980667544798491, + "step": 668 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8542, + "image_masks_loss": 0.0181, + "llm_learning_rate": 4.980564434876991e-06, + "loss": 0.8095, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 0.0004980564434876991, + "step": 669 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.854, + "image_masks_loss": 0.046, + "llm_learning_rate": 4.980461051790212e-06, + "loss": 0.8448, + "output_text_mask_loss": 0.0334, + "predictor_learning_rate": 0.0004980461051790211, + "step": 670 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8538, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.980357395549537e-06, + "loss": 0.859, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004980357395549536, + "step": 671 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8536, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.980253466166379e-06, + "loss": 0.7933, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004980253466166379, + "step": 672 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8534, + "image_masks_loss": 0.036, + "llm_learning_rate": 4.980149263652186e-06, + "loss": 0.7862, + "output_text_mask_loss": 0.0128, + "predictor_learning_rate": 0.0004980149263652187, + "step": 673 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8532, + "image_masks_loss": 0.0401, + "llm_learning_rate": 4.980044788018432e-06, + "loss": 0.8582, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004980044788018431, + "step": 674 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.853, + "image_masks_loss": 0.0462, + "llm_learning_rate": 4.979940039276622e-06, + "loss": 0.6898, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004979940039276621, + "step": 675 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8528, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.97983501743829e-06, + "loss": 0.7758, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.000497983501743829, + "step": 676 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8526, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.979729722515003e-06, + "loss": 0.6818, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004979729722515003, + "step": 677 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8524, + "image_masks_loss": 0.0549, + "llm_learning_rate": 4.979624154518355e-06, + "loss": 0.915, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.0004979624154518355, + "step": 678 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8522, + "image_masks_loss": 0.0397, + "llm_learning_rate": 4.979518313459973e-06, + "loss": 0.7331, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004979518313459972, + "step": 679 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.852, + "image_masks_loss": 0.0195, + "llm_learning_rate": 4.97941219935151e-06, + "loss": 0.705, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.000497941219935151, + "step": 680 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8518, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.979305812204654e-06, + "loss": 0.8212, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.0004979305812204654, + "step": 681 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8516, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.979199152031119e-06, + "loss": 0.8705, + "output_text_mask_loss": 0.0694, + "predictor_learning_rate": 0.000497919915203112, + "step": 682 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8514, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.979092218842652e-06, + "loss": 0.8584, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004979092218842651, + "step": 683 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8512, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.978985012651027e-06, + "loss": 0.8464, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004978985012651028, + "step": 684 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.851, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.978877533468051e-06, + "loss": 0.87, + "output_text_mask_loss": 0.0212, + "predictor_learning_rate": 0.0004978877533468051, + "step": 685 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8508, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.9787697813055596e-06, + "loss": 0.8831, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004978769781305559, + "step": 686 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8506, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.978661756175419e-06, + "loss": 0.8447, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004978661756175418, + "step": 687 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8504, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.978553458089524e-06, + "loss": 0.7843, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004978553458089524, + "step": 688 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8502, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.978444887059802e-06, + "loss": 0.7417, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0004978444887059802, + "step": 689 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.85, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.978336043098208e-06, + "loss": 0.8235, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.0004978336043098208, + "step": 690 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8498, + "image_masks_loss": 0.0328, + "llm_learning_rate": 4.978226926216729e-06, + "loss": 0.7612, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004978226926216729, + "step": 691 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8496, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.9781175364273805e-06, + "loss": 0.7779, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004978117536427381, + "step": 692 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8494, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.97800787374221e-06, + "loss": 0.7398, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.000497800787374221, + "step": 693 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8492, + "image_masks_loss": 0.0322, + "llm_learning_rate": 4.977897938173292e-06, + "loss": 0.7993, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004977897938173291, + "step": 694 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.849, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.977787729732734e-06, + "loss": 0.8604, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004977787729732734, + "step": 695 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8488, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.977677248432672e-06, + "loss": 0.7878, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004977677248432672, + "step": 696 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8486, + "image_masks_loss": 0.0549, + "llm_learning_rate": 4.977566494285273e-06, + "loss": 0.8279, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004977566494285273, + "step": 697 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8484, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.9774554673027335e-06, + "loss": 0.8414, + "output_text_mask_loss": 0.1099, + "predictor_learning_rate": 0.0004977455467302733, + "step": 698 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8482, + "image_masks_loss": 0.0393, + "llm_learning_rate": 4.977344167497279e-06, + "loss": 0.9254, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0004977344167497278, + "step": 699 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.848, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.9772325948811665e-06, + "loss": 0.802, + "output_text_mask_loss": 0.0477, + "predictor_learning_rate": 0.0004977232594881166, + "step": 700 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8478, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.977120749466683e-06, + "loss": 0.9364, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004977120749466682, + "step": 701 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8476, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.977008631266145e-06, + "loss": 0.8238, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004977008631266145, + "step": 702 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8474, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.976896240291899e-06, + "loss": 0.8254, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004976896240291899, + "step": 703 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8472, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.976783576556322e-06, + "loss": 0.8344, + "output_text_mask_loss": 0.0877, + "predictor_learning_rate": 0.0004976783576556322, + "step": 704 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.847, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.976670640071821e-06, + "loss": 0.7857, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004976670640071821, + "step": 705 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8468, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.976557430850833e-06, + "loss": 0.7079, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004976557430850832, + "step": 706 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8466, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.976443948905824e-06, + "loss": 0.7234, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.0004976443948905824, + "step": 707 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8464, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.976330194249291e-06, + "loss": 0.6684, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.000497633019424929, + "step": 708 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8462, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.9762161668937615e-06, + "loss": 0.7176, + "output_text_mask_loss": 0.0043, + "predictor_learning_rate": 0.0004976216166893761, + "step": 709 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.846, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.976101866851792e-06, + "loss": 0.7795, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004976101866851791, + "step": 710 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8458, + "image_masks_loss": 0.0207, + "llm_learning_rate": 4.975987294135969e-06, + "loss": 0.8465, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0004975987294135969, + "step": 711 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8456, + "image_masks_loss": 0.0212, + "llm_learning_rate": 4.975872448758911e-06, + "loss": 0.7697, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0004975872448758911, + "step": 712 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8454, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.9757573307332644e-06, + "loss": 0.7049, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004975757330733264, + "step": 713 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8452, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.975641940071705e-06, + "loss": 0.868, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 0.0004975641940071705, + "step": 714 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.845, + "image_masks_loss": 0.0158, + "llm_learning_rate": 4.975526276786942e-06, + "loss": 0.7993, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004975526276786941, + "step": 715 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8448, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.97541034089171e-06, + "loss": 0.733, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 0.000497541034089171, + "step": 716 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8446, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.975294132398779e-06, + "loss": 0.7819, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004975294132398778, + "step": 717 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8444, + "image_masks_loss": 0.0027, + "llm_learning_rate": 4.975177651320944e-06, + "loss": 0.7774, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004975177651320943, + "step": 718 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8442, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.975060897671032e-06, + "loss": 0.7434, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004975060897671032, + "step": 719 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.844, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.974943871461902e-06, + "loss": 0.7889, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004974943871461901, + "step": 720 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8438, + "image_masks_loss": 0.0215, + "llm_learning_rate": 4.9748265727064395e-06, + "loss": 0.8041, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000497482657270644, + "step": 721 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8436, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.974709001417563e-06, + "loss": 0.8003, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004974709001417562, + "step": 722 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8434, + "image_masks_loss": 0.0236, + "llm_learning_rate": 4.9745911576082186e-06, + "loss": 0.8224, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004974591157608218, + "step": 723 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8432, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.974473041291385e-06, + "loss": 0.8719, + "output_text_mask_loss": 0.0717, + "predictor_learning_rate": 0.0004974473041291384, + "step": 724 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.843, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.9743546524800665e-06, + "loss": 0.7955, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004974354652480066, + "step": 725 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8428, + "image_masks_loss": 0.0052, + "llm_learning_rate": 4.974235991187304e-06, + "loss": 0.7758, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004974235991187304, + "step": 726 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8426, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.9741170574261625e-06, + "loss": 0.829, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004974117057426162, + "step": 727 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8424, + "image_masks_loss": 0.0215, + "llm_learning_rate": 4.973997851209739e-06, + "loss": 0.7391, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004973997851209738, + "step": 728 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8422, + "image_masks_loss": 0.0507, + "llm_learning_rate": 4.973878372551162e-06, + "loss": 0.898, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.0004973878372551161, + "step": 729 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.842, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.973758621463588e-06, + "loss": 0.7738, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004973758621463588, + "step": 730 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8418, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.973638597960205e-06, + "loss": 0.7549, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0004973638597960205, + "step": 731 + }, + { + "epoch": 0.07, + "gumbel_tau": 0.8416, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.97351830205423e-06, + "loss": 0.8025, + "output_text_mask_loss": 0.0824, + "predictor_learning_rate": 0.000497351830205423, + "step": 732 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8414, + "image_masks_loss": 0.038, + "llm_learning_rate": 4.973397733758909e-06, + "loss": 0.7491, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000497339773375891, + "step": 733 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8412, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.973276893087522e-06, + "loss": 0.7841, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0004973276893087522, + "step": 734 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.841, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.9731557800533734e-06, + "loss": 0.8116, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004973155780053373, + "step": 735 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8408, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.973034394669801e-06, + "loss": 0.8397, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004973034394669801, + "step": 736 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8406, + "image_masks_loss": 0.0355, + "llm_learning_rate": 4.972912736950174e-06, + "loss": 0.8547, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.0004972912736950174, + "step": 737 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8404, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.972790806907887e-06, + "loss": 0.8669, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004972790806907887, + "step": 738 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8402, + "image_masks_loss": 0.0303, + "llm_learning_rate": 4.972668604556368e-06, + "loss": 0.7801, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.0004972668604556369, + "step": 739 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.84, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.972546129909077e-06, + "loss": 0.8469, + "output_text_mask_loss": 0.1122, + "predictor_learning_rate": 0.0004972546129909076, + "step": 740 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8398, + "image_masks_loss": 0.0223, + "llm_learning_rate": 4.9724233829794966e-06, + "loss": 0.7922, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0004972423382979496, + "step": 741 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8396, + "image_masks_loss": 0.038, + "llm_learning_rate": 4.972300363781148e-06, + "loss": 0.8158, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004972300363781147, + "step": 742 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8394, + "image_masks_loss": 0.0519, + "llm_learning_rate": 4.972177072327574e-06, + "loss": 0.8342, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 0.0004972177072327574, + "step": 743 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8392, + "image_masks_loss": 0.0427, + "llm_learning_rate": 4.972053508632356e-06, + "loss": 0.8887, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004972053508632356, + "step": 744 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.839, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.9719296727090986e-06, + "loss": 0.8401, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0004971929672709099, + "step": 745 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8388, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.9718055645714405e-06, + "loss": 0.7448, + "output_text_mask_loss": 0.0778, + "predictor_learning_rate": 0.000497180556457144, + "step": 746 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8386, + "image_masks_loss": 0.0383, + "llm_learning_rate": 4.971681184233047e-06, + "loss": 0.788, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004971681184233047, + "step": 747 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8384, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.971556531707616e-06, + "loss": 0.7757, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004971556531707616, + "step": 748 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8382, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.9714316070088756e-06, + "loss": 0.7744, + "output_text_mask_loss": 0.0214, + "predictor_learning_rate": 0.0004971431607008875, + "step": 749 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.838, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.971306410150581e-06, + "loss": 0.7939, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.000497130641015058, + "step": 750 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8378, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.97118094114652e-06, + "loss": 0.7698, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000497118094114652, + "step": 751 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8376, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.97105520001051e-06, + "loss": 0.7788, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.0004971055200010509, + "step": 752 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8374, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.970929186756396e-06, + "loss": 0.8066, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004970929186756396, + "step": 753 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8372, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.9708029013980565e-06, + "loss": 0.8434, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.0004970802901398056, + "step": 754 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.837, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.9706763439493985e-06, + "loss": 0.8164, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004970676343949398, + "step": 755 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8368, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.970549514424358e-06, + "loss": 0.7068, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004970549514424358, + "step": 756 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8366, + "image_masks_loss": 0.0207, + "llm_learning_rate": 4.970422412836903e-06, + "loss": 0.8491, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004970422412836902, + "step": 757 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8364, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.970295039201029e-06, + "loss": 0.7364, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004970295039201029, + "step": 758 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8362, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.970167393530762e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.0004970167393530761, + "step": 759 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8361, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.970039475840159e-06, + "loss": 0.8758, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004970039475840159, + "step": 760 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8359, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.969911286143309e-06, + "loss": 0.6774, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004969911286143308, + "step": 761 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8357, + "image_masks_loss": 0.0058, + "llm_learning_rate": 4.969782824454325e-06, + "loss": 0.8369, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0004969782824454325, + "step": 762 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8355, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.969654090787356e-06, + "loss": 0.746, + "output_text_mask_loss": 0.0214, + "predictor_learning_rate": 0.0004969654090787355, + "step": 763 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8353, + "image_masks_loss": 0.0349, + "llm_learning_rate": 4.969525085156578e-06, + "loss": 0.7757, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004969525085156578, + "step": 764 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8351, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.969395807576196e-06, + "loss": 0.8237, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004969395807576196, + "step": 765 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8349, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.969266258060448e-06, + "loss": 0.7281, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004969266258060448, + "step": 766 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8347, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.969136436623599e-06, + "loss": 0.8226, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0004969136436623599, + "step": 767 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8345, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.969006343279946e-06, + "loss": 0.7179, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004969006343279946, + "step": 768 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8343, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.968875978043815e-06, + "loss": 0.6959, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0004968875978043815, + "step": 769 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8341, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.968745340929563e-06, + "loss": 0.8561, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 0.0004968745340929563, + "step": 770 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8339, + "image_masks_loss": 0.0057, + "llm_learning_rate": 4.968614431951574e-06, + "loss": 0.6856, + "output_text_mask_loss": 0.0919, + "predictor_learning_rate": 0.0004968614431951574, + "step": 771 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8337, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.968483251124265e-06, + "loss": 0.7013, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004968483251124265, + "step": 772 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8335, + "image_masks_loss": 0.0053, + "llm_learning_rate": 4.968351798462082e-06, + "loss": 0.7933, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004968351798462083, + "step": 773 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8333, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.9682200739795025e-06, + "loss": 0.8579, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0004968220073979502, + "step": 774 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8331, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.968088077691029e-06, + "loss": 0.8151, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004968088077691029, + "step": 775 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8329, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.9679558096112e-06, + "loss": 0.8366, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00049679558096112, + "step": 776 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8327, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.96782326975458e-06, + "loss": 0.8274, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.000496782326975458, + "step": 777 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8325, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.967690458135764e-06, + "loss": 0.7309, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004967690458135764, + "step": 778 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8323, + "image_masks_loss": 0.0687, + "llm_learning_rate": 4.967557374769378e-06, + "loss": 0.8401, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0004967557374769378, + "step": 779 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8321, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.9674240196700785e-06, + "loss": 0.8336, + "output_text_mask_loss": 0.0341, + "predictor_learning_rate": 0.0004967424019670079, + "step": 780 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8319, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.967290392852549e-06, + "loss": 0.8623, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.000496729039285255, + "step": 781 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8317, + "image_masks_loss": 0.0328, + "llm_learning_rate": 4.967156494331507e-06, + "loss": 0.8437, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.0004967156494331507, + "step": 782 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8315, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.967022324121697e-06, + "loss": 0.8236, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004967022324121696, + "step": 783 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8313, + "image_masks_loss": 0.0213, + "llm_learning_rate": 4.9668878822378914e-06, + "loss": 0.7861, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004966887882237891, + "step": 784 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8311, + "image_masks_loss": 0.0213, + "llm_learning_rate": 4.966753168694899e-06, + "loss": 0.7425, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0004966753168694899, + "step": 785 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8309, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.966618183507552e-06, + "loss": 0.8368, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 0.0004966618183507552, + "step": 786 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8307, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.966482926690718e-06, + "loss": 0.8306, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004966482926690717, + "step": 787 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8305, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.966347398259288e-06, + "loss": 0.8152, + "output_text_mask_loss": 0.0839, + "predictor_learning_rate": 0.0004966347398259288, + "step": 788 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8304, + "image_masks_loss": 0.0471, + "llm_learning_rate": 4.9662115982281914e-06, + "loss": 0.8807, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.0004966211598228191, + "step": 789 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8302, + "image_masks_loss": 0.0215, + "llm_learning_rate": 4.966075526612379e-06, + "loss": 0.7774, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.0004966075526612379, + "step": 790 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.83, + "image_masks_loss": 0.0229, + "llm_learning_rate": 4.9659391834268365e-06, + "loss": 0.8541, + "output_text_mask_loss": 0.0916, + "predictor_learning_rate": 0.0004965939183426836, + "step": 791 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8298, + "image_masks_loss": 0.0298, + "llm_learning_rate": 4.9658025686865795e-06, + "loss": 0.832, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004965802568686579, + "step": 792 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8296, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.96566568240665e-06, + "loss": 0.8092, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000496566568240665, + "step": 793 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8294, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.965528524602124e-06, + "loss": 0.7413, + "output_text_mask_loss": 0.0954, + "predictor_learning_rate": 0.0004965528524602124, + "step": 794 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8292, + "image_masks_loss": 0.0357, + "llm_learning_rate": 4.965391095288106e-06, + "loss": 0.7795, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004965391095288106, + "step": 795 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.829, + "image_masks_loss": 0.0431, + "llm_learning_rate": 4.965253394479728e-06, + "loss": 0.8008, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0004965253394479727, + "step": 796 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8288, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.965115422192155e-06, + "loss": 0.8041, + "output_text_mask_loss": 0.0843, + "predictor_learning_rate": 0.0004965115422192154, + "step": 797 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8286, + "image_masks_loss": 0.0205, + "llm_learning_rate": 4.9649771784405816e-06, + "loss": 0.7155, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.0004964977178440581, + "step": 798 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8284, + "image_masks_loss": 0.0193, + "llm_learning_rate": 4.96483866324023e-06, + "loss": 0.8369, + "output_text_mask_loss": 0.0288, + "predictor_learning_rate": 0.000496483866324023, + "step": 799 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8282, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.964699876606356e-06, + "loss": 0.7948, + "output_text_mask_loss": 0.1083, + "predictor_learning_rate": 0.0004964699876606356, + "step": 800 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.828, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.964560818554241e-06, + "loss": 0.8668, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000496456081855424, + "step": 801 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8278, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.964421489099198e-06, + "loss": 0.8655, + "output_text_mask_loss": 0.0851, + "predictor_learning_rate": 0.0004964421489099198, + "step": 802 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8276, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.964281888256573e-06, + "loss": 0.8092, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004964281888256572, + "step": 803 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8274, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.964142016041736e-06, + "loss": 0.9016, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 0.0004964142016041736, + "step": 804 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8272, + "image_masks_loss": 0.024, + "llm_learning_rate": 4.9640018724700925e-06, + "loss": 0.6737, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004964001872470092, + "step": 805 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.827, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.9638614575570735e-06, + "loss": 0.7796, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004963861457557073, + "step": 806 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8268, + "image_masks_loss": 0.034, + "llm_learning_rate": 4.9637207713181435e-06, + "loss": 0.9148, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0004963720771318143, + "step": 807 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8266, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.963579813768795e-06, + "loss": 0.8138, + "output_text_mask_loss": 0.1122, + "predictor_learning_rate": 0.0004963579813768794, + "step": 808 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8264, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.963438584924549e-06, + "loss": 0.7513, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 0.0004963438584924549, + "step": 809 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8262, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.963297084800959e-06, + "loss": 0.7724, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004963297084800959, + "step": 810 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8261, + "image_masks_loss": 0.0234, + "llm_learning_rate": 4.963155313413608e-06, + "loss": 0.8317, + "output_text_mask_loss": 0.0626, + "predictor_learning_rate": 0.0004963155313413608, + "step": 811 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8259, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.963013270778106e-06, + "loss": 0.914, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 0.0004963013270778106, + "step": 812 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8257, + "image_masks_loss": 0.0542, + "llm_learning_rate": 4.962870956910098e-06, + "loss": 0.8923, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004962870956910098, + "step": 813 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8255, + "image_masks_loss": 0.023, + "llm_learning_rate": 4.962728371825254e-06, + "loss": 0.7638, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004962728371825253, + "step": 814 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8253, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.962585515539275e-06, + "loss": 0.8061, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004962585515539275, + "step": 815 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8251, + "image_masks_loss": 0.0192, + "llm_learning_rate": 4.962442388067896e-06, + "loss": 0.7434, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004962442388067895, + "step": 816 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8249, + "image_masks_loss": 0.0097, + "llm_learning_rate": 4.9622989894268745e-06, + "loss": 0.818, + "output_text_mask_loss": 0.0999, + "predictor_learning_rate": 0.0004962298989426875, + "step": 817 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8247, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.962155319632005e-06, + "loss": 0.8026, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0004962155319632004, + "step": 818 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8245, + "image_masks_loss": 0.042, + "llm_learning_rate": 4.962011378699107e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004962011378699106, + "step": 819 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8243, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.961867166644032e-06, + "loss": 0.7269, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004961867166644031, + "step": 820 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8241, + "image_masks_loss": 0.0515, + "llm_learning_rate": 4.96172268348266e-06, + "loss": 0.7039, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000496172268348266, + "step": 821 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8239, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.9615779292309045e-06, + "loss": 0.7814, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004961577929230904, + "step": 822 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8237, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.961432903904704e-06, + "loss": 0.8027, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.0004961432903904703, + "step": 823 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8235, + "image_masks_loss": 0.0079, + "llm_learning_rate": 4.961287607520029e-06, + "loss": 0.9118, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.0004961287607520028, + "step": 824 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8233, + "image_masks_loss": 0.006, + "llm_learning_rate": 4.96114204009288e-06, + "loss": 0.7812, + "output_text_mask_loss": 0.0227, + "predictor_learning_rate": 0.000496114204009288, + "step": 825 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8231, + "image_masks_loss": 0.0322, + "llm_learning_rate": 4.9609962016392885e-06, + "loss": 0.8184, + "output_text_mask_loss": 0.0645, + "predictor_learning_rate": 0.0004960996201639288, + "step": 826 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8229, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.960850092175313e-06, + "loss": 0.7856, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004960850092175313, + "step": 827 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8227, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.960703711717045e-06, + "loss": 0.8443, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0004960703711717044, + "step": 828 + }, + { + "epoch": 0.08, + "gumbel_tau": 0.8226, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.960557060280602e-06, + "loss": 0.7818, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004960557060280601, + "step": 829 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8224, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.9604101378821355e-06, + "loss": 0.8261, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004960410137882135, + "step": 830 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8222, + "image_masks_loss": 0.0319, + "llm_learning_rate": 4.9602629445378244e-06, + "loss": 0.7237, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004960262944537824, + "step": 831 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.822, + "image_masks_loss": 0.0319, + "llm_learning_rate": 4.9601154802638774e-06, + "loss": 0.7705, + "output_text_mask_loss": 0.0938, + "predictor_learning_rate": 0.0004960115480263877, + "step": 832 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8218, + "image_masks_loss": 0.0198, + "llm_learning_rate": 4.959967745076535e-06, + "loss": 0.8051, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004959967745076534, + "step": 833 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8216, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.959819738992064e-06, + "loss": 0.8327, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.0004959819738992063, + "step": 834 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8214, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.959671462026765e-06, + "loss": 0.7601, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004959671462026764, + "step": 835 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8212, + "image_masks_loss": 0.0315, + "llm_learning_rate": 4.959522914196964e-06, + "loss": 0.7589, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004959522914196965, + "step": 836 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.821, + "image_masks_loss": 0.0376, + "llm_learning_rate": 4.959374095519024e-06, + "loss": 0.8977, + "output_text_mask_loss": 0.0923, + "predictor_learning_rate": 0.0004959374095519024, + "step": 837 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8208, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.959225006009329e-06, + "loss": 0.7417, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.0004959225006009329, + "step": 838 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8206, + "image_masks_loss": 0.0047, + "llm_learning_rate": 4.9590756456842994e-06, + "loss": 0.8413, + "output_text_mask_loss": 0.174, + "predictor_learning_rate": 0.0004959075645684298, + "step": 839 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8204, + "image_masks_loss": 0.0217, + "llm_learning_rate": 4.958926014560382e-06, + "loss": 0.8432, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.0004958926014560382, + "step": 840 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8202, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.958776112654055e-06, + "loss": 0.8376, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004958776112654055, + "step": 841 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.82, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.958625939981826e-06, + "loss": 0.813, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0004958625939981826, + "step": 842 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8198, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.958475496560231e-06, + "loss": 0.8305, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 0.0004958475496560231, + "step": 843 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8196, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.95832478240584e-06, + "loss": 0.7789, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004958324782405839, + "step": 844 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8195, + "image_masks_loss": 0.0182, + "llm_learning_rate": 4.958173797535247e-06, + "loss": 0.7053, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004958173797535246, + "step": 845 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8193, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.95802254196508e-06, + "loss": 0.8643, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.000495802254196508, + "step": 846 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8191, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.9578710157119956e-06, + "loss": 0.7618, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004957871015711995, + "step": 847 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8189, + "image_masks_loss": 0.0139, + "llm_learning_rate": 4.9577192187926806e-06, + "loss": 0.8313, + "output_text_mask_loss": 0.1053, + "predictor_learning_rate": 0.000495771921879268, + "step": 848 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8187, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.957567151223851e-06, + "loss": 0.7428, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004957567151223851, + "step": 849 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8185, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.957414813022253e-06, + "loss": 0.7132, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004957414813022252, + "step": 850 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8183, + "image_masks_loss": 0.0322, + "llm_learning_rate": 4.9572622042046614e-06, + "loss": 0.6715, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004957262204204661, + "step": 851 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8181, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.957109324787883e-06, + "loss": 0.7183, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004957109324787883, + "step": 852 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8179, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.956956174788753e-06, + "loss": 0.8682, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.0004956956174788753, + "step": 853 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8177, + "image_masks_loss": 0.0181, + "llm_learning_rate": 4.956802754224136e-06, + "loss": 0.8596, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004956802754224136, + "step": 854 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8175, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.956649063110927e-06, + "loss": 0.7331, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.0004956649063110926, + "step": 855 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8173, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.956495101466051e-06, + "loss": 0.7818, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.0004956495101466051, + "step": 856 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8171, + "image_masks_loss": 0.0381, + "llm_learning_rate": 4.956340869306464e-06, + "loss": 0.8031, + "output_text_mask_loss": 0.1434, + "predictor_learning_rate": 0.0004956340869306464, + "step": 857 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8169, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.956186366649148e-06, + "loss": 0.8427, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0004956186366649148, + "step": 858 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8168, + "image_masks_loss": 0.0385, + "llm_learning_rate": 4.956031593511119e-06, + "loss": 0.7833, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004956031593511119, + "step": 859 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8166, + "image_masks_loss": 0.0242, + "llm_learning_rate": 4.955876549909421e-06, + "loss": 0.8398, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.000495587654990942, + "step": 860 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8164, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.955721235861126e-06, + "loss": 0.8122, + "output_text_mask_loss": 0.0725, + "predictor_learning_rate": 0.0004955721235861126, + "step": 861 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8162, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.955565651383339e-06, + "loss": 0.9379, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.0004955565651383339, + "step": 862 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.816, + "image_masks_loss": 0.0492, + "llm_learning_rate": 4.955409796493192e-06, + "loss": 0.6948, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004955409796493192, + "step": 863 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8158, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.95525367120785e-06, + "loss": 0.8131, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.000495525367120785, + "step": 864 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8156, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.955097275544505e-06, + "loss": 0.8416, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004955097275544505, + "step": 865 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8154, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.954940609520379e-06, + "loss": 0.7159, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004954940609520379, + "step": 866 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8152, + "image_masks_loss": 0.0067, + "llm_learning_rate": 4.954783673152725e-06, + "loss": 0.7359, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0004954783673152725, + "step": 867 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.815, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.954626466458826e-06, + "loss": 0.7945, + "output_text_mask_loss": 0.0364, + "predictor_learning_rate": 0.0004954626466458826, + "step": 868 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8148, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.9544689894559915e-06, + "loss": 0.6948, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004954468989455991, + "step": 869 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8146, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.954311242161566e-06, + "loss": 0.8484, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004954311242161565, + "step": 870 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8144, + "image_masks_loss": 0.0399, + "llm_learning_rate": 4.954153224592919e-06, + "loss": 0.712, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004954153224592919, + "step": 871 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8143, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.953994936767453e-06, + "loss": 0.7254, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.0004953994936767452, + "step": 872 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8141, + "image_masks_loss": 0.0242, + "llm_learning_rate": 4.953836378702598e-06, + "loss": 0.7859, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004953836378702597, + "step": 873 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8139, + "image_masks_loss": 0.0414, + "llm_learning_rate": 4.953677550415815e-06, + "loss": 0.8117, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.0004953677550415815, + "step": 874 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8137, + "image_masks_loss": 0.0229, + "llm_learning_rate": 4.953518451924596e-06, + "loss": 0.7998, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004953518451924595, + "step": 875 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8135, + "image_masks_loss": 0.0359, + "llm_learning_rate": 4.953359083246459e-06, + "loss": 0.787, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004953359083246459, + "step": 876 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8133, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.953199444398956e-06, + "loss": 0.7107, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004953199444398956, + "step": 877 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8131, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.9530395353996655e-06, + "loss": 0.8534, + "output_text_mask_loss": 0.0477, + "predictor_learning_rate": 0.0004953039535399665, + "step": 878 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8129, + "image_masks_loss": 0.0233, + "llm_learning_rate": 4.952879356266197e-06, + "loss": 0.8171, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004952879356266197, + "step": 879 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8127, + "image_masks_loss": 0.0614, + "llm_learning_rate": 4.952718907016191e-06, + "loss": 0.7641, + "output_text_mask_loss": 0.0406, + "predictor_learning_rate": 0.000495271890701619, + "step": 880 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8125, + "image_masks_loss": 0.0243, + "llm_learning_rate": 4.9525581876673144e-06, + "loss": 0.8032, + "output_text_mask_loss": 0.0067, + "predictor_learning_rate": 0.0004952558187667314, + "step": 881 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8123, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.952397198237269e-06, + "loss": 0.6368, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004952397198237268, + "step": 882 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8121, + "image_masks_loss": 0.0439, + "llm_learning_rate": 4.95223593874378e-06, + "loss": 0.852, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000495223593874378, + "step": 883 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8119, + "image_masks_loss": 0.0366, + "llm_learning_rate": 4.952074409204609e-06, + "loss": 0.6592, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0004952074409204608, + "step": 884 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8118, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.951912609637542e-06, + "loss": 0.782, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004951912609637542, + "step": 885 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8116, + "image_masks_loss": 0.0192, + "llm_learning_rate": 4.9517505400603955e-06, + "loss": 0.7918, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0004951750540060395, + "step": 886 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8114, + "image_masks_loss": 0.0322, + "llm_learning_rate": 4.95158820049102e-06, + "loss": 0.7751, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004951588200491019, + "step": 887 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8112, + "image_masks_loss": 0.0473, + "llm_learning_rate": 4.95142559094729e-06, + "loss": 0.7983, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.0004951425590947289, + "step": 888 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.811, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.951262711447114e-06, + "loss": 0.7446, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0004951262711447114, + "step": 889 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8108, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.951099562008429e-06, + "loss": 0.8252, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004951099562008428, + "step": 890 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8106, + "image_masks_loss": 0.0159, + "llm_learning_rate": 4.9509361426491994e-06, + "loss": 0.8666, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004950936142649199, + "step": 891 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8104, + "image_masks_loss": 0.0381, + "llm_learning_rate": 4.950772453387423e-06, + "loss": 0.8193, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004950772453387423, + "step": 892 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8102, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.950608494241125e-06, + "loss": 0.6976, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0004950608494241125, + "step": 893 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.81, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.950444265228361e-06, + "loss": 0.9611, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.0004950444265228361, + "step": 894 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8098, + "image_masks_loss": 0.0444, + "llm_learning_rate": 4.950279766367218e-06, + "loss": 0.7933, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004950279766367217, + "step": 895 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8097, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.950114997675808e-06, + "loss": 0.7718, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.0004950114997675808, + "step": 896 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8095, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.949949959172277e-06, + "loss": 0.7962, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004949949959172276, + "step": 897 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8093, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.949784650874799e-06, + "loss": 0.9204, + "output_text_mask_loss": 0.0448, + "predictor_learning_rate": 0.0004949784650874799, + "step": 898 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8091, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.949619072801579e-06, + "loss": 0.8368, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.0004949619072801579, + "step": 899 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8089, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.94945322497085e-06, + "loss": 0.7803, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000494945322497085, + "step": 900 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8087, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.949287107400877e-06, + "loss": 0.7935, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.0004949287107400877, + "step": 901 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8085, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.949120720109952e-06, + "loss": 0.8477, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004949120720109952, + "step": 902 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8083, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.948954063116397e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004948954063116397, + "step": 903 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8081, + "image_masks_loss": 0.0414, + "llm_learning_rate": 4.9487871364385664e-06, + "loss": 0.6975, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004948787136438566, + "step": 904 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8079, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.948619940094842e-06, + "loss": 0.732, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.0004948619940094842, + "step": 905 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8077, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.948452474103636e-06, + "loss": 0.8501, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004948452474103636, + "step": 906 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8076, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.948284738483389e-06, + "loss": 0.8062, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0004948284738483389, + "step": 907 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8074, + "image_masks_loss": 0.0303, + "llm_learning_rate": 4.948116733252574e-06, + "loss": 0.7707, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.0004948116733252574, + "step": 908 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8072, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.947948458429691e-06, + "loss": 0.7233, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.000494794845842969, + "step": 909 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.807, + "image_masks_loss": 0.0355, + "llm_learning_rate": 4.9477799140332715e-06, + "loss": 0.7792, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004947779914033271, + "step": 910 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8068, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.947611100081875e-06, + "loss": 0.6996, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004947611100081875, + "step": 911 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8066, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.947442016594093e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004947442016594092, + "step": 912 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8064, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.947272663588544e-06, + "loss": 0.6667, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004947272663588544, + "step": 913 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8062, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.947103041083879e-06, + "loss": 0.826, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.0004947103041083879, + "step": 914 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.806, + "image_masks_loss": 0.0105, + "llm_learning_rate": 4.946933149098776e-06, + "loss": 0.7518, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0004946933149098775, + "step": 915 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8058, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.946762987651944e-06, + "loss": 0.6325, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004946762987651943, + "step": 916 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8057, + "image_masks_loss": 0.0152, + "llm_learning_rate": 4.946592556762122e-06, + "loss": 0.804, + "output_text_mask_loss": 0.0009, + "predictor_learning_rate": 0.0004946592556762121, + "step": 917 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8055, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.946421856448078e-06, + "loss": 0.8085, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004946421856448078, + "step": 918 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8053, + "image_masks_loss": 0.0067, + "llm_learning_rate": 4.9462508867286104e-06, + "loss": 0.7472, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000494625088672861, + "step": 919 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8051, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.946079647622547e-06, + "loss": 0.8075, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004946079647622547, + "step": 920 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8049, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.945908139148743e-06, + "loss": 0.8423, + "output_text_mask_loss": 0.0043, + "predictor_learning_rate": 0.0004945908139148742, + "step": 921 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8047, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.9457363613260865e-06, + "loss": 0.85, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.0004945736361326086, + "step": 922 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8045, + "image_masks_loss": 0.0368, + "llm_learning_rate": 4.945564314173496e-06, + "loss": 0.6841, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004945564314173495, + "step": 923 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8043, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.945391997709914e-06, + "loss": 0.7416, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0004945391997709914, + "step": 924 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8041, + "image_masks_loss": 0.0141, + "llm_learning_rate": 4.94521941195432e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.1694, + "predictor_learning_rate": 0.000494521941195432, + "step": 925 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8039, + "image_masks_loss": 0.0147, + "llm_learning_rate": 4.945046556925716e-06, + "loss": 0.8643, + "output_text_mask_loss": 0.0414, + "predictor_learning_rate": 0.0004945046556925716, + "step": 926 + }, + { + "epoch": 0.09, + "gumbel_tau": 0.8038, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.94487343264314e-06, + "loss": 0.8119, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004944873432643139, + "step": 927 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8036, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.944700039125656e-06, + "loss": 0.7879, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0004944700039125656, + "step": 928 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8034, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.944526376392358e-06, + "loss": 0.7845, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.0004944526376392358, + "step": 929 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8032, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.944352444462371e-06, + "loss": 0.7191, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004944352444462371, + "step": 930 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.803, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.944178243354848e-06, + "loss": 0.8, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.0004944178243354848, + "step": 931 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8028, + "image_masks_loss": 0.0152, + "llm_learning_rate": 4.944003773088972e-06, + "loss": 0.8423, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004944003773088972, + "step": 932 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8026, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.943829033683958e-06, + "loss": 0.7804, + "output_text_mask_loss": 0.1434, + "predictor_learning_rate": 0.0004943829033683957, + "step": 933 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8024, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.943654025159047e-06, + "loss": 0.8126, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 0.0004943654025159046, + "step": 934 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8022, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.943478747533511e-06, + "loss": 0.7669, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.000494347874753351, + "step": 935 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.802, + "image_masks_loss": 0.005, + "llm_learning_rate": 4.943303200826653e-06, + "loss": 0.728, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004943303200826653, + "step": 936 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8019, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.943127385057804e-06, + "loss": 0.8788, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004943127385057804, + "step": 937 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8017, + "image_masks_loss": 0.016, + "llm_learning_rate": 4.9429513002463255e-06, + "loss": 0.9239, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.0004942951300246325, + "step": 938 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8015, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.94277494641161e-06, + "loss": 0.7668, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.0004942774946411609, + "step": 939 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8013, + "image_masks_loss": 0.0311, + "llm_learning_rate": 4.942598323573074e-06, + "loss": 0.8292, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0004942598323573075, + "step": 940 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8011, + "image_masks_loss": 0.0414, + "llm_learning_rate": 4.942421431750171e-06, + "loss": 0.8129, + "output_text_mask_loss": 0.0992, + "predictor_learning_rate": 0.0004942421431750171, + "step": 941 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8009, + "image_masks_loss": 0.0059, + "llm_learning_rate": 4.94224427096238e-06, + "loss": 0.7252, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.000494224427096238, + "step": 942 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8007, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.94206684122921e-06, + "loss": 0.78, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004942066841229209, + "step": 943 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8005, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.941889142570199e-06, + "loss": 0.7856, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004941889142570198, + "step": 944 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8003, + "image_masks_loss": 0.0241, + "llm_learning_rate": 4.941711175004917e-06, + "loss": 0.7489, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004941711175004917, + "step": 945 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8002, + "image_masks_loss": 0.0376, + "llm_learning_rate": 4.941532938552962e-06, + "loss": 0.8594, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004941532938552962, + "step": 946 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.8, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.9413544332339615e-06, + "loss": 0.7576, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004941354433233962, + "step": 947 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7998, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.941175659067572e-06, + "loss": 0.8596, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004941175659067572, + "step": 948 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7996, + "image_masks_loss": 0.0243, + "llm_learning_rate": 4.940996616073482e-06, + "loss": 0.8395, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.0004940996616073482, + "step": 949 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7994, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.940817304271408e-06, + "loss": 0.7642, + "output_text_mask_loss": 0.0923, + "predictor_learning_rate": 0.0004940817304271408, + "step": 950 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7992, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.940637723681095e-06, + "loss": 0.8059, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004940637723681096, + "step": 951 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.799, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.9404578743223204e-06, + "loss": 0.8614, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004940457874322321, + "step": 952 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7988, + "image_masks_loss": 0.0189, + "llm_learning_rate": 4.940277756214888e-06, + "loss": 0.7634, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004940277756214888, + "step": 953 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7987, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.940097369378634e-06, + "loss": 0.8277, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0004940097369378635, + "step": 954 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7985, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.939916713833423e-06, + "loss": 0.7622, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004939916713833423, + "step": 955 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7983, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.939735789599148e-06, + "loss": 0.8145, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.0004939735789599148, + "step": 956 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7981, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.9395545966957335e-06, + "loss": 0.7696, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0004939554596695734, + "step": 957 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7979, + "image_masks_loss": 0.0427, + "llm_learning_rate": 4.939373135143134e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004939373135143133, + "step": 958 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7977, + "image_masks_loss": 0.0209, + "llm_learning_rate": 4.9391914049613305e-06, + "loss": 0.8301, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004939191404961331, + "step": 959 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7975, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.939009406170337e-06, + "loss": 0.8675, + "output_text_mask_loss": 0.0916, + "predictor_learning_rate": 0.0004939009406170336, + "step": 960 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7973, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.9388271387901945e-06, + "loss": 0.762, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004938827138790194, + "step": 961 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7971, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.938644602840974e-06, + "loss": 0.8564, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0004938644602840974, + "step": 962 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.797, + "image_masks_loss": 0.0234, + "llm_learning_rate": 4.93846179834278e-06, + "loss": 0.82, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004938461798342779, + "step": 963 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7968, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.9382787253157404e-06, + "loss": 0.8675, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000493827872531574, + "step": 964 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7966, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.938095383780017e-06, + "loss": 0.7337, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004938095383780017, + "step": 965 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7964, + "image_masks_loss": 0.0228, + "llm_learning_rate": 4.937911773755798e-06, + "loss": 0.6821, + "output_text_mask_loss": 0.0595, + "predictor_learning_rate": 0.0004937911773755799, + "step": 966 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7962, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.9377278952633055e-06, + "loss": 0.7919, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.0004937727895263306, + "step": 967 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.796, + "image_masks_loss": 0.0212, + "llm_learning_rate": 4.937543748322787e-06, + "loss": 0.8752, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0004937543748322787, + "step": 968 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7958, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.937359332954521e-06, + "loss": 0.806, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0004937359332954521, + "step": 969 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7956, + "image_masks_loss": 0.0068, + "llm_learning_rate": 4.937174649178818e-06, + "loss": 0.7511, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004937174649178817, + "step": 970 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7955, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.936989697016013e-06, + "loss": 0.8125, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.0004936989697016012, + "step": 971 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7953, + "image_masks_loss": 0.0186, + "llm_learning_rate": 4.9368044764864735e-06, + "loss": 0.8519, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.0004936804476486472, + "step": 972 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7951, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.936618987610597e-06, + "loss": 0.667, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004936618987610597, + "step": 973 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7949, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.936433230408812e-06, + "loss": 0.8384, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004936433230408811, + "step": 974 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7947, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.936247204901572e-06, + "loss": 0.7922, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.0004936247204901571, + "step": 975 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7945, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.936060911109362e-06, + "loss": 0.8327, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004936060911109362, + "step": 976 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7943, + "image_masks_loss": 0.0236, + "llm_learning_rate": 4.9358743490527e-06, + "loss": 0.8189, + "output_text_mask_loss": 0.0412, + "predictor_learning_rate": 0.00049358743490527, + "step": 977 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7941, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.9356875187521295e-06, + "loss": 0.803, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004935687518752129, + "step": 978 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.794, + "image_masks_loss": 0.0211, + "llm_learning_rate": 4.935500420228223e-06, + "loss": 0.8374, + "output_text_mask_loss": 0.0393, + "predictor_learning_rate": 0.0004935500420228222, + "step": 979 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7938, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.935313053501586e-06, + "loss": 0.6618, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004935313053501585, + "step": 980 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7936, + "image_masks_loss": 0.0387, + "llm_learning_rate": 4.93512541859285e-06, + "loss": 0.8467, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.000493512541859285, + "step": 981 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7934, + "image_masks_loss": 0.0288, + "llm_learning_rate": 4.93493751552268e-06, + "loss": 0.6798, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004934937515522679, + "step": 982 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7932, + "image_masks_loss": 0.016, + "llm_learning_rate": 4.934749344311766e-06, + "loss": 0.8034, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004934749344311766, + "step": 983 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.793, + "image_masks_loss": 0.0481, + "llm_learning_rate": 4.934560904980832e-06, + "loss": 0.7421, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.0004934560904980832, + "step": 984 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7928, + "image_masks_loss": 0.0484, + "llm_learning_rate": 4.934372197550628e-06, + "loss": 0.8084, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004934372197550628, + "step": 985 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7926, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.934183222041935e-06, + "loss": 0.8003, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0004934183222041935, + "step": 986 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7925, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.933993978475564e-06, + "loss": 0.7152, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 0.0004933993978475564, + "step": 987 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7923, + "image_masks_loss": 0.0307, + "llm_learning_rate": 4.933804466872355e-06, + "loss": 0.8596, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 0.0004933804466872355, + "step": 988 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7921, + "image_masks_loss": 0.017, + "llm_learning_rate": 4.9336146872531766e-06, + "loss": 0.7828, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004933614687253177, + "step": 989 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7919, + "image_masks_loss": 0.0359, + "llm_learning_rate": 4.933424639638928e-06, + "loss": 0.8216, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004933424639638927, + "step": 990 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7917, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.933234324050537e-06, + "loss": 0.8065, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.0004933234324050538, + "step": 991 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7915, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.933043740508964e-06, + "loss": 0.6957, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004933043740508964, + "step": 992 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7913, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.9328528890351946e-06, + "loss": 0.7722, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.0004932852889035194, + "step": 993 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7911, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.932661769650245e-06, + "loss": 0.8414, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0004932661769650245, + "step": 994 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.791, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.932470382375164e-06, + "loss": 0.8745, + "output_text_mask_loss": 0.0492, + "predictor_learning_rate": 0.0004932470382375164, + "step": 995 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7908, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.932278727231027e-06, + "loss": 0.7978, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.0004932278727231026, + "step": 996 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7906, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.932086804238937e-06, + "loss": 0.6856, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004932086804238936, + "step": 997 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7904, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.931894613420032e-06, + "loss": 0.7739, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.0004931894613420031, + "step": 998 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7902, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.931702154795475e-06, + "loss": 0.7705, + "output_text_mask_loss": 0.119, + "predictor_learning_rate": 0.0004931702154795474, + "step": 999 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.79, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.9315094283864605e-06, + "loss": 0.7831, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004931509428386461, + "step": 1000 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7898, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.931316434214212e-06, + "loss": 0.8472, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004931316434214211, + "step": 1001 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7897, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.931123172299982e-06, + "loss": 0.7867, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.0004931123172299981, + "step": 1002 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7895, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.930929642665052e-06, + "loss": 0.7162, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004930929642665052, + "step": 1003 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7893, + "image_masks_loss": 0.0139, + "llm_learning_rate": 4.930735845330736e-06, + "loss": 0.8248, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004930735845330737, + "step": 1004 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7891, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.930541780318375e-06, + "loss": 0.8278, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004930541780318375, + "step": 1005 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7889, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.930347447649339e-06, + "loss": 0.8019, + "output_text_mask_loss": 0.0591, + "predictor_learning_rate": 0.0004930347447649338, + "step": 1006 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7887, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.930152847345028e-06, + "loss": 0.7247, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0004930152847345028, + "step": 1007 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7885, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.929957979426874e-06, + "loss": 0.8264, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004929957979426874, + "step": 1008 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7884, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.929762843916333e-06, + "loss": 0.7535, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004929762843916333, + "step": 1009 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7882, + "image_masks_loss": 0.0238, + "llm_learning_rate": 4.9295674408348965e-06, + "loss": 0.8067, + "output_text_mask_loss": 0.0273, + "predictor_learning_rate": 0.0004929567440834896, + "step": 1010 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.788, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.929371770204082e-06, + "loss": 0.8705, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.0004929371770204081, + "step": 1011 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7878, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.929175832045436e-06, + "loss": 0.7503, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004929175832045436, + "step": 1012 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7876, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.928979626380537e-06, + "loss": 0.803, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.0004928979626380537, + "step": 1013 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7874, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.928783153230991e-06, + "loss": 0.8228, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.0004928783153230991, + "step": 1014 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7872, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.928586412618436e-06, + "loss": 0.838, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 0.0004928586412618435, + "step": 1015 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7871, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.928389404564533e-06, + "loss": 0.8214, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0004928389404564534, + "step": 1016 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7869, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.928192129090982e-06, + "loss": 0.8585, + "output_text_mask_loss": 0.0946, + "predictor_learning_rate": 0.0004928192129090981, + "step": 1017 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7867, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.927994586219504e-06, + "loss": 0.8411, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004927994586219505, + "step": 1018 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7865, + "image_masks_loss": 0.0077, + "llm_learning_rate": 4.927796775971855e-06, + "loss": 0.7268, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004927796775971854, + "step": 1019 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7863, + "image_masks_loss": 0.0085, + "llm_learning_rate": 4.9275986983698165e-06, + "loss": 0.8738, + "output_text_mask_loss": 0.0587, + "predictor_learning_rate": 0.0004927598698369816, + "step": 1020 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7861, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.9274003534352025e-06, + "loss": 0.6479, + "output_text_mask_loss": 0.0961, + "predictor_learning_rate": 0.0004927400353435202, + "step": 1021 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7859, + "image_masks_loss": 0.0217, + "llm_learning_rate": 4.927201741189854e-06, + "loss": 0.756, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0004927201741189854, + "step": 1022 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7858, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.927002861655644e-06, + "loss": 0.7986, + "output_text_mask_loss": 0.0652, + "predictor_learning_rate": 0.0004927002861655643, + "step": 1023 + }, + { + "epoch": 0.1, + "gumbel_tau": 0.7856, + "image_masks_loss": 0.0366, + "llm_learning_rate": 4.926803714854472e-06, + "loss": 0.7941, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0004926803714854472, + "step": 1024 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7854, + "image_masks_loss": 0.0385, + "llm_learning_rate": 4.9266043008082705e-06, + "loss": 0.805, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004926604300808271, + "step": 1025 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7852, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.926404619538998e-06, + "loss": 0.8466, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004926404619538997, + "step": 1026 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.785, + "image_masks_loss": 0.0108, + "llm_learning_rate": 4.926204671068645e-06, + "loss": 0.6433, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004926204671068645, + "step": 1027 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7848, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.9260044554192285e-06, + "loss": 0.7157, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.0004926004455419228, + "step": 1028 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7846, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.925803972612798e-06, + "loss": 0.7883, + "output_text_mask_loss": 0.1442, + "predictor_learning_rate": 0.0004925803972612798, + "step": 1029 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7845, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.92560322267143e-06, + "loss": 0.7162, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004925603222671429, + "step": 1030 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7843, + "image_masks_loss": 0.0147, + "llm_learning_rate": 4.9254022056172335e-06, + "loss": 0.847, + "output_text_mask_loss": 0.0854, + "predictor_learning_rate": 0.0004925402205617234, + "step": 1031 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7841, + "image_masks_loss": 0.0303, + "llm_learning_rate": 4.925200921472343e-06, + "loss": 0.8808, + "output_text_mask_loss": 0.0458, + "predictor_learning_rate": 0.0004925200921472343, + "step": 1032 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7839, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.924999370258926e-06, + "loss": 0.7277, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004924999370258925, + "step": 1033 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7837, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.924797551999175e-06, + "loss": 0.8272, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004924797551999175, + "step": 1034 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7835, + "image_masks_loss": 0.004, + "llm_learning_rate": 4.924595466715318e-06, + "loss": 0.7803, + "output_text_mask_loss": 0.0349, + "predictor_learning_rate": 0.0004924595466715318, + "step": 1035 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7833, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.9243931144296075e-06, + "loss": 0.7591, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004924393114429607, + "step": 1036 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7832, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.924190495164327e-06, + "loss": 0.8173, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0004924190495164326, + "step": 1037 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.783, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.923987608941789e-06, + "loss": 0.8556, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000492398760894179, + "step": 1038 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7828, + "image_masks_loss": 0.0349, + "llm_learning_rate": 4.9237844557843375e-06, + "loss": 0.7563, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004923784455784337, + "step": 1039 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7826, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.923581035714342e-06, + "loss": 0.7734, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.0004923581035714342, + "step": 1040 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7824, + "image_masks_loss": 0.0336, + "llm_learning_rate": 4.923377348754205e-06, + "loss": 0.8212, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004923377348754204, + "step": 1041 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7822, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.923173394926356e-06, + "loss": 0.7035, + "output_text_mask_loss": 0.0736, + "predictor_learning_rate": 0.0004923173394926355, + "step": 1042 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7821, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.9229691742532556e-06, + "loss": 0.6574, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004922969174253255, + "step": 1043 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7819, + "image_masks_loss": 0.0366, + "llm_learning_rate": 4.922764686757393e-06, + "loss": 0.7539, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004922764686757393, + "step": 1044 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7817, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.922559932461286e-06, + "loss": 0.7111, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004922559932461286, + "step": 1045 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7815, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.922354911387484e-06, + "loss": 0.7032, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004922354911387483, + "step": 1046 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7813, + "image_masks_loss": 0.0059, + "llm_learning_rate": 4.922149623558563e-06, + "loss": 0.8235, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004922149623558563, + "step": 1047 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7811, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.92194406899713e-06, + "loss": 0.8386, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.000492194406899713, + "step": 1048 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7809, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.921738247725823e-06, + "loss": 0.8669, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004921738247725822, + "step": 1049 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7808, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.921532159767304e-06, + "loss": 0.8342, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0004921532159767304, + "step": 1050 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7806, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.921325805144271e-06, + "loss": 0.7311, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000492132580514427, + "step": 1051 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7804, + "image_masks_loss": 0.0158, + "llm_learning_rate": 4.921119183879447e-06, + "loss": 0.8645, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0004921119183879446, + "step": 1052 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7802, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.920912295995585e-06, + "loss": 0.7201, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004920912295995584, + "step": 1053 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.78, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.920705141515469e-06, + "loss": 0.7468, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.0004920705141515469, + "step": 1054 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7798, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.920497720461911e-06, + "loss": 0.7724, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.000492049772046191, + "step": 1055 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7797, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.920290032857752e-06, + "loss": 0.792, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004920290032857751, + "step": 1056 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7795, + "image_masks_loss": 0.0111, + "llm_learning_rate": 4.920082078725864e-06, + "loss": 0.8197, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004920082078725864, + "step": 1057 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7793, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.919873858089147e-06, + "loss": 0.7845, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004919873858089147, + "step": 1058 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7791, + "image_masks_loss": 0.0205, + "llm_learning_rate": 4.919665370970531e-06, + "loss": 0.8222, + "output_text_mask_loss": 0.0087, + "predictor_learning_rate": 0.0004919665370970531, + "step": 1059 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7789, + "image_masks_loss": 0.0186, + "llm_learning_rate": 4.919456617392975e-06, + "loss": 0.7463, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0004919456617392974, + "step": 1060 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7787, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.9192475973794666e-06, + "loss": 0.7315, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004919247597379466, + "step": 1061 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7786, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.919038310953024e-06, + "loss": 0.806, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004919038310953023, + "step": 1062 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7784, + "image_masks_loss": 0.0383, + "llm_learning_rate": 4.918828758136694e-06, + "loss": 0.7443, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004918828758136693, + "step": 1063 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7782, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.918618938953555e-06, + "loss": 0.85, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.0004918618938953554, + "step": 1064 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.778, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.918408853426709e-06, + "loss": 0.8179, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0004918408853426709, + "step": 1065 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7778, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.9181985015792945e-06, + "loss": 0.8299, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.0004918198501579294, + "step": 1066 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7776, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.917987883434475e-06, + "loss": 0.7334, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004917987883434475, + "step": 1067 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7775, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.917776999015443e-06, + "loss": 0.9328, + "output_text_mask_loss": 0.1442, + "predictor_learning_rate": 0.0004917776999015443, + "step": 1068 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7773, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.917565848345423e-06, + "loss": 0.7868, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004917565848345423, + "step": 1069 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7771, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.917354431447667e-06, + "loss": 0.8958, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004917354431447667, + "step": 1070 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7769, + "image_masks_loss": 0.023, + "llm_learning_rate": 4.917142748345457e-06, + "loss": 0.8439, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004917142748345456, + "step": 1071 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7767, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.916930799062103e-06, + "loss": 0.7227, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.0004916930799062102, + "step": 1072 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7765, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.9167185836209455e-06, + "loss": 0.7548, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004916718583620945, + "step": 1073 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7764, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.916506102045355e-06, + "loss": 0.8493, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004916506102045354, + "step": 1074 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7762, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.9162933543587295e-06, + "loss": 0.7211, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004916293354358729, + "step": 1075 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.776, + "image_masks_loss": 0.0072, + "llm_learning_rate": 4.916080340584498e-06, + "loss": 0.8409, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004916080340584498, + "step": 1076 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7758, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.915867060746118e-06, + "loss": 0.7858, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.0004915867060746118, + "step": 1077 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7756, + "image_masks_loss": 0.0343, + "llm_learning_rate": 4.915653514867076e-06, + "loss": 0.7399, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004915653514867076, + "step": 1078 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7754, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.915439702970889e-06, + "loss": 0.8416, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.0004915439702970889, + "step": 1079 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7753, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.9152256250811e-06, + "loss": 0.8535, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 0.00049152256250811, + "step": 1080 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7751, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.915011281221287e-06, + "loss": 0.7945, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 0.0004915011281221287, + "step": 1081 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7749, + "image_masks_loss": 0.008, + "llm_learning_rate": 4.914796671415052e-06, + "loss": 0.8142, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0004914796671415052, + "step": 1082 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7747, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.9145817956860286e-06, + "loss": 0.6463, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004914581795686029, + "step": 1083 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7745, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.914366654057881e-06, + "loss": 0.7822, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 0.000491436665405788, + "step": 1084 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7743, + "image_masks_loss": 0.0584, + "llm_learning_rate": 4.914151246554298e-06, + "loss": 0.8066, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004914151246554298, + "step": 1085 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7742, + "image_masks_loss": 0.0479, + "llm_learning_rate": 4.913935573199003e-06, + "loss": 0.8394, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.0004913935573199002, + "step": 1086 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.774, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.913719634015745e-06, + "loss": 0.8014, + "output_text_mask_loss": 0.0877, + "predictor_learning_rate": 0.0004913719634015745, + "step": 1087 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7738, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.913503429028305e-06, + "loss": 0.8676, + "output_text_mask_loss": 0.0877, + "predictor_learning_rate": 0.0004913503429028306, + "step": 1088 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7736, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.913286958260493e-06, + "loss": 0.7339, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004913286958260492, + "step": 1089 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7734, + "image_masks_loss": 0.0359, + "llm_learning_rate": 4.913070221736144e-06, + "loss": 0.8598, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.0004913070221736144, + "step": 1090 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7732, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.912853219479128e-06, + "loss": 0.7695, + "output_text_mask_loss": 0.0112, + "predictor_learning_rate": 0.0004912853219479127, + "step": 1091 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7731, + "image_masks_loss": 0.0228, + "llm_learning_rate": 4.912635951513341e-06, + "loss": 0.6625, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004912635951513341, + "step": 1092 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7729, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.912418417862709e-06, + "loss": 0.8088, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004912418417862708, + "step": 1093 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7727, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.912200618551187e-06, + "loss": 0.7647, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004912200618551187, + "step": 1094 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7725, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.911982553602761e-06, + "loss": 0.8084, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.0004911982553602761, + "step": 1095 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7723, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.911764223041443e-06, + "loss": 0.7168, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004911764223041443, + "step": 1096 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7722, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.9115456268912765e-06, + "loss": 0.7502, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0004911545626891277, + "step": 1097 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.772, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.911326765176334e-06, + "loss": 0.8359, + "output_text_mask_loss": 0.2121, + "predictor_learning_rate": 0.0004911326765176334, + "step": 1098 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7718, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.911107637920718e-06, + "loss": 0.8962, + "output_text_mask_loss": 0.0443, + "predictor_learning_rate": 0.0004911107637920718, + "step": 1099 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7716, + "image_masks_loss": 0.0397, + "llm_learning_rate": 4.910888245148557e-06, + "loss": 0.8477, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.0004910888245148557, + "step": 1100 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7714, + "image_masks_loss": 0.0039, + "llm_learning_rate": 4.910668586884014e-06, + "loss": 0.7788, + "output_text_mask_loss": 0.1404, + "predictor_learning_rate": 0.0004910668586884014, + "step": 1101 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7712, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.910448663151276e-06, + "loss": 0.8079, + "output_text_mask_loss": 0.0851, + "predictor_learning_rate": 0.0004910448663151275, + "step": 1102 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7711, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.910228473974562e-06, + "loss": 0.8448, + "output_text_mask_loss": 0.038, + "predictor_learning_rate": 0.0004910228473974562, + "step": 1103 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7709, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.91000801937812e-06, + "loss": 0.6759, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000491000801937812, + "step": 1104 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7707, + "image_masks_loss": 0.0425, + "llm_learning_rate": 4.9097872993862265e-06, + "loss": 0.8338, + "output_text_mask_loss": 0.1434, + "predictor_learning_rate": 0.0004909787299386226, + "step": 1105 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7705, + "image_masks_loss": 0.0238, + "llm_learning_rate": 4.909566314023188e-06, + "loss": 0.6622, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004909566314023187, + "step": 1106 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7703, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.90934506331334e-06, + "loss": 0.8627, + "output_text_mask_loss": 0.0412, + "predictor_learning_rate": 0.000490934506331334, + "step": 1107 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7702, + "image_masks_loss": 0.0198, + "llm_learning_rate": 4.909123547281047e-06, + "loss": 0.8082, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004909123547281047, + "step": 1108 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.77, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.908901765950703e-06, + "loss": 0.826, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.0004908901765950703, + "step": 1109 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7698, + "image_masks_loss": 0.0193, + "llm_learning_rate": 4.908679719346732e-06, + "loss": 0.8036, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004908679719346731, + "step": 1110 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7696, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.908457407493583e-06, + "loss": 0.7548, + "output_text_mask_loss": 0.0626, + "predictor_learning_rate": 0.0004908457407493583, + "step": 1111 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7694, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.908234830415741e-06, + "loss": 0.6821, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004908234830415741, + "step": 1112 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7692, + "image_masks_loss": 0.0429, + "llm_learning_rate": 4.908011988137715e-06, + "loss": 0.8817, + "output_text_mask_loss": 0.0847, + "predictor_learning_rate": 0.0004908011988137715, + "step": 1113 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7691, + "image_masks_loss": 0.0481, + "llm_learning_rate": 4.907788880684045e-06, + "loss": 0.8048, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004907788880684045, + "step": 1114 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7689, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.907565508079302e-06, + "loss": 0.8042, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004907565508079302, + "step": 1115 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7687, + "image_masks_loss": 0.0319, + "llm_learning_rate": 4.90734187034808e-06, + "loss": 0.8044, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004907341870348079, + "step": 1116 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7685, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.907117967515011e-06, + "loss": 0.804, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.000490711796751501, + "step": 1117 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7683, + "image_masks_loss": 0.0448, + "llm_learning_rate": 4.90689379960475e-06, + "loss": 0.746, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.000490689379960475, + "step": 1118 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7682, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.906669366641981e-06, + "loss": 0.7618, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0004906669366641981, + "step": 1119 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.768, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.906444668651421e-06, + "loss": 0.7622, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004906444668651421, + "step": 1120 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7678, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.906219705657815e-06, + "loss": 0.8947, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004906219705657815, + "step": 1121 + }, + { + "epoch": 0.11, + "gumbel_tau": 0.7676, + "image_masks_loss": 0.0241, + "llm_learning_rate": 4.905994477685934e-06, + "loss": 0.7512, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004905994477685935, + "step": 1122 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7674, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.905768984760583e-06, + "loss": 0.749, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.0004905768984760583, + "step": 1123 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7673, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.905543226906593e-06, + "loss": 0.7638, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004905543226906593, + "step": 1124 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7671, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.905317204148824e-06, + "loss": 0.7505, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0004905317204148824, + "step": 1125 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7669, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.905090916512167e-06, + "loss": 0.7139, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004905090916512168, + "step": 1126 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7667, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.904864364021541e-06, + "loss": 0.7739, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.000490486436402154, + "step": 1127 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7665, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.904637546701895e-06, + "loss": 0.8849, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004904637546701894, + "step": 1128 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7663, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.9044104645782066e-06, + "loss": 0.7011, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004904410464578207, + "step": 1129 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7662, + "image_masks_loss": 0.0315, + "llm_learning_rate": 4.904183117675481e-06, + "loss": 0.8161, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.000490418311767548, + "step": 1130 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.766, + "image_masks_loss": 0.0391, + "llm_learning_rate": 4.903955506018756e-06, + "loss": 0.8023, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004903955506018755, + "step": 1131 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7658, + "image_masks_loss": 0.0353, + "llm_learning_rate": 4.9037276296330965e-06, + "loss": 0.8086, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004903727629633096, + "step": 1132 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7656, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.903499488543596e-06, + "loss": 0.7569, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.0004903499488543596, + "step": 1133 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7654, + "image_masks_loss": 0.0431, + "llm_learning_rate": 4.903271082775378e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0004903271082775378, + "step": 1134 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7653, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.903042412353596e-06, + "loss": 0.7479, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004903042412353596, + "step": 1135 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7651, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.9028134773034305e-06, + "loss": 0.8502, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.000490281347730343, + "step": 1136 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7649, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.902584277650092e-06, + "loss": 0.7727, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.0004902584277650092, + "step": 1137 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7647, + "image_masks_loss": 0.0193, + "llm_learning_rate": 4.902354813418822e-06, + "loss": 0.8642, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004902354813418823, + "step": 1138 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7645, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.902125084634889e-06, + "loss": 0.8105, + "output_text_mask_loss": 0.0729, + "predictor_learning_rate": 0.0004902125084634889, + "step": 1139 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7644, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.901895091323592e-06, + "loss": 0.8409, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004901895091323591, + "step": 1140 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7642, + "image_masks_loss": 0.034, + "llm_learning_rate": 4.901664833510257e-06, + "loss": 0.7963, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004901664833510256, + "step": 1141 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.764, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.90143431122024e-06, + "loss": 0.8109, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.000490143431122024, + "step": 1142 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7638, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.901203524478929e-06, + "loss": 0.7597, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0004901203524478928, + "step": 1143 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7636, + "image_masks_loss": 0.0173, + "llm_learning_rate": 4.900972473311737e-06, + "loss": 0.7284, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004900972473311737, + "step": 1144 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7635, + "image_masks_loss": 0.0383, + "llm_learning_rate": 4.900741157744108e-06, + "loss": 0.8172, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004900741157744107, + "step": 1145 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7633, + "image_masks_loss": 0.016, + "llm_learning_rate": 4.900509577801515e-06, + "loss": 0.786, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004900509577801514, + "step": 1146 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7631, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.900277733509461e-06, + "loss": 0.738, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004900277733509461, + "step": 1147 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7629, + "image_masks_loss": 0.038, + "llm_learning_rate": 4.900045624893476e-06, + "loss": 0.8011, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004900045624893476, + "step": 1148 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7627, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.899813251979121e-06, + "loss": 0.881, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.000489981325197912, + "step": 1149 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7626, + "image_masks_loss": 0.0147, + "llm_learning_rate": 4.899580614791986e-06, + "loss": 0.7684, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004899580614791986, + "step": 1150 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7624, + "image_masks_loss": 0.022, + "llm_learning_rate": 4.899347713357688e-06, + "loss": 0.732, + "output_text_mask_loss": 0.0523, + "predictor_learning_rate": 0.0004899347713357688, + "step": 1151 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7622, + "image_masks_loss": 0.022, + "llm_learning_rate": 4.899114547701876e-06, + "loss": 0.7863, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004899114547701875, + "step": 1152 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.762, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.898881117850226e-06, + "loss": 0.8253, + "output_text_mask_loss": 0.0698, + "predictor_learning_rate": 0.0004898881117850226, + "step": 1153 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7618, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.898647423828444e-06, + "loss": 0.6214, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004898647423828443, + "step": 1154 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7617, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.898413465662265e-06, + "loss": 0.7243, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004898413465662264, + "step": 1155 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7615, + "image_masks_loss": 0.0399, + "llm_learning_rate": 4.8981792433774524e-06, + "loss": 0.7005, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004898179243377452, + "step": 1156 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7613, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.897944756999801e-06, + "loss": 0.8428, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.00048979447569998, + "step": 1157 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7611, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.89771000655513e-06, + "loss": 0.8416, + "output_text_mask_loss": 0.0227, + "predictor_learning_rate": 0.000489771000655513, + "step": 1158 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7609, + "image_masks_loss": 0.0028, + "llm_learning_rate": 4.897474992069294e-06, + "loss": 0.6967, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004897474992069294, + "step": 1159 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7608, + "image_masks_loss": 0.0205, + "llm_learning_rate": 4.89723971356817e-06, + "loss": 0.8471, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000489723971356817, + "step": 1160 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7606, + "image_masks_loss": 0.034, + "llm_learning_rate": 4.89700417107767e-06, + "loss": 0.6902, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000489700417107767, + "step": 1161 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7604, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.896768364623733e-06, + "loss": 0.8039, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004896768364623733, + "step": 1162 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7602, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.8965322942323234e-06, + "loss": 0.8218, + "output_text_mask_loss": 0.025, + "predictor_learning_rate": 0.0004896532294232323, + "step": 1163 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.76, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.89629595992944e-06, + "loss": 0.7968, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.000489629595992944, + "step": 1164 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7599, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.896059361741108e-06, + "loss": 0.7817, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004896059361741107, + "step": 1165 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7597, + "image_masks_loss": 0.024, + "llm_learning_rate": 4.895822499693381e-06, + "loss": 0.7963, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004895822499693381, + "step": 1166 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7595, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.895585373812346e-06, + "loss": 0.7986, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0004895585373812345, + "step": 1167 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7593, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.895347984124112e-06, + "loss": 0.7434, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0004895347984124112, + "step": 1168 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7592, + "image_masks_loss": 0.0319, + "llm_learning_rate": 4.895110330654823e-06, + "loss": 0.863, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004895110330654823, + "step": 1169 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.759, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.8948724134306495e-06, + "loss": 0.8022, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004894872413430649, + "step": 1170 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7588, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.894634232477792e-06, + "loss": 0.7855, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004894634232477791, + "step": 1171 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7586, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.894395787822478e-06, + "loss": 0.8221, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004894395787822478, + "step": 1172 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7584, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.894157079490967e-06, + "loss": 0.7409, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.0004894157079490967, + "step": 1173 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7583, + "image_masks_loss": 0.0086, + "llm_learning_rate": 4.893918107509546e-06, + "loss": 0.8053, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004893918107509545, + "step": 1174 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7581, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.89367887190453e-06, + "loss": 0.7906, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.000489367887190453, + "step": 1175 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7579, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.893439372702265e-06, + "loss": 0.8097, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004893439372702265, + "step": 1176 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7577, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.8931996099291254e-06, + "loss": 0.8149, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004893199609929125, + "step": 1177 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7575, + "image_masks_loss": 0.0322, + "llm_learning_rate": 4.892959583611514e-06, + "loss": 0.7333, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004892959583611513, + "step": 1178 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7574, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.892719293775864e-06, + "loss": 0.8674, + "output_text_mask_loss": 0.0725, + "predictor_learning_rate": 0.0004892719293775864, + "step": 1179 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7572, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.892478740448635e-06, + "loss": 0.8304, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004892478740448634, + "step": 1180 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.757, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.8922379236563175e-06, + "loss": 0.7462, + "output_text_mask_loss": 0.0755, + "predictor_learning_rate": 0.0004892237923656318, + "step": 1181 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7568, + "image_masks_loss": 0.0315, + "llm_learning_rate": 4.891996843425432e-06, + "loss": 0.8361, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004891996843425432, + "step": 1182 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7566, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.891755499782527e-06, + "loss": 0.8205, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004891755499782526, + "step": 1183 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7565, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.8915138927541775e-06, + "loss": 0.7453, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004891513892754177, + "step": 1184 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7563, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.891272022366992e-06, + "loss": 0.6413, + "output_text_mask_loss": 0.0507, + "predictor_learning_rate": 0.0004891272022366992, + "step": 1185 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7561, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.891029888647605e-06, + "loss": 0.7682, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.0004891029888647604, + "step": 1186 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7559, + "image_masks_loss": 0.0173, + "llm_learning_rate": 4.8907874916226805e-06, + "loss": 0.7514, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000489078749162268, + "step": 1187 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7558, + "image_masks_loss": 0.0393, + "llm_learning_rate": 4.890544831318912e-06, + "loss": 0.778, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004890544831318912, + "step": 1188 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7556, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.890301907763022e-06, + "loss": 0.7903, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004890301907763022, + "step": 1189 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7554, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.890058720981762e-06, + "loss": 0.8148, + "output_text_mask_loss": 0.0793, + "predictor_learning_rate": 0.0004890058720981762, + "step": 1190 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7552, + "image_masks_loss": 0.0241, + "llm_learning_rate": 4.889815271001912e-06, + "loss": 0.8048, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004889815271001912, + "step": 1191 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.755, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.889571557850281e-06, + "loss": 0.8346, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.000488957155785028, + "step": 1192 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7549, + "image_masks_loss": 0.0049, + "llm_learning_rate": 4.889327581553709e-06, + "loss": 0.7939, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 0.0004889327581553708, + "step": 1193 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7547, + "image_masks_loss": 0.0193, + "llm_learning_rate": 4.889083342139059e-06, + "loss": 0.8616, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004889083342139059, + "step": 1194 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7545, + "image_masks_loss": 0.0441, + "llm_learning_rate": 4.888838839633232e-06, + "loss": 0.7362, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004888838839633232, + "step": 1195 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7543, + "image_masks_loss": 0.0328, + "llm_learning_rate": 4.88859407406315e-06, + "loss": 0.8226, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.000488859407406315, + "step": 1196 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7542, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.888349045455769e-06, + "loss": 0.7102, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.0004888349045455768, + "step": 1197 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.754, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.888103753838071e-06, + "loss": 0.8289, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.000488810375383807, + "step": 1198 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7538, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.8878581992370675e-06, + "loss": 0.8528, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0004887858199237068, + "step": 1199 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7536, + "image_masks_loss": 0.061, + "llm_learning_rate": 4.887612381679802e-06, + "loss": 0.6867, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004887612381679801, + "step": 1200 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7534, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.887366301193342e-06, + "loss": 0.8225, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004887366301193341, + "step": 1201 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7533, + "image_masks_loss": 0.0041, + "llm_learning_rate": 4.887119957804787e-06, + "loss": 0.7508, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004887119957804787, + "step": 1202 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7531, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.886873351541264e-06, + "loss": 0.7107, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0004886873351541265, + "step": 1203 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7529, + "image_masks_loss": 0.0069, + "llm_learning_rate": 4.886626482429933e-06, + "loss": 0.7994, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004886626482429932, + "step": 1204 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7527, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.886379350497978e-06, + "loss": 0.7385, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 0.0004886379350497977, + "step": 1205 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7526, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.8861319557726115e-06, + "loss": 0.8102, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 0.0004886131955772611, + "step": 1206 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7524, + "image_masks_loss": 0.0158, + "llm_learning_rate": 4.88588429828108e-06, + "loss": 0.8492, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000488588429828108, + "step": 1207 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7522, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.885636378050656e-06, + "loss": 0.7765, + "output_text_mask_loss": 0.0633, + "predictor_learning_rate": 0.0004885636378050655, + "step": 1208 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.752, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.885388195108639e-06, + "loss": 0.84, + "output_text_mask_loss": 0.0473, + "predictor_learning_rate": 0.0004885388195108639, + "step": 1209 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7518, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.885139749482361e-06, + "loss": 0.7903, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004885139749482361, + "step": 1210 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7517, + "image_masks_loss": 0.0359, + "llm_learning_rate": 4.884891041199181e-06, + "loss": 0.8944, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004884891041199181, + "step": 1211 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7515, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.884642070286489e-06, + "loss": 0.8159, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004884642070286488, + "step": 1212 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7513, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.884392836771699e-06, + "loss": 0.8286, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.0004884392836771699, + "step": 1213 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7511, + "image_masks_loss": 0.0429, + "llm_learning_rate": 4.884143340682259e-06, + "loss": 0.7962, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004884143340682259, + "step": 1214 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.751, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.883893582045644e-06, + "loss": 0.7771, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.0004883893582045644, + "step": 1215 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7508, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.883643560889358e-06, + "loss": 0.8217, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0004883643560889357, + "step": 1216 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7506, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.883393277240933e-06, + "loss": 0.6504, + "output_text_mask_loss": 0.1076, + "predictor_learning_rate": 0.0004883393277240933, + "step": 1217 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7504, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.8831427311279325e-06, + "loss": 0.8591, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 0.0004883142731127932, + "step": 1218 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7502, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.882891922577946e-06, + "loss": 0.8293, + "output_text_mask_loss": 0.0257, + "predictor_learning_rate": 0.0004882891922577946, + "step": 1219 + }, + { + "epoch": 0.12, + "gumbel_tau": 0.7501, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.882640851618593e-06, + "loss": 0.6726, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00048826408516185925, + "step": 1220 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7499, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.882389518277523e-06, + "loss": 0.7991, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00048823895182775225, + "step": 1221 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7497, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.882137922582412e-06, + "loss": 0.805, + "output_text_mask_loss": 0.0542, + "predictor_learning_rate": 0.00048821379225824114, + "step": 1222 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7495, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.881886064560967e-06, + "loss": 0.7342, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004881886064560966, + "step": 1223 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7494, + "image_masks_loss": 0.0345, + "llm_learning_rate": 4.881633944240923e-06, + "loss": 0.6771, + "output_text_mask_loss": 0.0523, + "predictor_learning_rate": 0.00048816339442409225, + "step": 1224 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7492, + "image_masks_loss": 0.058, + "llm_learning_rate": 4.8813815616500435e-06, + "loss": 0.7802, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 0.0004881381561650043, + "step": 1225 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.749, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.881128916816124e-06, + "loss": 0.7403, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004881128916816123, + "step": 1226 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7488, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.880876009766983e-06, + "loss": 0.7197, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048808760097669826, + "step": 1227 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7487, + "image_masks_loss": 0.0233, + "llm_learning_rate": 4.880622840530473e-06, + "loss": 0.7853, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048806228405304724, + "step": 1228 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7485, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.8803694091344735e-06, + "loss": 0.7867, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004880369409134473, + "step": 1229 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7483, + "image_masks_loss": 0.0079, + "llm_learning_rate": 4.880115715606892e-06, + "loss": 0.8314, + "output_text_mask_loss": 0.004, + "predictor_learning_rate": 0.0004880115715606892, + "step": 1230 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7481, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.879861759975666e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004879861759975666, + "step": 1231 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.748, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.879607542268763e-06, + "loss": 0.8667, + "output_text_mask_loss": 0.1099, + "predictor_learning_rate": 0.0004879607542268763, + "step": 1232 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7478, + "image_masks_loss": 0.0311, + "llm_learning_rate": 4.879353062514177e-06, + "loss": 0.7249, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048793530625141767, + "step": 1233 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7476, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.879098320739933e-06, + "loss": 0.7802, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004879098320739932, + "step": 1234 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7474, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.87884331697408e-06, + "loss": 0.8276, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.000487884331697408, + "step": 1235 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7472, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.878588051244704e-06, + "loss": 0.7336, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004878588051244703, + "step": 1236 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7471, + "image_masks_loss": 0.0463, + "llm_learning_rate": 4.878332523579913e-06, + "loss": 0.8903, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.0004878332523579913, + "step": 1237 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7469, + "image_masks_loss": 0.024, + "llm_learning_rate": 4.878076734007848e-06, + "loss": 0.7405, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048780767340078483, + "step": 1238 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7467, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.8778206825566755e-06, + "loss": 0.7342, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048778206825566753, + "step": 1239 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7465, + "image_masks_loss": 0.0412, + "llm_learning_rate": 4.877564369254593e-06, + "loss": 0.7767, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048775643692545924, + "step": 1240 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7464, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.877307794129826e-06, + "loss": 0.8274, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0004877307794129826, + "step": 1241 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7462, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.87705095721063e-06, + "loss": 0.8502, + "output_text_mask_loss": 0.0633, + "predictor_learning_rate": 0.000487705095721063, + "step": 1242 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.746, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.876793858525288e-06, + "loss": 0.9114, + "output_text_mask_loss": 0.177, + "predictor_learning_rate": 0.0004876793858525288, + "step": 1243 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7458, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.876536498102112e-06, + "loss": 0.6677, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.0004876536498102112, + "step": 1244 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7457, + "image_masks_loss": 0.0389, + "llm_learning_rate": 4.8762788759694435e-06, + "loss": 0.7238, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0004876278875969443, + "step": 1245 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7455, + "image_masks_loss": 0.0209, + "llm_learning_rate": 4.876020992155652e-06, + "loss": 0.7848, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00048760209921556514, + "step": 1246 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7453, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.875762846689136e-06, + "loss": 0.785, + "output_text_mask_loss": 0.0507, + "predictor_learning_rate": 0.0004875762846689136, + "step": 1247 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7451, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.8755044395983244e-06, + "loss": 0.6388, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004875504439598324, + "step": 1248 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.745, + "image_masks_loss": 0.0401, + "llm_learning_rate": 4.875245770911672e-06, + "loss": 0.8005, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.00048752457709116715, + "step": 1249 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7448, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.874986840657665e-06, + "loss": 0.7112, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004874986840657665, + "step": 1250 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7446, + "image_masks_loss": 0.0387, + "llm_learning_rate": 4.874727648864817e-06, + "loss": 0.7878, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004874727648864816, + "step": 1251 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7444, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.8744681955616705e-06, + "loss": 0.7706, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000487446819556167, + "step": 1252 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7443, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.874208480776798e-06, + "loss": 0.7893, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004874208480776797, + "step": 1253 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7441, + "image_masks_loss": 0.0349, + "llm_learning_rate": 4.873948504538798e-06, + "loss": 0.8407, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.00048739485045387983, + "step": 1254 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7439, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.873688266876302e-06, + "loss": 0.7886, + "output_text_mask_loss": 0.0725, + "predictor_learning_rate": 0.00048736882668763015, + "step": 1255 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7437, + "image_masks_loss": 0.0399, + "llm_learning_rate": 4.873427767817967e-06, + "loss": 0.788, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004873427767817967, + "step": 1256 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7436, + "image_masks_loss": 0.0061, + "llm_learning_rate": 4.873167007392479e-06, + "loss": 0.7867, + "output_text_mask_loss": 0.0408, + "predictor_learning_rate": 0.0004873167007392479, + "step": 1257 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7434, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.872905985628555e-06, + "loss": 0.8016, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004872905985628554, + "step": 1258 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7432, + "image_masks_loss": 0.0186, + "llm_learning_rate": 4.872644702554937e-06, + "loss": 0.8208, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.0004872644702554937, + "step": 1259 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.743, + "image_masks_loss": 0.0483, + "llm_learning_rate": 4.8723831582004015e-06, + "loss": 0.814, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048723831582004007, + "step": 1260 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7429, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.872121352593746e-06, + "loss": 0.8598, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 0.00048721213525937455, + "step": 1261 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7427, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.871859285763805e-06, + "loss": 0.7568, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00048718592857638045, + "step": 1262 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7425, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.871596957739436e-06, + "loss": 0.8337, + "output_text_mask_loss": 0.09, + "predictor_learning_rate": 0.0004871596957739436, + "step": 1263 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7423, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.8713343685495274e-06, + "loss": 0.8093, + "output_text_mask_loss": 0.0629, + "predictor_learning_rate": 0.0004871334368549527, + "step": 1264 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7422, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.871071518222996e-06, + "loss": 0.7271, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0004871071518222996, + "step": 1265 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.742, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.870808406788789e-06, + "loss": 0.7658, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004870808406788788, + "step": 1266 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7418, + "image_masks_loss": 0.0234, + "llm_learning_rate": 4.870545034275878e-06, + "loss": 0.8567, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004870545034275877, + "step": 1267 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7416, + "image_masks_loss": 0.0322, + "llm_learning_rate": 4.870281400713268e-06, + "loss": 0.8057, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.00048702814007132674, + "step": 1268 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7415, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.8700175061299905e-06, + "loss": 0.7415, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000487001750612999, + "step": 1269 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7413, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.8697533505551055e-06, + "loss": 0.8222, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004869753350555106, + "step": 1270 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7411, + "image_masks_loss": 0.0217, + "llm_learning_rate": 4.869488934017704e-06, + "loss": 0.6937, + "output_text_mask_loss": 0.0362, + "predictor_learning_rate": 0.0004869488934017704, + "step": 1271 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7409, + "image_masks_loss": 0.0084, + "llm_learning_rate": 4.8692242565469036e-06, + "loss": 0.7244, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.00048692242565469035, + "step": 1272 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7408, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.868959318171851e-06, + "loss": 0.8755, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000486895931817185, + "step": 1273 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7406, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.868694118921721e-06, + "loss": 0.8272, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.000486869411892172, + "step": 1274 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7404, + "image_masks_loss": 0.0431, + "llm_learning_rate": 4.868428658825718e-06, + "loss": 0.7726, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004868428658825718, + "step": 1275 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7402, + "image_masks_loss": 0.0546, + "llm_learning_rate": 4.868162937913076e-06, + "loss": 0.8049, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004868162937913076, + "step": 1276 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7401, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.867896956213057e-06, + "loss": 0.7677, + "output_text_mask_loss": 0.0288, + "predictor_learning_rate": 0.0004867896956213056, + "step": 1277 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7399, + "image_masks_loss": 0.0538, + "llm_learning_rate": 4.867630713754951e-06, + "loss": 0.7834, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.000486763071375495, + "step": 1278 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7397, + "image_masks_loss": 0.0357, + "llm_learning_rate": 4.867364210568076e-06, + "loss": 0.6868, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.0004867364210568076, + "step": 1279 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7395, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.867097446681782e-06, + "loss": 0.8013, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00048670974466817813, + "step": 1280 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7394, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.866830422125444e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00048668304221254435, + "step": 1281 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7392, + "image_masks_loss": 0.069, + "llm_learning_rate": 4.866563136928469e-06, + "loss": 0.7997, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 0.0004866563136928468, + "step": 1282 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.739, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.866295591120289e-06, + "loss": 0.7949, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00048662955911202884, + "step": 1283 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7388, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.866027784730367e-06, + "loss": 0.7366, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004866027784730367, + "step": 1284 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7387, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.8657597177881976e-06, + "loss": 0.7096, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.00048657597177881974, + "step": 1285 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7385, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.865491390323297e-06, + "loss": 0.8536, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.00048654913903232967, + "step": 1286 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7383, + "image_masks_loss": 0.0173, + "llm_learning_rate": 4.865222802365216e-06, + "loss": 0.734, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048652228023652157, + "step": 1287 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7381, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.864953953943532e-06, + "loss": 0.736, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.00048649539539435315, + "step": 1288 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.738, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.86468484508785e-06, + "loss": 0.7455, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000486468484508785, + "step": 1289 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7378, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.864415475827806e-06, + "loss": 0.754, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0004864415475827806, + "step": 1290 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7376, + "image_masks_loss": 0.0744, + "llm_learning_rate": 4.864145846193064e-06, + "loss": 0.8673, + "output_text_mask_loss": 0.0923, + "predictor_learning_rate": 0.0004864145846193063, + "step": 1291 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7374, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.863875956213315e-06, + "loss": 0.7577, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048638759562133146, + "step": 1292 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7373, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.86360580591828e-06, + "loss": 0.8342, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00048636058059182795, + "step": 1293 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7371, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.86333539533771e-06, + "loss": 0.8206, + "output_text_mask_loss": 0.0138, + "predictor_learning_rate": 0.0004863335395337709, + "step": 1294 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7369, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.86306472450138e-06, + "loss": 0.7937, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00048630647245013803, + "step": 1295 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7367, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.862793793439101e-06, + "loss": 0.7982, + "output_text_mask_loss": 0.0317, + "predictor_learning_rate": 0.00048627937934391007, + "step": 1296 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7366, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.862522602180706e-06, + "loss": 0.8744, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00048625226021807057, + "step": 1297 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7364, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.86225115075606e-06, + "loss": 0.8694, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00048622511507560596, + "step": 1298 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7362, + "image_masks_loss": 0.0053, + "llm_learning_rate": 4.861979439195055e-06, + "loss": 0.7358, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.00048619794391950545, + "step": 1299 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.736, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.861707467527613e-06, + "loss": 0.7334, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.0004861707467527613, + "step": 1300 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7359, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.861435235783685e-06, + "loss": 0.7184, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00048614352357836846, + "step": 1301 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7357, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.861162743993249e-06, + "loss": 0.7492, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0004861162743993248, + "step": 1302 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7355, + "image_masks_loss": 0.0195, + "llm_learning_rate": 4.860889992186311e-06, + "loss": 0.8128, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004860889992186311, + "step": 1303 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7354, + "image_masks_loss": 0.0385, + "llm_learning_rate": 4.86061698039291e-06, + "loss": 0.8039, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 0.0004860616980392909, + "step": 1304 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7352, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.860343708643107e-06, + "loss": 0.7897, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00048603437086431066, + "step": 1305 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.735, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.860070176966999e-06, + "loss": 0.6783, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00048600701769669984, + "step": 1306 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7348, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.859796385394706e-06, + "loss": 0.7822, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.0004859796385394706, + "step": 1307 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7347, + "image_masks_loss": 0.0355, + "llm_learning_rate": 4.859522333956378e-06, + "loss": 0.8326, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00048595223339563783, + "step": 1308 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7345, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.859248022682196e-06, + "loss": 0.7612, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00048592480226821957, + "step": 1309 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7343, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.8589734516023666e-06, + "loss": 0.7815, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.00048589734516023666, + "step": 1310 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7341, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.858698620747126e-06, + "loss": 0.8153, + "output_text_mask_loss": 0.1572, + "predictor_learning_rate": 0.00048586986207471255, + "step": 1311 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.734, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.8584235301467395e-06, + "loss": 0.7906, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004858423530146739, + "step": 1312 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7338, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.8581481798315006e-06, + "loss": 0.7272, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00048581481798315, + "step": 1313 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7336, + "image_masks_loss": 0.017, + "llm_learning_rate": 4.857872569831732e-06, + "loss": 0.8174, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004857872569831732, + "step": 1314 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7334, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.857596700177784e-06, + "loss": 0.7755, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.00048575967001777834, + "step": 1315 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7333, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.857320570900035e-06, + "loss": 0.7042, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00048573205709000346, + "step": 1316 + }, + { + "epoch": 0.13, + "gumbel_tau": 0.7331, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.857044182028895e-06, + "loss": 0.7014, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048570441820288947, + "step": 1317 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7329, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.8567675335948e-06, + "loss": 0.7733, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00048567675335947993, + "step": 1318 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7328, + "image_masks_loss": 0.0233, + "llm_learning_rate": 4.856490625628213e-06, + "loss": 0.7625, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048564906256282124, + "step": 1319 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7326, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.856213458159631e-06, + "loss": 0.7762, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048562134581596304, + "step": 1320 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7324, + "image_masks_loss": 0.0584, + "llm_learning_rate": 4.855936031219573e-06, + "loss": 0.8456, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 0.0004855936031219573, + "step": 1321 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7322, + "image_masks_loss": 0.0102, + "llm_learning_rate": 4.8556583448385936e-06, + "loss": 0.7863, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 0.0004855658344838593, + "step": 1322 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7321, + "image_masks_loss": 0.0046, + "llm_learning_rate": 4.855380399047269e-06, + "loss": 0.688, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.0004855380399047269, + "step": 1323 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7319, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.855102193876209e-06, + "loss": 0.8324, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00048551021938762084, + "step": 1324 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7317, + "image_masks_loss": 0.0383, + "llm_learning_rate": 4.85482372935605e-06, + "loss": 0.8114, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00048548237293560495, + "step": 1325 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7315, + "image_masks_loss": 0.042, + "llm_learning_rate": 4.854545005517456e-06, + "loss": 0.7757, + "output_text_mask_loss": 0.0626, + "predictor_learning_rate": 0.0004854545005517456, + "step": 1326 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7314, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.854266022391122e-06, + "loss": 0.7757, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.00048542660223911216, + "step": 1327 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7312, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.8539867800077685e-06, + "loss": 0.6208, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00048539867800077684, + "step": 1328 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.731, + "image_masks_loss": 0.0105, + "llm_learning_rate": 4.853707278398149e-06, + "loss": 0.7614, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00048537072783981484, + "step": 1329 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7309, + "image_masks_loss": 0.0152, + "llm_learning_rate": 4.8534275175930405e-06, + "loss": 0.7879, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00048534275175930405, + "step": 1330 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7307, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.853147497623252e-06, + "loss": 0.775, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.0004853147497623251, + "step": 1331 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7305, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.85286721851962e-06, + "loss": 0.723, + "output_text_mask_loss": 0.1335, + "predictor_learning_rate": 0.0004852867218519619, + "step": 1332 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7303, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.852586680313008e-06, + "loss": 0.8317, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004852586680313008, + "step": 1333 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7302, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.852305883034312e-06, + "loss": 0.7991, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048523058830343116, + "step": 1334 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.73, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.852024826714451e-06, + "loss": 0.7464, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004852024826714451, + "step": 1335 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7298, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.851743511384378e-06, + "loss": 0.7922, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00048517435113843777, + "step": 1336 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7297, + "image_masks_loss": 0.0108, + "llm_learning_rate": 4.851461937075071e-06, + "loss": 0.7483, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0004851461937075071, + "step": 1337 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7295, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.851180103817539e-06, + "loss": 0.7645, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.0004851180103817539, + "step": 1338 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7293, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.850898011642815e-06, + "loss": 0.6461, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00048508980116428147, + "step": 1339 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7291, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.850615660581967e-06, + "loss": 0.7091, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.00048506156605819664, + "step": 1340 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.729, + "image_masks_loss": 0.0687, + "llm_learning_rate": 4.850333050666087e-06, + "loss": 0.7775, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004850333050666086, + "step": 1341 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7288, + "image_masks_loss": 0.0026, + "llm_learning_rate": 4.850050181926295e-06, + "loss": 0.733, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0004850050181926294, + "step": 1342 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7286, + "image_masks_loss": 0.0393, + "llm_learning_rate": 4.849767054393742e-06, + "loss": 0.7809, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004849767054393742, + "step": 1343 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7284, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.849483668099608e-06, + "loss": 0.7905, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.0004849483668099608, + "step": 1344 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7283, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.8492000230750995e-06, + "loss": 0.7423, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004849200023075099, + "step": 1345 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7281, + "image_masks_loss": 0.0028, + "llm_learning_rate": 4.848916119351451e-06, + "loss": 0.6543, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00048489161193514507, + "step": 1346 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7279, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.848631956959928e-06, + "loss": 0.7274, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004848631956959928, + "step": 1347 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7278, + "image_masks_loss": 0.0486, + "llm_learning_rate": 4.848347535931823e-06, + "loss": 0.8815, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00048483475359318223, + "step": 1348 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7276, + "image_masks_loss": 0.0288, + "llm_learning_rate": 4.848062856298456e-06, + "loss": 0.8079, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0004848062856298455, + "step": 1349 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7274, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.8477779180911765e-06, + "loss": 0.8048, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0004847777918091176, + "step": 1350 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7272, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.847492721341364e-06, + "loss": 0.7997, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004847492721341364, + "step": 1351 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7271, + "image_masks_loss": 0.05, + "llm_learning_rate": 4.847207266080424e-06, + "loss": 0.7709, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 0.0004847207266080424, + "step": 1352 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7269, + "image_masks_loss": 0.0404, + "llm_learning_rate": 4.846921552339792e-06, + "loss": 0.8517, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.00048469215523397915, + "step": 1353 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7267, + "image_masks_loss": 0.0045, + "llm_learning_rate": 4.8466355801509305e-06, + "loss": 0.8602, + "output_text_mask_loss": 0.0885, + "predictor_learning_rate": 0.00048466355801509304, + "step": 1354 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7266, + "image_masks_loss": 0.0052, + "llm_learning_rate": 4.846349349545333e-06, + "loss": 0.815, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048463493495453324, + "step": 1355 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7264, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.846062860554518e-06, + "loss": 0.7392, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0004846062860554518, + "step": 1356 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7262, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.845776113210035e-06, + "loss": 0.8038, + "output_text_mask_loss": 0.0418, + "predictor_learning_rate": 0.00048457761132100345, + "step": 1357 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.726, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.8454891075434614e-06, + "loss": 0.7575, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 0.00048454891075434614, + "step": 1358 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7259, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.845201843586403e-06, + "loss": 0.6966, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00048452018435864034, + "step": 1359 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7257, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.8449143213704945e-06, + "loss": 0.656, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.0004844914321370494, + "step": 1360 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7255, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.844626540927397e-06, + "loss": 0.7142, + "output_text_mask_loss": 0.0423, + "predictor_learning_rate": 0.0004844626540927397, + "step": 1361 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7254, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.844338502288803e-06, + "loss": 0.6796, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.0004844338502288803, + "step": 1362 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7252, + "image_masks_loss": 0.0039, + "llm_learning_rate": 4.844050205486431e-06, + "loss": 0.7315, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004844050205486431, + "step": 1363 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.725, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.843761650552029e-06, + "loss": 0.7955, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00048437616505520286, + "step": 1364 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7248, + "image_masks_loss": 0.0519, + "llm_learning_rate": 4.843472837517374e-06, + "loss": 0.7925, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00048434728375173743, + "step": 1365 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7247, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.84318376641427e-06, + "loss": 0.6844, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000484318376641427, + "step": 1366 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7245, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.84289443727455e-06, + "loss": 0.8038, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00048428944372745506, + "step": 1367 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7243, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.842604850130077e-06, + "loss": 0.8059, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.0004842604850130077, + "step": 1368 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7242, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.84231500501274e-06, + "loss": 0.8314, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 0.00048423150050127395, + "step": 1369 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.724, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.842024901954456e-06, + "loss": 0.758, + "output_text_mask_loss": 0.0938, + "predictor_learning_rate": 0.0004842024901954456, + "step": 1370 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7238, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.841734540987174e-06, + "loss": 0.8088, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00048417345409871736, + "step": 1371 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7237, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.841443922142868e-06, + "loss": 0.731, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004841443922142868, + "step": 1372 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7235, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.841153045453543e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00048411530454535424, + "step": 1373 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7233, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.840861910951228e-06, + "loss": 0.6668, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0004840861910951228, + "step": 1374 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7231, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.840570518667987e-06, + "loss": 0.787, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00048405705186679865, + "step": 1375 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.723, + "image_masks_loss": 0.0056, + "llm_learning_rate": 4.840278868635906e-06, + "loss": 0.6518, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004840278868635906, + "step": 1376 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7228, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.839986960887104e-06, + "loss": 0.7663, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0004839986960887104, + "step": 1377 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7226, + "image_masks_loss": 0.0058, + "llm_learning_rate": 4.839694795453725e-06, + "loss": 0.8295, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00048396947954537254, + "step": 1378 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7225, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.839402372367944e-06, + "loss": 0.7221, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00048394023723679437, + "step": 1379 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7223, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.839109691661963e-06, + "loss": 0.8408, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.00048391096916619624, + "step": 1380 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7221, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.838816753368013e-06, + "loss": 0.8592, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 0.00048388167533680126, + "step": 1381 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7219, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.838523557518352e-06, + "loss": 0.7416, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048385235575183516, + "step": 1382 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7218, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.838230104145268e-06, + "loss": 0.6973, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.0004838230104145268, + "step": 1383 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7216, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.837936393281078e-06, + "loss": 0.654, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004837936393281077, + "step": 1384 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7214, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.837642424958123e-06, + "loss": 0.8026, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00048376424249581225, + "step": 1385 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7213, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.837348199208778e-06, + "loss": 0.9071, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.00048373481992087776, + "step": 1386 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7211, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.837053716065443e-06, + "loss": 0.7167, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.0004837053716065443, + "step": 1387 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7209, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.836758975560547e-06, + "loss": 0.7193, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004836758975560547, + "step": 1388 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7208, + "image_masks_loss": 0.0236, + "llm_learning_rate": 4.836463977726548e-06, + "loss": 0.7144, + "output_text_mask_loss": 0.0462, + "predictor_learning_rate": 0.0004836463977726547, + "step": 1389 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7206, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.836168722595931e-06, + "loss": 0.8881, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004836168722595931, + "step": 1390 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7204, + "image_masks_loss": 0.0213, + "llm_learning_rate": 4.8358732102012116e-06, + "loss": 0.8104, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.00048358732102012113, + "step": 1391 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7202, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.8355774405749315e-06, + "loss": 0.8582, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004835577440574931, + "step": 1392 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7201, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.83528141374966e-06, + "loss": 0.7752, + "output_text_mask_loss": 0.053, + "predictor_learning_rate": 0.000483528141374966, + "step": 1393 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7199, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.834985129757999e-06, + "loss": 0.8622, + "output_text_mask_loss": 0.0296, + "predictor_learning_rate": 0.0004834985129757998, + "step": 1394 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7197, + "image_masks_loss": 0.0241, + "llm_learning_rate": 4.834688588632574e-06, + "loss": 0.7656, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004834688588632573, + "step": 1395 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7196, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.83439179040604e-06, + "loss": 0.7408, + "output_text_mask_loss": 0.0196, + "predictor_learning_rate": 0.00048343917904060396, + "step": 1396 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7194, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.834094735111083e-06, + "loss": 0.719, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004834094735111083, + "step": 1397 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7192, + "image_masks_loss": 0.0389, + "llm_learning_rate": 4.833797422780416e-06, + "loss": 0.8549, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004833797422780416, + "step": 1398 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7191, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.833499853446778e-06, + "loss": 0.8251, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.00048334998534467775, + "step": 1399 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7189, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.833202027142938e-06, + "loss": 0.7597, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004833202027142938, + "step": 1400 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7187, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.832903943901694e-06, + "loss": 0.8685, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 0.0004832903943901694, + "step": 1401 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7185, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.832605603755871e-06, + "loss": 0.7914, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004832605603755871, + "step": 1402 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7184, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.8323070067383236e-06, + "loss": 0.8091, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.00048323070067383234, + "step": 1403 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7182, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.832008152881934e-06, + "loss": 0.8421, + "output_text_mask_loss": 0.0351, + "predictor_learning_rate": 0.0004832008152881934, + "step": 1404 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.718, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.831709042219611e-06, + "loss": 0.8256, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00048317090422196114, + "step": 1405 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7179, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.831409674784297e-06, + "loss": 0.6956, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048314096747842963, + "step": 1406 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7177, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.831110050608954e-06, + "loss": 0.6776, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0004831110050608954, + "step": 1407 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7175, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.830810169726581e-06, + "loss": 0.7335, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00048308101697265804, + "step": 1408 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7174, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.8305100321702e-06, + "loss": 0.738, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00048305100321702, + "step": 1409 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7172, + "image_masks_loss": 0.0186, + "llm_learning_rate": 4.830209637972864e-06, + "loss": 0.7604, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00048302096379728636, + "step": 1410 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.717, + "image_masks_loss": 0.004, + "llm_learning_rate": 4.829908987167652e-06, + "loss": 0.6886, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00048299089871676516, + "step": 1411 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7169, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.8296080797876725e-06, + "loss": 0.7961, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 0.0004829608079787672, + "step": 1412 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7167, + "image_masks_loss": 0.0213, + "llm_learning_rate": 4.829306915866062e-06, + "loss": 0.8349, + "output_text_mask_loss": 0.0946, + "predictor_learning_rate": 0.0004829306915866062, + "step": 1413 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7165, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.8290054954359865e-06, + "loss": 0.8069, + "output_text_mask_loss": 0.0362, + "predictor_learning_rate": 0.00048290054954359864, + "step": 1414 + }, + { + "epoch": 0.14, + "gumbel_tau": 0.7163, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.8287038185306375e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00048287038185306375, + "step": 1415 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7162, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.828401885183237e-06, + "loss": 0.742, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004828401885183237, + "step": 1416 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.716, + "image_masks_loss": 0.0118, + "llm_learning_rate": 4.828099695427035e-06, + "loss": 0.8646, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00048280996954270345, + "step": 1417 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7158, + "image_masks_loss": 0.0211, + "llm_learning_rate": 4.827797249295308e-06, + "loss": 0.7442, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.0004827797249295308, + "step": 1418 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7157, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.827494546821364e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004827494546821364, + "step": 1419 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7155, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.8271915880385355e-06, + "loss": 0.8462, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00048271915880385353, + "step": 1420 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7153, + "image_masks_loss": 0.0429, + "llm_learning_rate": 4.826888372980187e-06, + "loss": 0.8531, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0004826888372980186, + "step": 1421 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7152, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.8265849016797065e-06, + "loss": 0.8045, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00048265849016797057, + "step": 1422 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.715, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.826281174170515e-06, + "loss": 0.769, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004826281174170515, + "step": 1423 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7148, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.8259771904860585e-06, + "loss": 0.8039, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.00048259771904860584, + "step": 1424 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7147, + "image_masks_loss": 0.0198, + "llm_learning_rate": 4.825672950659813e-06, + "loss": 0.6872, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004825672950659813, + "step": 1425 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7145, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.825368454725282e-06, + "loss": 0.7941, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0004825368454725282, + "step": 1426 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7143, + "image_masks_loss": 0.0067, + "llm_learning_rate": 4.825063702715997e-06, + "loss": 0.8064, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00048250637027159974, + "step": 1427 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7142, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.824758694665519e-06, + "loss": 0.7369, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.0004824758694665519, + "step": 1428 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.714, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.824453430607435e-06, + "loss": 0.7404, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004824453430607435, + "step": 1429 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7138, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.824147910575362e-06, + "loss": 0.7599, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.0004824147910575362, + "step": 1430 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7136, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.8238421346029435e-06, + "loss": 0.7834, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048238421346029434, + "step": 1431 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7135, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.8235361027238536e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00048235361027238534, + "step": 1432 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7133, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.823229814971792e-06, + "loss": 0.7552, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0004823229814971792, + "step": 1433 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7131, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.822923271380488e-06, + "loss": 0.7412, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00048229232713804885, + "step": 1434 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.713, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.822616471983701e-06, + "loss": 0.8637, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004822616471983701, + "step": 1435 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7128, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.822309416815213e-06, + "loss": 0.7884, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004822309416815213, + "step": 1436 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7126, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.82200210590884e-06, + "loss": 0.8211, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.000482200210590884, + "step": 1437 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7125, + "image_masks_loss": 0.0086, + "llm_learning_rate": 4.821694539298423e-06, + "loss": 0.6899, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00048216945392984224, + "step": 1438 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7123, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.821386717017832e-06, + "loss": 0.7892, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004821386717017832, + "step": 1439 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7121, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.821078639100964e-06, + "loss": 0.7646, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.0004821078639100964, + "step": 1440 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.712, + "image_masks_loss": 0.0311, + "llm_learning_rate": 4.820770305581748e-06, + "loss": 0.7055, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.0004820770305581748, + "step": 1441 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7118, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.820461716494136e-06, + "loss": 0.7918, + "output_text_mask_loss": 0.1633, + "predictor_learning_rate": 0.00048204617164941357, + "step": 1442 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7116, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.820152871872111e-06, + "loss": 0.7256, + "output_text_mask_loss": 0.041, + "predictor_learning_rate": 0.00048201528718721105, + "step": 1443 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7115, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.819843771749684e-06, + "loss": 0.8244, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00048198437717496837, + "step": 1444 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7113, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.819534416160893e-06, + "loss": 0.726, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004819534416160892, + "step": 1445 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7111, + "image_masks_loss": 0.0198, + "llm_learning_rate": 4.819224805139806e-06, + "loss": 0.743, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 0.00048192248051398057, + "step": 1446 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.711, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.818914938720518e-06, + "loss": 0.7206, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00048189149387205173, + "step": 1447 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7108, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.818604816937151e-06, + "loss": 0.8297, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00048186048169371505, + "step": 1448 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7106, + "image_masks_loss": 0.0052, + "llm_learning_rate": 4.818294439823857e-06, + "loss": 0.8269, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.0004818294439823857, + "step": 1449 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7105, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.8179838074148154e-06, + "loss": 0.6648, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0004817983807414815, + "step": 1450 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7103, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.817672919744235e-06, + "loss": 0.7675, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00048176729197442345, + "step": 1451 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7101, + "image_masks_loss": 0.0402, + "llm_learning_rate": 4.8173617768463495e-06, + "loss": 0.8019, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.0004817361776846349, + "step": 1452 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.71, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.817050378755423e-06, + "loss": 0.8454, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 0.0004817050378755423, + "step": 1453 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7098, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.816738725505749e-06, + "loss": 0.784, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004816738725505748, + "step": 1454 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7096, + "image_masks_loss": 0.0211, + "llm_learning_rate": 4.8164268171316445e-06, + "loss": 0.709, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.00048164268171316444, + "step": 1455 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7095, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.816114653667461e-06, + "loss": 0.8165, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00048161146536674607, + "step": 1456 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7093, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.815802235147571e-06, + "loss": 0.7444, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004815802235147571, + "step": 1457 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7091, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.815489561606382e-06, + "loss": 0.7051, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048154895616063823, + "step": 1458 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7089, + "image_masks_loss": 0.0228, + "llm_learning_rate": 4.815176633078325e-06, + "loss": 0.7381, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.0004815176633078324, + "step": 1459 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7088, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.81486344959786e-06, + "loss": 0.7752, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00048148634495978593, + "step": 1460 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7086, + "image_masks_loss": 0.0919, + "llm_learning_rate": 4.814550011199475e-06, + "loss": 0.8592, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004814550011199475, + "step": 1461 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7084, + "image_masks_loss": 0.0378, + "llm_learning_rate": 4.814236317917689e-06, + "loss": 0.7745, + "output_text_mask_loss": 0.0128, + "predictor_learning_rate": 0.00048142363179176883, + "step": 1462 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7083, + "image_masks_loss": 0.0215, + "llm_learning_rate": 4.813922369787043e-06, + "loss": 0.8053, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00048139223697870427, + "step": 1463 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7081, + "image_masks_loss": 0.061, + "llm_learning_rate": 4.813608166842113e-06, + "loss": 0.8305, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00048136081668421126, + "step": 1464 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7079, + "image_masks_loss": 0.0429, + "llm_learning_rate": 4.813293709117497e-06, + "loss": 0.78, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048132937091174973, + "step": 1465 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7078, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.812978996647827e-06, + "loss": 0.7937, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00048129789966478266, + "step": 1466 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7076, + "image_masks_loss": 0.1236, + "llm_learning_rate": 4.8126640294677565e-06, + "loss": 0.736, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004812664029467756, + "step": 1467 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7074, + "image_masks_loss": 0.0182, + "llm_learning_rate": 4.812348807611972e-06, + "loss": 0.8341, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.00048123488076119714, + "step": 1468 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7073, + "image_masks_loss": 0.0222, + "llm_learning_rate": 4.812033331115186e-06, + "loss": 0.8051, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004812033331115186, + "step": 1469 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7071, + "image_masks_loss": 0.0622, + "llm_learning_rate": 4.81171760001214e-06, + "loss": 0.7532, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.000481171760001214, + "step": 1470 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7069, + "image_masks_loss": 0.0721, + "llm_learning_rate": 4.811401614337603e-06, + "loss": 0.7975, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00048114016143376026, + "step": 1471 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7068, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.811085374126372e-06, + "loss": 0.8068, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0004811085374126372, + "step": 1472 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7066, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.810768879413271e-06, + "loss": 0.7357, + "output_text_mask_loss": 0.0257, + "predictor_learning_rate": 0.0004810768879413271, + "step": 1473 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7064, + "image_masks_loss": 0.036, + "llm_learning_rate": 4.810452130233155e-06, + "loss": 0.8921, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.0004810452130233155, + "step": 1474 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7063, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.810135126620904e-06, + "loss": 0.761, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.0004810135126620903, + "step": 1475 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7061, + "image_masks_loss": 0.0397, + "llm_learning_rate": 4.809817868611426e-06, + "loss": 0.7888, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004809817868611426, + "step": 1476 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7059, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.809500356239661e-06, + "loss": 0.6945, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.000480950035623966, + "step": 1477 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7058, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.809182589540571e-06, + "loss": 0.6953, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048091825895405703, + "step": 1478 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7056, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.808864568549152e-06, + "loss": 0.8263, + "output_text_mask_loss": 0.0832, + "predictor_learning_rate": 0.00048088645685491515, + "step": 1479 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7054, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.808546293300423e-06, + "loss": 0.8438, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00048085462933004225, + "step": 1480 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7053, + "image_masks_loss": 0.0336, + "llm_learning_rate": 4.808227763829434e-06, + "loss": 0.7813, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.0004808227763829434, + "step": 1481 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7051, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.807908980171263e-06, + "loss": 0.7338, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0004807908980171263, + "step": 1482 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7049, + "image_masks_loss": 0.0315, + "llm_learning_rate": 4.8075899423610136e-06, + "loss": 0.7732, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00048075899423610135, + "step": 1483 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7048, + "image_masks_loss": 0.0229, + "llm_learning_rate": 4.807270650433821e-06, + "loss": 0.7425, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004807270650433821, + "step": 1484 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7046, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.806951104424844e-06, + "loss": 0.7567, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.0004806951104424844, + "step": 1485 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7044, + "image_masks_loss": 0.0401, + "llm_learning_rate": 4.806631304369274e-06, + "loss": 0.8015, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00048066313043692736, + "step": 1486 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7043, + "image_masks_loss": 0.0557, + "llm_learning_rate": 4.806311250302326e-06, + "loss": 0.7426, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004806311250302326, + "step": 1487 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7041, + "image_masks_loss": 0.0228, + "llm_learning_rate": 4.805990942259247e-06, + "loss": 0.8078, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0004805990942259246, + "step": 1488 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7039, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.805670380275308e-06, + "loss": 0.7633, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00048056703802753077, + "step": 1489 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7038, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.805349564385812e-06, + "loss": 0.7637, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00048053495643858113, + "step": 1490 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7036, + "image_masks_loss": 0.0215, + "llm_learning_rate": 4.805028494626086e-06, + "loss": 0.7348, + "output_text_mask_loss": 0.0751, + "predictor_learning_rate": 0.0004805028494626086, + "step": 1491 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7035, + "image_masks_loss": 0.0048, + "llm_learning_rate": 4.804707171031488e-06, + "loss": 0.6735, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0004804707171031488, + "step": 1492 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7033, + "image_masks_loss": 0.0683, + "llm_learning_rate": 4.804385593637403e-06, + "loss": 0.8722, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00048043855936374025, + "step": 1493 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7031, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.804063762479244e-06, + "loss": 0.9129, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00048040637624792437, + "step": 1494 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.703, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.803741677592451e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048037416775924507, + "step": 1495 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7028, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.803419339012493e-06, + "loss": 0.8467, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004803419339012493, + "step": 1496 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7026, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.803096746774867e-06, + "loss": 0.6501, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004803096746774867, + "step": 1497 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7025, + "image_masks_loss": 0.0159, + "llm_learning_rate": 4.802773900915098e-06, + "loss": 0.7526, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048027739009150975, + "step": 1498 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7023, + "image_masks_loss": 0.0576, + "llm_learning_rate": 4.802450801468737e-06, + "loss": 0.838, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00048024508014687364, + "step": 1499 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7021, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.802127448471364e-06, + "loss": 0.8477, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.0004802127448471364, + "step": 1500 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.702, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.801803841958591e-06, + "loss": 0.7997, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.000480180384195859, + "step": 1501 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7018, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.80147998196605e-06, + "loss": 0.7727, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00048014799819660496, + "step": 1502 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7016, + "image_masks_loss": 0.0406, + "llm_learning_rate": 4.801155868529408e-06, + "loss": 0.7052, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004801155868529408, + "step": 1503 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7015, + "image_masks_loss": 0.045, + "llm_learning_rate": 4.800831501684356e-06, + "loss": 0.8, + "output_text_mask_loss": 0.1068, + "predictor_learning_rate": 0.00048008315016843554, + "step": 1504 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7013, + "image_masks_loss": 0.0614, + "llm_learning_rate": 4.800506881466613e-06, + "loss": 0.7299, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00048005068814666133, + "step": 1505 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7011, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.80018200791193e-06, + "loss": 0.7868, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00048001820079119294, + "step": 1506 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.701, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.79985688105608e-06, + "loss": 0.7667, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.00047998568810560797, + "step": 1507 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7008, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.799531500934867e-06, + "loss": 0.7253, + "output_text_mask_loss": 0.103, + "predictor_learning_rate": 0.00047995315009348666, + "step": 1508 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7006, + "image_masks_loss": 0.0147, + "llm_learning_rate": 4.799205867584123e-06, + "loss": 0.703, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0004799205867584123, + "step": 1509 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7005, + "image_masks_loss": 0.008, + "llm_learning_rate": 4.798879981039709e-06, + "loss": 0.7316, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004798879981039708, + "step": 1510 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7003, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.79855384133751e-06, + "loss": 0.7883, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00047985538413375096, + "step": 1511 + }, + { + "epoch": 0.15, + "gumbel_tau": 0.7001, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.798227448513441e-06, + "loss": 0.7343, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00047982274485134415, + "step": 1512 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.7, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.797900802603448e-06, + "loss": 0.8544, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00047979008026034477, + "step": 1513 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6998, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.7975739036435e-06, + "loss": 0.72, + "output_text_mask_loss": 0.0206, + "predictor_learning_rate": 0.00047975739036434993, + "step": 1514 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6996, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.797246751669595e-06, + "loss": 0.85, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0004797246751669595, + "step": 1515 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6995, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.7969193467177624e-06, + "loss": 0.8039, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004796919346717762, + "step": 1516 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6993, + "image_masks_loss": 0.0572, + "llm_learning_rate": 4.796591688824054e-06, + "loss": 0.7549, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047965916888240536, + "step": 1517 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6991, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.796263778024554e-06, + "loss": 0.8026, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0004796263778024553, + "step": 1518 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.699, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.795935614355372e-06, + "loss": 0.7667, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004795935614355371, + "step": 1519 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6988, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.7956071978526454e-06, + "loss": 0.8822, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004795607197852645, + "step": 1520 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6987, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.795278528552541e-06, + "loss": 0.7703, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00047952785285525413, + "step": 1521 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6985, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.794949606491253e-06, + "loss": 0.7466, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.0004794949606491253, + "step": 1522 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6983, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.794620431705002e-06, + "loss": 0.6743, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004794620431705002, + "step": 1523 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6982, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.79429100423004e-06, + "loss": 0.7663, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.0004794291004230039, + "step": 1524 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.698, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.793961324102639e-06, + "loss": 0.751, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0004793961324102639, + "step": 1525 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6978, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.79363139135911e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000479363139135911, + "step": 1526 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6977, + "image_masks_loss": 0.0182, + "llm_learning_rate": 4.793301206035782e-06, + "loss": 0.7713, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00047933012060357816, + "step": 1527 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6975, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.792970768169017e-06, + "loss": 0.7448, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004792970768169017, + "step": 1528 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6973, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.792640077795205e-06, + "loss": 0.7009, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00047926400777952043, + "step": 1529 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6972, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.79230913495076e-06, + "loss": 0.7558, + "output_text_mask_loss": 0.0725, + "predictor_learning_rate": 0.00047923091349507595, + "step": 1530 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.697, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.791977939672127e-06, + "loss": 0.6596, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00047919779396721267, + "step": 1531 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6968, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.791646491995778e-06, + "loss": 0.8003, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004791646491995778, + "step": 1532 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6967, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.791314791958214e-06, + "loss": 0.7579, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047913147919582136, + "step": 1533 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6965, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.79098283959596e-06, + "loss": 0.7605, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.000479098283959596, + "step": 1534 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6964, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.790650634945573e-06, + "loss": 0.7379, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004790650634945573, + "step": 1535 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6962, + "image_masks_loss": 0.0303, + "llm_learning_rate": 4.790318178043637e-06, + "loss": 0.8678, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.00047903181780436366, + "step": 1536 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.696, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.789985468926761e-06, + "loss": 0.7688, + "output_text_mask_loss": 0.0393, + "predictor_learning_rate": 0.000478998546892676, + "step": 1537 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6959, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.789652507631584e-06, + "loss": 0.8082, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004789652507631583, + "step": 1538 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6957, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.789319294194773e-06, + "loss": 0.7718, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004789319294194773, + "step": 1539 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6955, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.788985828653022e-06, + "loss": 0.7899, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004788985828653022, + "step": 1540 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6954, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.788652111043054e-06, + "loss": 0.8559, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.00047886521110430535, + "step": 1541 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6952, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.7883181414016165e-06, + "loss": 0.9246, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00047883181414016163, + "step": 1542 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.695, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.787983919765489e-06, + "loss": 0.7357, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.0004787983919765488, + "step": 1543 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6949, + "image_masks_loss": 0.0399, + "llm_learning_rate": 4.787649446171476e-06, + "loss": 0.8206, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.00047876494461714754, + "step": 1544 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6947, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.78731472065641e-06, + "loss": 0.8151, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.00047873147206564095, + "step": 1545 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6945, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.786979743257152e-06, + "loss": 0.7578, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0004786979743257152, + "step": 1546 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6944, + "image_masks_loss": 0.0622, + "llm_learning_rate": 4.786644514010591e-06, + "loss": 0.7905, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.00047866445140105907, + "step": 1547 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6942, + "image_masks_loss": 0.0341, + "llm_learning_rate": 4.786309032953643e-06, + "loss": 0.7892, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.0004786309032953643, + "step": 1548 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6941, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.7859733001232515e-06, + "loss": 0.7422, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004785973300123251, + "step": 1549 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6939, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.785637315556388e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.061, + "predictor_learning_rate": 0.0004785637315556388, + "step": 1550 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6937, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.785301079290054e-06, + "loss": 0.7121, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004785301079290053, + "step": 1551 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6936, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.784964591361273e-06, + "loss": 0.749, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047849645913612725, + "step": 1552 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6934, + "image_masks_loss": 0.0204, + "llm_learning_rate": 4.784627851807102e-06, + "loss": 0.7886, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.0004784627851807102, + "step": 1553 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6932, + "image_masks_loss": 0.0141, + "llm_learning_rate": 4.784290860664623e-06, + "loss": 0.7412, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004784290860664623, + "step": 1554 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6931, + "image_masks_loss": 0.0076, + "llm_learning_rate": 4.783953617970946e-06, + "loss": 0.6318, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004783953617970946, + "step": 1555 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6929, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.783616123763211e-06, + "loss": 0.7918, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.00047836161237632105, + "step": 1556 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6927, + "image_masks_loss": 0.008, + "llm_learning_rate": 4.783278378078581e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.00047832783780785805, + "step": 1557 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6926, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.7829403809542495e-06, + "loss": 0.7456, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004782940380954249, + "step": 1558 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6924, + "image_masks_loss": 0.0507, + "llm_learning_rate": 4.782602132427439e-06, + "loss": 0.8659, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004782602132427439, + "step": 1559 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6923, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.782263632535397e-06, + "loss": 0.8388, + "output_text_mask_loss": 0.0448, + "predictor_learning_rate": 0.0004782263632535397, + "step": 1560 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6921, + "image_masks_loss": 0.0036, + "llm_learning_rate": 4.781924881315401e-06, + "loss": 0.751, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00047819248813154, + "step": 1561 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6919, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.781585878804753e-06, + "loss": 0.7989, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.00047815858788047527, + "step": 1562 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6918, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.7812466250407864e-06, + "loss": 0.7286, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.00047812466250407863, + "step": 1563 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6916, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.780907120060861e-06, + "loss": 0.7166, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00047809071200608606, + "step": 1564 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6914, + "image_masks_loss": 0.0395, + "llm_learning_rate": 4.780567363902363e-06, + "loss": 0.7465, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004780567363902362, + "step": 1565 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6913, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.7802273566027056e-06, + "loss": 0.8155, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.00047802273566027053, + "step": 1566 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6911, + "image_masks_loss": 0.0229, + "llm_learning_rate": 4.779887098199334e-06, + "loss": 0.7923, + "output_text_mask_loss": 0.0303, + "predictor_learning_rate": 0.00047798870981993335, + "step": 1567 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.691, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.779546588729717e-06, + "loss": 0.7417, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00047795465887297164, + "step": 1568 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6908, + "image_masks_loss": 0.0016, + "llm_learning_rate": 4.779205828231351e-06, + "loss": 0.67, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004779205828231351, + "step": 1569 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6906, + "image_masks_loss": 0.017, + "llm_learning_rate": 4.778864816741763e-06, + "loss": 0.7771, + "output_text_mask_loss": 0.032, + "predictor_learning_rate": 0.0004778864816741763, + "step": 1570 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6905, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.778523554298506e-06, + "loss": 0.7799, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00047785235542985055, + "step": 1571 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6903, + "image_masks_loss": 0.0572, + "llm_learning_rate": 4.77818204093916e-06, + "loss": 0.9395, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00047781820409391595, + "step": 1572 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6901, + "image_masks_loss": 0.0074, + "llm_learning_rate": 4.7778402767013325e-06, + "loss": 0.7606, + "output_text_mask_loss": 0.1175, + "predictor_learning_rate": 0.0004777840276701332, + "step": 1573 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.69, + "image_masks_loss": 0.0061, + "llm_learning_rate": 4.77749826162266e-06, + "loss": 0.6807, + "output_text_mask_loss": 0.0614, + "predictor_learning_rate": 0.00047774982616226595, + "step": 1574 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6898, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.777155995740807e-06, + "loss": 0.6939, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004777155995740806, + "step": 1575 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6896, + "image_masks_loss": 0.0072, + "llm_learning_rate": 4.7768134790934625e-06, + "loss": 0.7412, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00047768134790934617, + "step": 1576 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6895, + "image_masks_loss": 0.0526, + "llm_learning_rate": 4.776470711718346e-06, + "loss": 0.7639, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004776470711718346, + "step": 1577 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6893, + "image_masks_loss": 0.0519, + "llm_learning_rate": 4.776127693653205e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004776127693653205, + "step": 1578 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6892, + "image_masks_loss": 0.0102, + "llm_learning_rate": 4.7757844249358125e-06, + "loss": 0.7188, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047757844249358126, + "step": 1579 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.689, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.77544090560397e-06, + "loss": 0.7439, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047754409056039696, + "step": 1580 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6888, + "image_masks_loss": 0.0217, + "llm_learning_rate": 4.775097135695506e-06, + "loss": 0.8125, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0004775097135695506, + "step": 1581 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6887, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.774753115248279e-06, + "loss": 0.6976, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004774753115248278, + "step": 1582 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6885, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.774408844300171e-06, + "loss": 0.7329, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.0004774408844300171, + "step": 1583 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6883, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.774064322889095e-06, + "loss": 0.7443, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00047740643228890946, + "step": 1584 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6882, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.773719551052991e-06, + "loss": 0.7641, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004773719551052991, + "step": 1585 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.688, + "image_masks_loss": 0.0381, + "llm_learning_rate": 4.773374528829825e-06, + "loss": 0.8942, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.0004773374528829825, + "step": 1586 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6879, + "image_masks_loss": 0.066, + "llm_learning_rate": 4.773029256257593e-06, + "loss": 0.8016, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0004773029256257593, + "step": 1587 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6877, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.772683733374316e-06, + "loss": 0.8736, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00047726837333743154, + "step": 1588 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6875, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.772337960218045e-06, + "loss": 0.6937, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004772337960218044, + "step": 1589 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6874, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.771991936826854e-06, + "loss": 0.7277, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004771991936826854, + "step": 1590 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6872, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.771645663238852e-06, + "loss": 0.7049, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047716456632388517, + "step": 1591 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.687, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.771299139492169e-06, + "loss": 0.7908, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004771299139492169, + "step": 1592 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6869, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.770952365624966e-06, + "loss": 0.7863, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00047709523656249656, + "step": 1593 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6867, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.77060534167543e-06, + "loss": 0.7864, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00047706053416754295, + "step": 1594 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6866, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.770258067681776e-06, + "loss": 0.8341, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004770258067681776, + "step": 1595 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6864, + "image_masks_loss": 0.0105, + "llm_learning_rate": 4.769910543682247e-06, + "loss": 0.8387, + "output_text_mask_loss": 0.0338, + "predictor_learning_rate": 0.00047699105436822467, + "step": 1596 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6862, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.7695627697151135e-06, + "loss": 0.8048, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00047695627697151127, + "step": 1597 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6861, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.769214745818671e-06, + "loss": 0.8272, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.00047692147458186713, + "step": 1598 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6859, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.768866472031248e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00047688664720312474, + "step": 1599 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6858, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.768517948391195e-06, + "loss": 0.7079, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.00047685179483911947, + "step": 1600 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6856, + "image_masks_loss": 0.0219, + "llm_learning_rate": 4.768169174936892e-06, + "loss": 0.6301, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004768169174936892, + "step": 1601 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6854, + "image_masks_loss": 0.038, + "llm_learning_rate": 4.767820151706747e-06, + "loss": 0.8043, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00047678201517067464, + "step": 1602 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6853, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.767470878739196e-06, + "loss": 0.7792, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004767470878739196, + "step": 1603 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6851, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.767121356072701e-06, + "loss": 0.8583, + "output_text_mask_loss": 0.0492, + "predictor_learning_rate": 0.0004767121356072701, + "step": 1604 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6849, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.766771583745753e-06, + "loss": 0.7056, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004766771583745753, + "step": 1605 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6848, + "image_masks_loss": 0.0072, + "llm_learning_rate": 4.766421561796869e-06, + "loss": 0.832, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.00047664215617968685, + "step": 1606 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6846, + "image_masks_loss": 0.013, + "llm_learning_rate": 4.766071290264594e-06, + "loss": 0.7845, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004766071290264594, + "step": 1607 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6845, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.765720769187502e-06, + "loss": 0.8174, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.00047657207691875015, + "step": 1608 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6843, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.765369998604191e-06, + "loss": 0.7769, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004765369998604191, + "step": 1609 + }, + { + "epoch": 0.16, + "gumbel_tau": 0.6841, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.7650189785532905e-06, + "loss": 0.7057, + "output_text_mask_loss": 0.0212, + "predictor_learning_rate": 0.000476501897855329, + "step": 1610 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.684, + "image_masks_loss": 0.0345, + "llm_learning_rate": 4.764667709073455e-06, + "loss": 0.8576, + "output_text_mask_loss": 0.0362, + "predictor_learning_rate": 0.00047646677090734543, + "step": 1611 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6838, + "image_masks_loss": 0.0298, + "llm_learning_rate": 4.764316190203367e-06, + "loss": 0.682, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0004764316190203366, + "step": 1612 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6837, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.763964421981736e-06, + "loss": 0.8295, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0004763964421981735, + "step": 1613 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6835, + "image_masks_loss": 0.0471, + "llm_learning_rate": 4.7636124044473e-06, + "loss": 0.7043, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00047636124044472995, + "step": 1614 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6833, + "image_masks_loss": 0.0181, + "llm_learning_rate": 4.763260137638824e-06, + "loss": 0.709, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004763260137638824, + "step": 1615 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6832, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.762907621595102e-06, + "loss": 0.8487, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00047629076215951013, + "step": 1616 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.683, + "image_masks_loss": 0.0473, + "llm_learning_rate": 4.7625548563549505e-06, + "loss": 0.8069, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00047625548563549506, + "step": 1617 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6828, + "image_masks_loss": 0.034, + "llm_learning_rate": 4.762201841957219e-06, + "loss": 0.7447, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 0.0004762201841957219, + "step": 1618 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6827, + "image_masks_loss": 0.0402, + "llm_learning_rate": 4.761848578440782e-06, + "loss": 0.8158, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00047618485784407816, + "step": 1619 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6825, + "image_masks_loss": 0.0614, + "llm_learning_rate": 4.761495065844542e-06, + "loss": 0.8409, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.00047614950658445417, + "step": 1620 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6824, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.761141304207427e-06, + "loss": 0.7527, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 0.0004761141304207427, + "step": 1621 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6822, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.760787293568396e-06, + "loss": 0.7659, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004760787293568396, + "step": 1622 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.682, + "image_masks_loss": 0.0347, + "llm_learning_rate": 4.760433033966432e-06, + "loss": 0.727, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047604330339664315, + "step": 1623 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6819, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.760078525440548e-06, + "loss": 0.7338, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00047600785254405475, + "step": 1624 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6817, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.759723768029782e-06, + "loss": 0.7161, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00047597237680297816, + "step": 1625 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6816, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.7593687617732015e-06, + "loss": 0.8109, + "output_text_mask_loss": 0.0412, + "predictor_learning_rate": 0.00047593687617732014, + "step": 1626 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6814, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.7590135067099e-06, + "loss": 0.772, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00047590135067099, + "step": 1627 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6812, + "image_masks_loss": 0.0345, + "llm_learning_rate": 4.758658002879001e-06, + "loss": 0.7602, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004758658002879, + "step": 1628 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6811, + "image_masks_loss": 0.0456, + "llm_learning_rate": 4.75830225031965e-06, + "loss": 0.7826, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.00047583022503196496, + "step": 1629 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6809, + "image_masks_loss": 0.09, + "llm_learning_rate": 4.7579462490710256e-06, + "loss": 0.8814, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004757946249071025, + "step": 1630 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6808, + "image_masks_loss": 0.0046, + "llm_learning_rate": 4.75758999917233e-06, + "loss": 0.7837, + "output_text_mask_loss": 0.0234, + "predictor_learning_rate": 0.00047575899991723306, + "step": 1631 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6806, + "image_masks_loss": 0.0223, + "llm_learning_rate": 4.7572335006627964e-06, + "loss": 0.8075, + "output_text_mask_loss": 0.0094, + "predictor_learning_rate": 0.0004757233500662797, + "step": 1632 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6804, + "image_masks_loss": 0.0471, + "llm_learning_rate": 4.756876753581682e-06, + "loss": 0.7848, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004756876753581681, + "step": 1633 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6803, + "image_masks_loss": 0.0387, + "llm_learning_rate": 4.756519757968272e-06, + "loss": 0.7819, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004756519757968272, + "step": 1634 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6801, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.75616251386188e-06, + "loss": 0.8697, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00047561625138618796, + "step": 1635 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.68, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.755805021301847e-06, + "loss": 0.8084, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.0004755805021301847, + "step": 1636 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6798, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.75544728032754e-06, + "loss": 0.7247, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 0.00047554472803275393, + "step": 1637 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6796, + "image_masks_loss": 0.0717, + "llm_learning_rate": 4.755089290978354e-06, + "loss": 0.8555, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00047550892909783537, + "step": 1638 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6795, + "image_masks_loss": 0.0534, + "llm_learning_rate": 4.754731053293713e-06, + "loss": 0.7969, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00047547310532937133, + "step": 1639 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6793, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.754372567313067e-06, + "loss": 0.7432, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004754372567313067, + "step": 1640 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6792, + "image_masks_loss": 0.0182, + "llm_learning_rate": 4.754013833075891e-06, + "loss": 0.7508, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004754013833075891, + "step": 1641 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.679, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.753654850621692e-06, + "loss": 0.7574, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00047536548506216914, + "step": 1642 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6788, + "image_masks_loss": 0.0591, + "llm_learning_rate": 4.753295619990001e-06, + "loss": 0.8414, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00047532956199900005, + "step": 1643 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6787, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.752936141220376e-06, + "loss": 0.8407, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004752936141220376, + "step": 1644 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6785, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.752576414352407e-06, + "loss": 0.7687, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.0004752576414352406, + "step": 1645 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6784, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.752216439425704e-06, + "loss": 0.8268, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00047522164394257045, + "step": 1646 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6782, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.751856216479912e-06, + "loss": 0.7245, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00047518562164799115, + "step": 1647 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.678, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.751495745554696e-06, + "loss": 0.7576, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047514957455546957, + "step": 1648 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6779, + "image_masks_loss": 0.0395, + "llm_learning_rate": 4.751135026689754e-06, + "loss": 0.7347, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004751135026689754, + "step": 1649 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6777, + "image_masks_loss": 0.0587, + "llm_learning_rate": 4.7507740599248096e-06, + "loss": 0.6762, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0004750774059924809, + "step": 1650 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6776, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.750412845299611e-06, + "loss": 0.7563, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004750412845299611, + "step": 1651 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6774, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.750051382853938e-06, + "loss": 0.7189, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047500513828539374, + "step": 1652 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6772, + "image_masks_loss": 0.0347, + "llm_learning_rate": 4.749689672627594e-06, + "loss": 0.8173, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0004749689672627594, + "step": 1653 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6771, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.749327714660413e-06, + "loss": 0.8108, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004749327714660413, + "step": 1654 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6769, + "image_masks_loss": 0.0435, + "llm_learning_rate": 4.748965508992254e-06, + "loss": 0.8158, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00047489655089922537, + "step": 1655 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6768, + "image_masks_loss": 0.0687, + "llm_learning_rate": 4.7486030556630035e-06, + "loss": 0.769, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004748603055663003, + "step": 1656 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6766, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.748240354712575e-06, + "loss": 0.8144, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00047482403547125754, + "step": 1657 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6764, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.747877406180913e-06, + "loss": 0.8032, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004747877406180913, + "step": 1658 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6763, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.747514210107983e-06, + "loss": 0.768, + "output_text_mask_loss": 0.0427, + "predictor_learning_rate": 0.00047475142101079825, + "step": 1659 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6761, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.747150766533782e-06, + "loss": 0.7237, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.00047471507665337815, + "step": 1660 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.676, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.746787075498332e-06, + "loss": 0.8612, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004746787075498332, + "step": 1661 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6758, + "image_masks_loss": 0.0222, + "llm_learning_rate": 4.746423137041687e-06, + "loss": 0.8403, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.00047464231370416865, + "step": 1662 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6756, + "image_masks_loss": 0.0345, + "llm_learning_rate": 4.7460589512039215e-06, + "loss": 0.775, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.0004746058951203921, + "step": 1663 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6755, + "image_masks_loss": 0.0155, + "llm_learning_rate": 4.745694518025142e-06, + "loss": 0.7753, + "output_text_mask_loss": 0.0418, + "predictor_learning_rate": 0.0004745694518025141, + "step": 1664 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6753, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.745329837545478e-06, + "loss": 0.6884, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.00047453298375454777, + "step": 1665 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6752, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.744964909805093e-06, + "loss": 0.8724, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00047449649098050927, + "step": 1666 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.675, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.744599734844171e-06, + "loss": 0.784, + "output_text_mask_loss": 0.0664, + "predictor_learning_rate": 0.00047445997348441704, + "step": 1667 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6748, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.744234312702927e-06, + "loss": 0.6635, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004744234312702926, + "step": 1668 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6747, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.743868643421601e-06, + "loss": 0.7796, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00047438686434216005, + "step": 1669 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6745, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.743502727040462e-06, + "loss": 0.7643, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0004743502727040462, + "step": 1670 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6744, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.743136563599806e-06, + "loss": 0.7902, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004743136563599806, + "step": 1671 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6742, + "image_masks_loss": 0.0233, + "llm_learning_rate": 4.7427701531399554e-06, + "loss": 0.7845, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00047427701531399546, + "step": 1672 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.674, + "image_masks_loss": 0.0395, + "llm_learning_rate": 4.742403495701259e-06, + "loss": 0.785, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.00047424034957012584, + "step": 1673 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6739, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.742036591324095e-06, + "loss": 0.9102, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004742036591324095, + "step": 1674 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6737, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.741669440048869e-06, + "loss": 0.7932, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0004741669440048868, + "step": 1675 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6736, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.741302041916009e-06, + "loss": 0.7933, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.0004741302041916009, + "step": 1676 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6734, + "image_masks_loss": 0.036, + "llm_learning_rate": 4.740934396965977e-06, + "loss": 0.8752, + "output_text_mask_loss": 0.0372, + "predictor_learning_rate": 0.0004740934396965977, + "step": 1677 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6732, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.740566505239258e-06, + "loss": 0.8447, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004740566505239258, + "step": 1678 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6731, + "image_masks_loss": 0.017, + "llm_learning_rate": 4.740198366776364e-06, + "loss": 0.7809, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004740198366776364, + "step": 1679 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6729, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.739829981617837e-06, + "loss": 0.8003, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 0.0004739829981617837, + "step": 1680 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6728, + "image_masks_loss": 0.0288, + "llm_learning_rate": 4.739461349804243e-06, + "loss": 0.7949, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004739461349804243, + "step": 1681 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6726, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.739092471376176e-06, + "loss": 0.8274, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004739092471376176, + "step": 1682 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6725, + "image_masks_loss": 0.0315, + "llm_learning_rate": 4.7387233463742605e-06, + "loss": 0.736, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000473872334637426, + "step": 1683 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6723, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.738353974839142e-06, + "loss": 0.7963, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.00047383539748391425, + "step": 1684 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6721, + "image_masks_loss": 0.0053, + "llm_learning_rate": 4.737984356811499e-06, + "loss": 0.7524, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004737984356811499, + "step": 1685 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.672, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.737614492332033e-06, + "loss": 0.728, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0004737614492332033, + "step": 1686 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6718, + "image_masks_loss": 0.0341, + "llm_learning_rate": 4.737244381441476e-06, + "loss": 0.8162, + "output_text_mask_loss": 0.0067, + "predictor_learning_rate": 0.00047372443814414756, + "step": 1687 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6717, + "image_masks_loss": 0.0315, + "llm_learning_rate": 4.736874024180583e-06, + "loss": 0.8002, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004736874024180583, + "step": 1688 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6715, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.736503420590142e-06, + "loss": 0.8002, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.00047365034205901414, + "step": 1689 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6713, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.736132570710961e-06, + "loss": 0.7661, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.000473613257071096, + "step": 1690 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6712, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.735761474583881e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.0004735761474583881, + "step": 1691 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.671, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.735390132249768e-06, + "loss": 0.8126, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0004735390132249767, + "step": 1692 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6709, + "image_masks_loss": 0.0345, + "llm_learning_rate": 4.735018543749514e-06, + "loss": 0.7586, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 0.0004735018543749513, + "step": 1693 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6707, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.734646709124039e-06, + "loss": 0.6798, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00047346467091240386, + "step": 1694 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6706, + "image_masks_loss": 0.0213, + "llm_learning_rate": 4.734274628414292e-06, + "loss": 0.7871, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00047342746284142914, + "step": 1695 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6704, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.733902301661245e-06, + "loss": 0.8551, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047339023016612446, + "step": 1696 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6702, + "image_masks_loss": 0.016, + "llm_learning_rate": 4.733529728905902e-06, + "loss": 0.768, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.00047335297289059016, + "step": 1697 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6701, + "image_masks_loss": 0.0222, + "llm_learning_rate": 4.7331569101892894e-06, + "loss": 0.7071, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004733156910189289, + "step": 1698 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6699, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.732783845552464e-06, + "loss": 0.8236, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0004732783845552464, + "step": 1699 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6698, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.732410535036509e-06, + "loss": 0.7478, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004732410535036509, + "step": 1700 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6696, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.732036978682532e-06, + "loss": 0.7769, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.00047320369786825326, + "step": 1701 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6694, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.731663176531673e-06, + "loss": 0.6934, + "output_text_mask_loss": 0.0265, + "predictor_learning_rate": 0.0004731663176531672, + "step": 1702 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6693, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.731289128625092e-06, + "loss": 0.8368, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.0004731289128625092, + "step": 1703 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6691, + "image_masks_loss": 0.0068, + "llm_learning_rate": 4.730914835003983e-06, + "loss": 0.7459, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00047309148350039833, + "step": 1704 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.669, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.730540295709564e-06, + "loss": 0.7077, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00047305402957095635, + "step": 1705 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6688, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.730165510783079e-06, + "loss": 0.7744, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00047301655107830786, + "step": 1706 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6687, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.7297904802658e-06, + "loss": 0.6699, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00047297904802658, + "step": 1707 + }, + { + "epoch": 0.17, + "gumbel_tau": 0.6685, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.729415204199027e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004729415204199027, + "step": 1708 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6683, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.729039682624086e-06, + "loss": 0.7823, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00047290396826240856, + "step": 1709 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6682, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.728663915582329e-06, + "loss": 0.8658, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.0004728663915582329, + "step": 1710 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.668, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.728287903115139e-06, + "loss": 0.7539, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.00047282879031151383, + "step": 1711 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6679, + "image_masks_loss": 0.0097, + "llm_learning_rate": 4.727911645263921e-06, + "loss": 0.7818, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.00047279116452639204, + "step": 1712 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6677, + "image_masks_loss": 0.023, + "llm_learning_rate": 4.72753514207011e-06, + "loss": 0.7779, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 0.000472753514207011, + "step": 1713 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6676, + "image_masks_loss": 0.0307, + "llm_learning_rate": 4.727158393575167e-06, + "loss": 0.753, + "output_text_mask_loss": 0.0507, + "predictor_learning_rate": 0.0004727158393575167, + "step": 1714 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6674, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.726781399820583e-06, + "loss": 0.7507, + "output_text_mask_loss": 0.042, + "predictor_learning_rate": 0.0004726781399820582, + "step": 1715 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6672, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.72640416084787e-06, + "loss": 0.7676, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00047264041608478693, + "step": 1716 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6671, + "image_masks_loss": 0.0097, + "llm_learning_rate": 4.726026676698571e-06, + "loss": 0.8167, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004726026676698571, + "step": 1717 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6669, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.725648947414257e-06, + "loss": 0.662, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00047256489474142567, + "step": 1718 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6668, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.725270973036523e-06, + "loss": 0.7508, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00047252709730365226, + "step": 1719 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6666, + "image_masks_loss": 0.0111, + "llm_learning_rate": 4.724892753606994e-06, + "loss": 0.7579, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004724892753606993, + "step": 1720 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6665, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.724514289167318e-06, + "loss": 0.7496, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047245142891673176, + "step": 1721 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6663, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.724135579759173e-06, + "loss": 0.7558, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047241355797591735, + "step": 1722 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6661, + "image_masks_loss": 0.0052, + "llm_learning_rate": 4.723756625424265e-06, + "loss": 0.6677, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004723756625424265, + "step": 1723 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.666, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.723377426204324e-06, + "loss": 0.8269, + "output_text_mask_loss": 0.1091, + "predictor_learning_rate": 0.0004723377426204324, + "step": 1724 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6658, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.722997982141109e-06, + "loss": 0.8519, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00047229979821411083, + "step": 1725 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6657, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.7226182932764034e-06, + "loss": 0.7544, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004722618293276403, + "step": 1726 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6655, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.722238359652022e-06, + "loss": 0.8209, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00047222383596520214, + "step": 1727 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6654, + "image_masks_loss": 0.0086, + "llm_learning_rate": 4.721858181309802e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004721858181309801, + "step": 1728 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6652, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.721477758291609e-06, + "loss": 0.8, + "output_text_mask_loss": 0.0896, + "predictor_learning_rate": 0.0004721477758291609, + "step": 1729 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.665, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.721097090639339e-06, + "loss": 0.7975, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004721097090639338, + "step": 1730 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6649, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.720716178394908e-06, + "loss": 0.7271, + "output_text_mask_loss": 0.0473, + "predictor_learning_rate": 0.00047207161783949085, + "step": 1731 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6647, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.720335021600266e-06, + "loss": 0.8115, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.0004720335021600266, + "step": 1732 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6646, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.719953620297387e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0668, + "predictor_learning_rate": 0.00047199536202973866, + "step": 1733 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6644, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.719571974528269e-06, + "loss": 0.8071, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.0004719571974528269, + "step": 1734 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6643, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.719190084334942e-06, + "loss": 0.672, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00047191900843349414, + "step": 1735 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6641, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.7188079497594594e-06, + "loss": 0.7908, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00047188079497594595, + "step": 1736 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6639, + "image_masks_loss": 0.0385, + "llm_learning_rate": 4.718425570843904e-06, + "loss": 0.8429, + "output_text_mask_loss": 0.0599, + "predictor_learning_rate": 0.0004718425570843903, + "step": 1737 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6638, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.718042947630383e-06, + "loss": 0.8368, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00047180429476303823, + "step": 1738 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6636, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.717660080161032e-06, + "loss": 0.7925, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004717660080161031, + "step": 1739 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6635, + "image_masks_loss": 0.0307, + "llm_learning_rate": 4.717276968478013e-06, + "loss": 0.732, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.00047172769684780127, + "step": 1740 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6633, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.716893612623515e-06, + "loss": 0.7942, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 0.00047168936126235154, + "step": 1741 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6632, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.7165100126397555e-06, + "loss": 0.6749, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00047165100126397554, + "step": 1742 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.663, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.716126168568976e-06, + "loss": 0.8188, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004716126168568976, + "step": 1743 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6628, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.715742080453446e-06, + "loss": 0.7963, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0004715742080453446, + "step": 1744 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6627, + "image_masks_loss": 0.0159, + "llm_learning_rate": 4.715357748335464e-06, + "loss": 0.7843, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00047153577483354637, + "step": 1745 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6625, + "image_masks_loss": 0.0671, + "llm_learning_rate": 4.714973172257351e-06, + "loss": 0.7568, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004714973172257351, + "step": 1746 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6624, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.71458835226146e-06, + "loss": 0.8082, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00047145883522614596, + "step": 1747 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6622, + "image_masks_loss": 0.0102, + "llm_learning_rate": 4.714203288390166e-06, + "loss": 0.7103, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004714203288390165, + "step": 1748 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6621, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.713817980685874e-06, + "loss": 0.7481, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00047138179806858734, + "step": 1749 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6619, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.713432429191015e-06, + "loss": 0.7242, + "output_text_mask_loss": 0.0729, + "predictor_learning_rate": 0.00047134324291910144, + "step": 1750 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6618, + "image_masks_loss": 0.0207, + "llm_learning_rate": 4.7130466339480455e-06, + "loss": 0.8226, + "output_text_mask_loss": 0.0523, + "predictor_learning_rate": 0.00047130466339480456, + "step": 1751 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6616, + "image_masks_loss": 0.0633, + "llm_learning_rate": 4.712660594999453e-06, + "loss": 0.8056, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0004712660594999452, + "step": 1752 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6614, + "image_masks_loss": 0.0351, + "llm_learning_rate": 4.712274312387746e-06, + "loss": 0.6499, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.0004712274312387746, + "step": 1753 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6613, + "image_masks_loss": 0.0212, + "llm_learning_rate": 4.7118877861554645e-06, + "loss": 0.8366, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00047118877861554643, + "step": 1754 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6611, + "image_masks_loss": 0.045, + "llm_learning_rate": 4.7115010163451735e-06, + "loss": 0.7507, + "output_text_mask_loss": 0.0425, + "predictor_learning_rate": 0.0004711501016345173, + "step": 1755 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.661, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.711114002999464e-06, + "loss": 0.7827, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.0004711114002999464, + "step": 1756 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6608, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.7107267461609565e-06, + "loss": 0.9, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.0004710726746160956, + "step": 1757 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6607, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.710339245872295e-06, + "loss": 0.78, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00047103392458722947, + "step": 1758 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6605, + "image_masks_loss": 0.0155, + "llm_learning_rate": 4.709951502176152e-06, + "loss": 0.7938, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004709951502176152, + "step": 1759 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6604, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.7095635151152275e-06, + "loss": 0.7718, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.0004709563515115227, + "step": 1760 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6602, + "image_masks_loss": 0.0492, + "llm_learning_rate": 4.709175284732247e-06, + "loss": 0.8543, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00047091752847322466, + "step": 1761 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.66, + "image_masks_loss": 0.0061, + "llm_learning_rate": 4.7087868110699635e-06, + "loss": 0.7549, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004708786811069963, + "step": 1762 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6599, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.708398094171156e-06, + "loss": 0.7695, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004708398094171156, + "step": 1763 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6597, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.7080091340786326e-06, + "loss": 0.6857, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.00047080091340786324, + "step": 1764 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6596, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.707619930835224e-06, + "loss": 0.854, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004707619930835224, + "step": 1765 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6594, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.707230484483792e-06, + "loss": 0.8165, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.00047072304844837914, + "step": 1766 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6593, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.7068407950672225e-06, + "loss": 0.7602, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004706840795067222, + "step": 1767 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6591, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.706450862628429e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00047064508626284285, + "step": 1768 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6589, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.706060687210351e-06, + "loss": 0.8097, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00047060606872103514, + "step": 1769 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6588, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.705670268855957e-06, + "loss": 0.8964, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004705670268855957, + "step": 1770 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6586, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.705279607608241e-06, + "loss": 0.7826, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.000470527960760824, + "step": 1771 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6585, + "image_masks_loss": 0.0349, + "llm_learning_rate": 4.704888703510221e-06, + "loss": 0.7764, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00047048887035102213, + "step": 1772 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6583, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.704497556604946e-06, + "loss": 0.7612, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00047044975566049464, + "step": 1773 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6582, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.704106166935491e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00047041061669354914, + "step": 1774 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.658, + "image_masks_loss": 0.0077, + "llm_learning_rate": 4.703714534544955e-06, + "loss": 0.8145, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00047037145345449544, + "step": 1775 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6579, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.703322659476465e-06, + "loss": 0.7253, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004703322659476465, + "step": 1776 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6577, + "image_masks_loss": 0.045, + "llm_learning_rate": 4.702930541773177e-06, + "loss": 0.7933, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004702930541773177, + "step": 1777 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6576, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.70253818147827e-06, + "loss": 0.775, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00047025381814782695, + "step": 1778 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6574, + "image_masks_loss": 0.0584, + "llm_learning_rate": 4.702145578634953e-06, + "loss": 0.7382, + "output_text_mask_loss": 0.0427, + "predictor_learning_rate": 0.0004702145578634953, + "step": 1779 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6572, + "image_masks_loss": 0.0549, + "llm_learning_rate": 4.70175273328646e-06, + "loss": 0.7076, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00047017527332864593, + "step": 1780 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6571, + "image_masks_loss": 0.0303, + "llm_learning_rate": 4.701359645476052e-06, + "loss": 0.7632, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0004701359645476051, + "step": 1781 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6569, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.700966315247015e-06, + "loss": 0.6275, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.0004700966315247015, + "step": 1782 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6568, + "image_masks_loss": 0.0618, + "llm_learning_rate": 4.700572742642666e-06, + "loss": 0.7825, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0004700572742642666, + "step": 1783 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6566, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.700178927706346e-06, + "loss": 0.7344, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004700178927706346, + "step": 1784 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6565, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.69978487048142e-06, + "loss": 0.7559, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.000469978487048142, + "step": 1785 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6563, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.699390571011285e-06, + "loss": 0.7687, + "output_text_mask_loss": 0.0114, + "predictor_learning_rate": 0.0004699390571011285, + "step": 1786 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6562, + "image_masks_loss": 0.0391, + "llm_learning_rate": 4.698996029339361e-06, + "loss": 0.7645, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.00046989960293393614, + "step": 1787 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.656, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.698601245509097e-06, + "loss": 0.7843, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00046986012455090963, + "step": 1788 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6558, + "image_masks_loss": 0.0391, + "llm_learning_rate": 4.698206219563966e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.0599, + "predictor_learning_rate": 0.0004698206219563966, + "step": 1789 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6557, + "image_masks_loss": 0.0241, + "llm_learning_rate": 4.69781095154747e-06, + "loss": 0.846, + "output_text_mask_loss": 0.1411, + "predictor_learning_rate": 0.000469781095154747, + "step": 1790 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6555, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.697415441503137e-06, + "loss": 0.7812, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00046974154415031365, + "step": 1791 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6554, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.69701968947452e-06, + "loss": 0.826, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.000469701968947452, + "step": 1792 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6552, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.696623695505202e-06, + "loss": 0.8039, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00046966236955052015, + "step": 1793 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6551, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.696227459638789e-06, + "loss": 0.7821, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.0004696227459638789, + "step": 1794 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6549, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.695830981918918e-06, + "loss": 0.845, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0004695830981918917, + "step": 1795 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6548, + "image_masks_loss": 0.0572, + "llm_learning_rate": 4.695434262389246e-06, + "loss": 0.753, + "output_text_mask_loss": 0.0215, + "predictor_learning_rate": 0.0004695434262389246, + "step": 1796 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6546, + "image_masks_loss": 0.016, + "llm_learning_rate": 4.695037301093465e-06, + "loss": 0.7298, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046950373010934643, + "step": 1797 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6545, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.6946400980752854e-06, + "loss": 0.8474, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0004694640098075285, + "step": 1798 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6543, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.6942426533784504e-06, + "loss": 0.8384, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.000469424265337845, + "step": 1799 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6541, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.693844967046727e-06, + "loss": 0.7243, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.00046938449670467266, + "step": 1800 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.654, + "image_masks_loss": 0.0071, + "llm_learning_rate": 4.69344703912391e-06, + "loss": 0.76, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.0004693447039123909, + "step": 1801 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6538, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.693048869653818e-06, + "loss": 0.7561, + "output_text_mask_loss": 0.0858, + "predictor_learning_rate": 0.00046930488696538176, + "step": 1802 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6537, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.6926504586803e-06, + "loss": 0.7115, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.00046926504586803, + "step": 1803 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6535, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.69225180624723e-06, + "loss": 0.8286, + "output_text_mask_loss": 0.0885, + "predictor_learning_rate": 0.00046922518062472295, + "step": 1804 + }, + { + "epoch": 0.18, + "gumbel_tau": 0.6534, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.691852912398508e-06, + "loss": 0.892, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00046918529123985076, + "step": 1805 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6532, + "image_masks_loss": 0.0404, + "llm_learning_rate": 4.691453777178061e-06, + "loss": 0.8164, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00046914537771780606, + "step": 1806 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6531, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.691054400629843e-06, + "loss": 0.7569, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004691054400629843, + "step": 1807 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6529, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.690654782797833e-06, + "loss": 0.8365, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004690654782797833, + "step": 1808 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6528, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.69025492372604e-06, + "loss": 0.7599, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00046902549237260393, + "step": 1809 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6526, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.689854823458496e-06, + "loss": 0.7659, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046898548234584956, + "step": 1810 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6525, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.68945448203926e-06, + "loss": 0.7373, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00046894544820392603, + "step": 1811 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6523, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.689053899512421e-06, + "loss": 0.7524, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.0004689053899512421, + "step": 1812 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6521, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.688653075922091e-06, + "loss": 0.8365, + "output_text_mask_loss": 0.0992, + "predictor_learning_rate": 0.00046886530759220903, + "step": 1813 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.652, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.688252011312408e-06, + "loss": 0.8778, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.00046882520113124076, + "step": 1814 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6518, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.68785070572754e-06, + "loss": 0.7806, + "output_text_mask_loss": 0.0166, + "predictor_learning_rate": 0.000468785070572754, + "step": 1815 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6517, + "image_masks_loss": 0.0072, + "llm_learning_rate": 4.687449159211679e-06, + "loss": 0.6819, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00046874491592116785, + "step": 1816 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6515, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.687047371809044e-06, + "loss": 0.8717, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.00046870473718090434, + "step": 1817 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6514, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.686645343563881e-06, + "loss": 0.763, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00046866453435638806, + "step": 1818 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6512, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.686243074520462e-06, + "loss": 0.6971, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004686243074520462, + "step": 1819 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6511, + "image_masks_loss": 0.017, + "llm_learning_rate": 4.685840564723086e-06, + "loss": 0.806, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.0004685840564723086, + "step": 1820 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6509, + "image_masks_loss": 0.0152, + "llm_learning_rate": 4.685437814216079e-06, + "loss": 0.8329, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046854378142160784, + "step": 1821 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6508, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.6850348230437914e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004685034823043791, + "step": 1822 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6506, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.684631591250602e-06, + "loss": 0.7953, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 0.0004684631591250602, + "step": 1823 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6505, + "image_masks_loss": 0.0053, + "llm_learning_rate": 4.6842281188809165e-06, + "loss": 0.6881, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0004684228118880916, + "step": 1824 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6503, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.683824405979165e-06, + "loss": 0.7671, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004683824405979165, + "step": 1825 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6501, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.683420452589805e-06, + "loss": 0.8378, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 0.0004683420452589805, + "step": 1826 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.65, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.683016258757321e-06, + "loss": 0.768, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.0004683016258757321, + "step": 1827 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6498, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.6826118245262255e-06, + "loss": 0.7447, + "output_text_mask_loss": 0.0896, + "predictor_learning_rate": 0.0004682611824526225, + "step": 1828 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6497, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.682207149941054e-06, + "loss": 0.6563, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00046822071499410536, + "step": 1829 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6495, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.6818022350463695e-06, + "loss": 0.6911, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046818022350463695, + "step": 1830 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6494, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.681397079886765e-06, + "loss": 0.7066, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00046813970798867643, + "step": 1831 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6492, + "image_masks_loss": 0.0233, + "llm_learning_rate": 4.680991684506853e-06, + "loss": 0.7661, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004680991684506853, + "step": 1832 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6491, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.68058604895128e-06, + "loss": 0.8407, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00046805860489512795, + "step": 1833 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6489, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.680180173264714e-06, + "loss": 0.811, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00046801801732647137, + "step": 1834 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6488, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.679774057491851e-06, + "loss": 0.8169, + "output_text_mask_loss": 0.0225, + "predictor_learning_rate": 0.0004679774057491851, + "step": 1835 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6486, + "image_masks_loss": 0.0359, + "llm_learning_rate": 4.679367701677414e-06, + "loss": 0.7595, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004679367701677414, + "step": 1836 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6485, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.678961105866151e-06, + "loss": 0.8098, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.0004678961105866151, + "step": 1837 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6483, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.678554270102837e-06, + "loss": 0.8731, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.0004678554270102837, + "step": 1838 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6482, + "image_masks_loss": 0.0328, + "llm_learning_rate": 4.678147194432276e-06, + "loss": 0.763, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00046781471944322753, + "step": 1839 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.648, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.6777398788992925e-06, + "loss": 0.7633, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0004677739878899292, + "step": 1840 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6479, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.677332323548743e-06, + "loss": 0.6917, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0004677332323548743, + "step": 1841 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6477, + "image_masks_loss": 0.0469, + "llm_learning_rate": 4.676924528425508e-06, + "loss": 0.8171, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004676924528425508, + "step": 1842 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6475, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.676516493574496e-06, + "loss": 0.7293, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004676516493574496, + "step": 1843 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6474, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.67610821904064e-06, + "loss": 0.8505, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00046761082190406394, + "step": 1844 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6472, + "image_masks_loss": 0.0084, + "llm_learning_rate": 4.675699704868899e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004675699704868899, + "step": 1845 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6471, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.67529095110426e-06, + "loss": 0.8027, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.000467529095110426, + "step": 1846 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6469, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.674881957791737e-06, + "loss": 0.8292, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0004674881957791737, + "step": 1847 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6468, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.6744727249763686e-06, + "loss": 0.7795, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004674472724976368, + "step": 1848 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6466, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.674063252703219e-06, + "loss": 0.8672, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00046740632527032187, + "step": 1849 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6465, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.6736535410173826e-06, + "loss": 0.8104, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00046736535410173824, + "step": 1850 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6463, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.673243589963977e-06, + "loss": 0.7809, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046732435899639763, + "step": 1851 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6462, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.672833399588146e-06, + "loss": 0.742, + "output_text_mask_loss": 0.0427, + "predictor_learning_rate": 0.00046728333995881455, + "step": 1852 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.646, + "image_masks_loss": 0.0047, + "llm_learning_rate": 4.672422969935061e-06, + "loss": 0.7046, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00046724229699350605, + "step": 1853 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6459, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.67201230104992e-06, + "loss": 0.6683, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00046720123010499196, + "step": 1854 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6457, + "image_masks_loss": 0.0443, + "llm_learning_rate": 4.6716013929779455e-06, + "loss": 0.7876, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00046716013929779454, + "step": 1855 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6456, + "image_masks_loss": 0.0102, + "llm_learning_rate": 4.67119024576439e-06, + "loss": 0.8093, + "output_text_mask_loss": 0.0748, + "predictor_learning_rate": 0.00046711902457643897, + "step": 1856 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6454, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.670778859454528e-06, + "loss": 0.7594, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00046707788594545273, + "step": 1857 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6453, + "image_masks_loss": 0.0243, + "llm_learning_rate": 4.670367234093662e-06, + "loss": 0.6773, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004670367234093662, + "step": 1858 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6451, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.669955369727123e-06, + "loss": 0.883, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004669955369727122, + "step": 1859 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.645, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.6695432664002645e-06, + "loss": 0.7938, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004669543266400264, + "step": 1860 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6448, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.6691309241584695e-06, + "loss": 0.7776, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.0004669130924158469, + "step": 1861 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6446, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.668718343047145e-06, + "loss": 0.6627, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046687183430471445, + "step": 1862 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6445, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.668305523111727e-06, + "loss": 0.7985, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.0004668305523111726, + "step": 1863 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6443, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.667892464397674e-06, + "loss": 0.8252, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004667892464397674, + "step": 1864 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6442, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.667479166950474e-06, + "loss": 0.7135, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.0004667479166950474, + "step": 1865 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.644, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.667065630815641e-06, + "loss": 0.8472, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00046670656308156407, + "step": 1866 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6439, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.666651856038713e-06, + "loss": 0.7587, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.0004666651856038713, + "step": 1867 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6437, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.666237842665257e-06, + "loss": 0.6855, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046662378426652574, + "step": 1868 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6436, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.6658235907408655e-06, + "loss": 0.7147, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046658235907408654, + "step": 1869 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6434, + "image_masks_loss": 0.0229, + "llm_learning_rate": 4.665409100311155e-06, + "loss": 0.7827, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.0004665409100311155, + "step": 1870 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6433, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.664994371421772e-06, + "loss": 0.7903, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0004664994371421771, + "step": 1871 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6431, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.664579404118386e-06, + "loss": 0.8363, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.00046645794041183855, + "step": 1872 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.643, + "image_masks_loss": 0.0351, + "llm_learning_rate": 4.664164198446693e-06, + "loss": 0.7976, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004664164198446693, + "step": 1873 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6428, + "image_masks_loss": 0.017, + "llm_learning_rate": 4.66374875445242e-06, + "loss": 0.758, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000466374875445242, + "step": 1874 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6427, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.663333072181314e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 0.0004663333072181314, + "step": 1875 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6425, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.662917151679152e-06, + "loss": 0.8219, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046629171516791515, + "step": 1876 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6424, + "image_masks_loss": 0.0105, + "llm_learning_rate": 4.662500992991735e-06, + "loss": 0.7208, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00046625009929917346, + "step": 1877 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6422, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.662084596164891e-06, + "loss": 0.7158, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0004662084596164891, + "step": 1878 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6421, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.661667961244477e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 0.0004661667961244477, + "step": 1879 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6419, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.661251088276372e-06, + "loss": 0.6799, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00046612510882763713, + "step": 1880 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6418, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.660833977306482e-06, + "loss": 0.7134, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046608339773064817, + "step": 1881 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6416, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.660416628380741e-06, + "loss": 0.7656, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00046604166283807413, + "step": 1882 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6415, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.659999041545111e-06, + "loss": 0.7671, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00046599990415451106, + "step": 1883 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6413, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.659581216845574e-06, + "loss": 0.847, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004659581216845574, + "step": 1884 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6412, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.659163154328143e-06, + "loss": 0.7383, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00046591631543281434, + "step": 1885 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.641, + "image_masks_loss": 0.0322, + "llm_learning_rate": 4.658744854038856e-06, + "loss": 0.7427, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004658744854038856, + "step": 1886 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6409, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.658326316023778e-06, + "loss": 0.8204, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.0004658326316023778, + "step": 1887 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6407, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.657907540328998e-06, + "loss": 0.733, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00046579075403289976, + "step": 1888 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6406, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.657488527000633e-06, + "loss": 0.6316, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046574885270006327, + "step": 1889 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6404, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.657069276084826e-06, + "loss": 0.6904, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004657069276084826, + "step": 1890 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6403, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.656649787627746e-06, + "loss": 0.8403, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00046566497876277457, + "step": 1891 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6401, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.656230061675588e-06, + "loss": 0.6778, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00046562300616755873, + "step": 1892 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.64, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.655810098274571e-06, + "loss": 0.7802, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004655810098274571, + "step": 1893 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6398, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.655389897470946e-06, + "loss": 0.7568, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00046553898974709454, + "step": 1894 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6397, + "image_masks_loss": 0.0393, + "llm_learning_rate": 4.654969459310984e-06, + "loss": 0.8064, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.0004654969459310983, + "step": 1895 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6395, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.654548783840984e-06, + "loss": 0.7087, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004654548783840984, + "step": 1896 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6393, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.654127871107274e-06, + "loss": 0.6071, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046541278711072735, + "step": 1897 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6392, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.653706721156205e-06, + "loss": 0.6754, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00046537067211562043, + "step": 1898 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.639, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.6532853340341536e-06, + "loss": 0.7799, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00046532853340341534, + "step": 1899 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6389, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.652863709787526e-06, + "loss": 0.7235, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004652863709787525, + "step": 1900 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6387, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.65244184846275e-06, + "loss": 0.7824, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.000465244184846275, + "step": 1901 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6386, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.652019750106284e-06, + "loss": 0.7428, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046520197501062843, + "step": 1902 + }, + { + "epoch": 0.19, + "gumbel_tau": 0.6384, + "image_masks_loss": 0.0288, + "llm_learning_rate": 4.6515974147646105e-06, + "loss": 0.8374, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.000465159741476461, + "step": 1903 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6383, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.6511748424842365e-06, + "loss": 0.8426, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0004651174842484236, + "step": 1904 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6381, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.6507520333116975e-06, + "loss": 0.7878, + "output_text_mask_loss": 0.014, + "predictor_learning_rate": 0.0004650752033311697, + "step": 1905 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.638, + "image_masks_loss": 0.0207, + "llm_learning_rate": 4.650328987293553e-06, + "loss": 0.829, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004650328987293553, + "step": 1906 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6378, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.649905704476392e-06, + "loss": 0.8267, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00046499057044763914, + "step": 1907 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6377, + "image_masks_loss": 0.0054, + "llm_learning_rate": 4.649482184906826e-06, + "loss": 0.7734, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046494821849068254, + "step": 1908 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6375, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.649058428631493e-06, + "loss": 0.7707, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0004649058428631493, + "step": 1909 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6374, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.648634435697059e-06, + "loss": 0.7676, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00046486344356970596, + "step": 1910 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6372, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.648210206150217e-06, + "loss": 0.7218, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00046482102061502165, + "step": 1911 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6371, + "image_masks_loss": 0.0288, + "llm_learning_rate": 4.64778574003768e-06, + "loss": 0.7617, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000464778574003768, + "step": 1912 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6369, + "image_masks_loss": 0.013, + "llm_learning_rate": 4.647361037406194e-06, + "loss": 0.7626, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00046473610374061944, + "step": 1913 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6368, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.646936098302528e-06, + "loss": 0.7335, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004646936098302528, + "step": 1914 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6366, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.6465109227734765e-06, + "loss": 0.7819, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004646510922773476, + "step": 1915 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6365, + "image_masks_loss": 0.0599, + "llm_learning_rate": 4.646085510865861e-06, + "loss": 0.8006, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00046460855108658607, + "step": 1916 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6363, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.645659862626528e-06, + "loss": 0.8174, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00046456598626265277, + "step": 1917 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6362, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.645233978102352e-06, + "loss": 0.6839, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004645233978102352, + "step": 1918 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.636, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.644807857340234e-06, + "loss": 0.6668, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004644807857340233, + "step": 1919 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6359, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.644381500387095e-06, + "loss": 0.8336, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004644381500387095, + "step": 1920 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6357, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.643954907289889e-06, + "loss": 0.6488, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004643954907289889, + "step": 1921 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6356, + "image_masks_loss": 0.036, + "llm_learning_rate": 4.643528078095594e-06, + "loss": 0.8235, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00046435280780955933, + "step": 1922 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6354, + "image_masks_loss": 0.0454, + "llm_learning_rate": 4.6431010128512126e-06, + "loss": 0.8309, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004643101012851212, + "step": 1923 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6353, + "image_masks_loss": 0.0401, + "llm_learning_rate": 4.642673711603773e-06, + "loss": 0.7337, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004642673711603773, + "step": 1924 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6351, + "image_masks_loss": 0.0064, + "llm_learning_rate": 4.6422461744003325e-06, + "loss": 0.7474, + "output_text_mask_loss": 0.0614, + "predictor_learning_rate": 0.0004642246174400332, + "step": 1925 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.635, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.641818401287972e-06, + "loss": 0.7113, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004641818401287972, + "step": 1926 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6348, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.641390392313797e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00046413903923137967, + "step": 1927 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6347, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.640962147524944e-06, + "loss": 0.7182, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0004640962147524943, + "step": 1928 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6345, + "image_masks_loss": 0.0637, + "llm_learning_rate": 4.640533666968569e-06, + "loss": 0.7912, + "output_text_mask_loss": 0.0351, + "predictor_learning_rate": 0.0004640533666968569, + "step": 1929 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6344, + "image_masks_loss": 0.0209, + "llm_learning_rate": 4.640104950691859e-06, + "loss": 0.6675, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0004640104950691859, + "step": 1930 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6342, + "image_masks_loss": 0.0035, + "llm_learning_rate": 4.639675998742026e-06, + "loss": 0.7403, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.0004639675998742026, + "step": 1931 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6341, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.639246811166306e-06, + "loss": 0.8189, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0004639246811166305, + "step": 1932 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6339, + "image_masks_loss": 0.0378, + "llm_learning_rate": 4.6388173880119605e-06, + "loss": 0.6893, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000463881738801196, + "step": 1933 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6338, + "image_masks_loss": 0.0057, + "llm_learning_rate": 4.6383877293262815e-06, + "loss": 0.7915, + "output_text_mask_loss": 0.0477, + "predictor_learning_rate": 0.0004638387729326281, + "step": 1934 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6336, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.6379578351565824e-06, + "loss": 0.7679, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004637957835156582, + "step": 1935 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6335, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.637527705550205e-06, + "loss": 0.6877, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004637527705550204, + "step": 1936 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6333, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.637097340554514e-06, + "loss": 0.7699, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0004637097340554514, + "step": 1937 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6332, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.636666740216905e-06, + "loss": 0.8087, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00046366667402169046, + "step": 1938 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.633, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.6362359045847945e-06, + "loss": 0.8202, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004636235904584794, + "step": 1939 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6329, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.635804833705628e-06, + "loss": 0.8123, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004635804833705628, + "step": 1940 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6327, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.635373527626876e-06, + "loss": 0.7841, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.0004635373527626876, + "step": 1941 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6326, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.634941986396035e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.0004634941986396035, + "step": 1942 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6324, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.634510210060628e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00046345102100606275, + "step": 1943 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6323, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.634078198668201e-06, + "loss": 0.7295, + "output_text_mask_loss": 0.0104, + "predictor_learning_rate": 0.00046340781986682, + "step": 1944 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6321, + "image_masks_loss": 0.0041, + "llm_learning_rate": 4.633645952266329e-06, + "loss": 0.7156, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046336459522663287, + "step": 1945 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.632, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.6332134709026135e-06, + "loss": 0.7388, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004633213470902613, + "step": 1946 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6319, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.632780754624678e-06, + "loss": 0.8127, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00046327807546246773, + "step": 1947 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6317, + "image_masks_loss": 0.013, + "llm_learning_rate": 4.632347803480175e-06, + "loss": 0.7353, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046323478034801746, + "step": 1948 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6316, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.631914617516783e-06, + "loss": 0.7644, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00046319146175167824, + "step": 1949 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6314, + "image_masks_loss": 0.0085, + "llm_learning_rate": 4.631481196782205e-06, + "loss": 0.6595, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.00046314811967822044, + "step": 1950 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6313, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.631047541324169e-06, + "loss": 0.779, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004631047541324169, + "step": 1951 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6311, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.630613651190431e-06, + "loss": 0.7195, + "output_text_mask_loss": 0.1823, + "predictor_learning_rate": 0.0004630613651190431, + "step": 1952 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.631, + "image_masks_loss": 0.0462, + "llm_learning_rate": 4.630179526428773e-06, + "loss": 0.7604, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00046301795264287726, + "step": 1953 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6308, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.6297451670869996e-06, + "loss": 0.7347, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004629745167086999, + "step": 1954 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6307, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.629310573212945e-06, + "loss": 0.7855, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.0004629310573212945, + "step": 1955 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6305, + "image_masks_loss": 0.0044, + "llm_learning_rate": 4.628875744854467e-06, + "loss": 0.8291, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.0004628875744854467, + "step": 1956 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6304, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.628440682059451e-06, + "loss": 0.6701, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.00046284406820594503, + "step": 1957 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6302, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.6280053848758054e-06, + "loss": 0.818, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004628005384875805, + "step": 1958 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6301, + "image_masks_loss": 0.024, + "llm_learning_rate": 4.627569853351468e-06, + "loss": 0.7294, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00046275698533514674, + "step": 1959 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6299, + "image_masks_loss": 0.0423, + "llm_learning_rate": 4.627134087534398e-06, + "loss": 0.7292, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00046271340875343975, + "step": 1960 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6298, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.626698087472585e-06, + "loss": 0.7366, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0004626698087472584, + "step": 1961 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6296, + "image_masks_loss": 0.0209, + "llm_learning_rate": 4.626261853214041e-06, + "loss": 0.5943, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046262618532140405, + "step": 1962 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6295, + "image_masks_loss": 0.0381, + "llm_learning_rate": 4.625825384806806e-06, + "loss": 0.7493, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004625825384806806, + "step": 1963 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6293, + "image_masks_loss": 0.0393, + "llm_learning_rate": 4.625388682298945e-06, + "loss": 0.8167, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004625388682298944, + "step": 1964 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6292, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.624951745738547e-06, + "loss": 0.7637, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004624951745738547, + "step": 1965 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.629, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.62451457517373e-06, + "loss": 0.8527, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.000462451457517373, + "step": 1966 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6289, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.624077170652637e-06, + "loss": 0.7203, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00046240771706526363, + "step": 1967 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6287, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.623639532223434e-06, + "loss": 0.8228, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.0004623639532223433, + "step": 1968 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6286, + "image_masks_loss": 0.0031, + "llm_learning_rate": 4.623201659934314e-06, + "loss": 0.7492, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004623201659934314, + "step": 1969 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6284, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.6227635538335e-06, + "loss": 0.6951, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00046227635538335, + "step": 1970 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6283, + "image_masks_loss": 0.0454, + "llm_learning_rate": 4.622325213969234e-06, + "loss": 0.7705, + "output_text_mask_loss": 0.0387, + "predictor_learning_rate": 0.0004622325213969234, + "step": 1971 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6281, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.62188664038979e-06, + "loss": 0.6579, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.0004621886640389789, + "step": 1972 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.628, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.621447833143461e-06, + "loss": 0.8296, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00046214478331434606, + "step": 1973 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6278, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.621008792278572e-06, + "loss": 0.9745, + "output_text_mask_loss": 0.161, + "predictor_learning_rate": 0.00046210087922785714, + "step": 1974 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6277, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.620569517843469e-06, + "loss": 0.9967, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.00046205695178434694, + "step": 1975 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6275, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.620130009886529e-06, + "loss": 1.2464, + "output_text_mask_loss": 0.3525, + "predictor_learning_rate": 0.0004620130009886529, + "step": 1976 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6274, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.61969026845615e-06, + "loss": 0.8836, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00046196902684561493, + "step": 1977 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6272, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.619250293600756e-06, + "loss": 0.9582, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.00046192502936007564, + "step": 1978 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6271, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.6188100853688005e-06, + "loss": 0.9284, + "output_text_mask_loss": 0.2472, + "predictor_learning_rate": 0.00046188100853688, + "step": 1979 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.627, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.618369643808758e-06, + "loss": 0.8684, + "output_text_mask_loss": 0.1038, + "predictor_learning_rate": 0.00046183696438087575, + "step": 1980 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6268, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.617928968969132e-06, + "loss": 0.8475, + "output_text_mask_loss": 0.2182, + "predictor_learning_rate": 0.00046179289689691317, + "step": 1981 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6267, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.617488060898451e-06, + "loss": 0.8047, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00046174880608984504, + "step": 1982 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6265, + "image_masks_loss": 0.0353, + "llm_learning_rate": 4.617046919645267e-06, + "loss": 0.7812, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.0004617046919645267, + "step": 1983 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6264, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.616605545258161e-06, + "loss": 0.7637, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00046166055452581613, + "step": 1984 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6262, + "image_masks_loss": 0.0219, + "llm_learning_rate": 4.616163937785739e-06, + "loss": 0.7238, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004616163937785739, + "step": 1985 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6261, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.6157220972766305e-06, + "loss": 0.8106, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.000461572209727663, + "step": 1986 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6259, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.615280023779491e-06, + "loss": 0.8264, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.0004615280023779491, + "step": 1987 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6258, + "image_masks_loss": 0.0182, + "llm_learning_rate": 4.614837717343004e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004614837717343003, + "step": 1988 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6256, + "image_masks_loss": 0.0311, + "llm_learning_rate": 4.614395178015877e-06, + "loss": 0.823, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00046143951780158764, + "step": 1989 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6255, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.6139524058468435e-06, + "loss": 0.7671, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.0004613952405846843, + "step": 1990 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6253, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.6135094008846626e-06, + "loss": 0.785, + "output_text_mask_loss": 0.1228, + "predictor_learning_rate": 0.00046135094008846624, + "step": 1991 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6252, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.6130661631781184e-06, + "loss": 0.7555, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.0004613066163178118, + "step": 1992 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.625, + "image_masks_loss": 0.0702, + "llm_learning_rate": 4.612622692776021e-06, + "loss": 0.7289, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004612622692776021, + "step": 1993 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6249, + "image_masks_loss": 0.0141, + "llm_learning_rate": 4.612178989727209e-06, + "loss": 0.8095, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004612178989727208, + "step": 1994 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6247, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.61173505408054e-06, + "loss": 0.8424, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00046117350540805395, + "step": 1995 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6246, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.611290885884904e-06, + "loss": 0.743, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00046112908858849034, + "step": 1996 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6244, + "image_masks_loss": 0.0068, + "llm_learning_rate": 4.6108464851892135e-06, + "loss": 0.7474, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004610846485189213, + "step": 1997 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6243, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.6104018520424045e-06, + "loss": 0.8592, + "output_text_mask_loss": 0.0847, + "predictor_learning_rate": 0.00046104018520424047, + "step": 1998 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.6241, + "image_masks_loss": 0.0111, + "llm_learning_rate": 4.6099569864934445e-06, + "loss": 0.7868, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.00046099569864934445, + "step": 1999 + }, + { + "epoch": 0.2, + "gumbel_tau": 0.624, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.609511888591321e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.1572, + "predictor_learning_rate": 0.0004609511888591321, + "step": 2000 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6239, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.609066558385049e-06, + "loss": 0.8651, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004609066558385049, + "step": 2001 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6237, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.60862099592367e-06, + "loss": 0.8154, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00046086209959236705, + "step": 2002 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6236, + "image_masks_loss": 0.0204, + "llm_learning_rate": 4.608175201256251e-06, + "loss": 0.8029, + "output_text_mask_loss": 0.1244, + "predictor_learning_rate": 0.00046081752012562505, + "step": 2003 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6234, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.607729174431882e-06, + "loss": 0.8088, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004607729174431882, + "step": 2004 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6233, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.607282915499682e-06, + "loss": 0.794, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004607282915499682, + "step": 2005 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6231, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.606836424508794e-06, + "loss": 0.8047, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00046068364245087937, + "step": 2006 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.623, + "image_masks_loss": 0.0192, + "llm_learning_rate": 4.606389701508386e-06, + "loss": 0.8041, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00046063897015083855, + "step": 2007 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6228, + "image_masks_loss": 0.0037, + "llm_learning_rate": 4.605942746547652e-06, + "loss": 0.8262, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004605942746547651, + "step": 2008 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6227, + "image_masks_loss": 0.0086, + "llm_learning_rate": 4.605495559675811e-06, + "loss": 0.8649, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 0.0004605495559675811, + "step": 2009 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6225, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.60504814094211e-06, + "loss": 0.7712, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000460504814094211, + "step": 2010 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6224, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.60460049039582e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0005, + "predictor_learning_rate": 0.00046046004903958196, + "step": 2011 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6222, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.604152608086235e-06, + "loss": 0.7327, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00046041526080862344, + "step": 2012 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6221, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.603704494062678e-06, + "loss": 0.7332, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00046037044940626777, + "step": 2013 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6219, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.603256148374496e-06, + "loss": 0.7475, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004603256148374496, + "step": 2014 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6218, + "image_masks_loss": 0.0069, + "llm_learning_rate": 4.6028075710710625e-06, + "loss": 0.7597, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00046028075710710626, + "step": 2015 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6216, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.602358762201776e-06, + "loss": 0.7957, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004602358762201776, + "step": 2016 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6215, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.6019097218160594e-06, + "loss": 0.7887, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00046019097218160586, + "step": 2017 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6214, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.601460449963362e-06, + "loss": 0.6686, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004601460449963362, + "step": 2018 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6212, + "image_masks_loss": 0.0236, + "llm_learning_rate": 4.6010109466931594e-06, + "loss": 0.7805, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00046010109466931593, + "step": 2019 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6211, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.600561212054951e-06, + "loss": 0.8403, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004600561212054951, + "step": 2020 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6209, + "image_masks_loss": 0.05, + "llm_learning_rate": 4.600111246098264e-06, + "loss": 0.7672, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004600111246098264, + "step": 2021 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6208, + "image_masks_loss": 0.003, + "llm_learning_rate": 4.599661048872648e-06, + "loss": 0.8675, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004599661048872648, + "step": 2022 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6206, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.599210620427681e-06, + "loss": 0.7824, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00045992106204276813, + "step": 2023 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6205, + "image_masks_loss": 0.0152, + "llm_learning_rate": 4.598759960812965e-06, + "loss": 0.8125, + "output_text_mask_loss": 0.0197, + "predictor_learning_rate": 0.0004598759960812965, + "step": 2024 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6203, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.598309070078127e-06, + "loss": 0.7318, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004598309070078126, + "step": 2025 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6202, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.59785794827282e-06, + "loss": 0.7552, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.000459785794827282, + "step": 2026 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.62, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.597406595446723e-06, + "loss": 0.8186, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0004597406595446723, + "step": 2027 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6199, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.59695501164954e-06, + "loss": 0.7973, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045969550116495403, + "step": 2028 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6197, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.596503196931e-06, + "loss": 0.8485, + "output_text_mask_loss": 0.0538, + "predictor_learning_rate": 0.0004596503196931, + "step": 2029 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6196, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.596051151340859e-06, + "loss": 0.7496, + "output_text_mask_loss": 0.1701, + "predictor_learning_rate": 0.0004596051151340859, + "step": 2030 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6195, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.595598874928896e-06, + "loss": 0.749, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045955988749288957, + "step": 2031 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6193, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.595146367744917e-06, + "loss": 0.7508, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004595146367744917, + "step": 2032 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6192, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.594693629838754e-06, + "loss": 0.759, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.0004594693629838754, + "step": 2033 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.619, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.594240661260262e-06, + "loss": 0.8007, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045942406612602615, + "step": 2034 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6189, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.593787462059324e-06, + "loss": 0.7745, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00045937874620593236, + "step": 2035 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6187, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.593334032285847e-06, + "loss": 0.8144, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0004593334032285846, + "step": 2036 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6186, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.592880371989763e-06, + "loss": 0.7645, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004592880371989763, + "step": 2037 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6184, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.592426481221031e-06, + "loss": 0.7709, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00045924264812210303, + "step": 2038 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6183, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.591972360029634e-06, + "loss": 0.7962, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.0004591972360029634, + "step": 2039 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6181, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.591518008465582e-06, + "loss": 0.6216, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045915180084655816, + "step": 2040 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.618, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.591063426578907e-06, + "loss": 0.795, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004591063426578907, + "step": 2041 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6178, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.590608614419671e-06, + "loss": 0.6962, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00045906086144196703, + "step": 2042 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6177, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.590153572037957e-06, + "loss": 0.7267, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045901535720379563, + "step": 2043 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6176, + "image_masks_loss": 0.0051, + "llm_learning_rate": 4.589698299483875e-06, + "loss": 0.6184, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045896982994838757, + "step": 2044 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6174, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.589242796807564e-06, + "loss": 0.7999, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00045892427968075635, + "step": 2045 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6173, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.588787064059181e-06, + "loss": 0.8566, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.00045887870640591805, + "step": 2046 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6171, + "image_masks_loss": 0.0433, + "llm_learning_rate": 4.588331101288914e-06, + "loss": 0.7565, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004588331101288914, + "step": 2047 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.617, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.587874908546975e-06, + "loss": 0.8659, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 0.00045878749085469743, + "step": 2048 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6168, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.5874184858836006e-06, + "loss": 0.7426, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00045874184858836, + "step": 2049 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6167, + "image_masks_loss": 0.0147, + "llm_learning_rate": 4.586961833349052e-06, + "loss": 0.7834, + "output_text_mask_loss": 0.0889, + "predictor_learning_rate": 0.00045869618333490523, + "step": 2050 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6165, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.586504950993619e-06, + "loss": 0.8424, + "output_text_mask_loss": 0.0755, + "predictor_learning_rate": 0.00045865049509936185, + "step": 2051 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6164, + "image_masks_loss": 0.0414, + "llm_learning_rate": 4.586047838867612e-06, + "loss": 0.7304, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00045860478388676126, + "step": 2052 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6162, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.585590497021372e-06, + "loss": 0.8355, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00045855904970213714, + "step": 2053 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6161, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.5851329255052605e-06, + "loss": 0.8677, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.000458513292550526, + "step": 2054 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.616, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.584675124369667e-06, + "loss": 0.7757, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0004584675124369666, + "step": 2055 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6158, + "image_masks_loss": 0.0425, + "llm_learning_rate": 4.584217093665004e-06, + "loss": 0.6602, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004584217093665004, + "step": 2056 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6157, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.583758833441713e-06, + "loss": 0.8237, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045837588334417137, + "step": 2057 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6155, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.5833003437502585e-06, + "loss": 0.8393, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.00045833003437502584, + "step": 2058 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6154, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.582841624641129e-06, + "loss": 0.6173, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00045828416246411285, + "step": 2059 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6152, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.582382676164842e-06, + "loss": 0.8038, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.0004582382676164841, + "step": 2060 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6151, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.581923498371934e-06, + "loss": 0.8323, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.0004581923498371934, + "step": 2061 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6149, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.581464091312974e-06, + "loss": 0.8432, + "output_text_mask_loss": 0.05, + "predictor_learning_rate": 0.0004581464091312974, + "step": 2062 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6148, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.581004455038552e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00045810044550385524, + "step": 2063 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6146, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.5805445895992854e-06, + "loss": 0.6484, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004580544589599285, + "step": 2064 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6145, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.580084495045813e-06, + "loss": 0.8305, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004580084495045813, + "step": 2065 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6144, + "image_masks_loss": 0.0195, + "llm_learning_rate": 4.5796241714288035e-06, + "loss": 0.709, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00045796241714288026, + "step": 2066 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6142, + "image_masks_loss": 0.0077, + "llm_learning_rate": 4.579163618798948e-06, + "loss": 0.8121, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004579163618798947, + "step": 2067 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6141, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.578702837206962e-06, + "loss": 0.7635, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.0004578702837206962, + "step": 2068 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6139, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.578241826703591e-06, + "loss": 0.8095, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.0004578241826703591, + "step": 2069 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6138, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.5777805873396005e-06, + "loss": 0.744, + "output_text_mask_loss": 0.0507, + "predictor_learning_rate": 0.00045777805873396007, + "step": 2070 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6136, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.577319119165784e-06, + "loss": 0.882, + "output_text_mask_loss": 0.0927, + "predictor_learning_rate": 0.0004577319119165784, + "step": 2071 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6135, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.5768574222329596e-06, + "loss": 0.7455, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 0.0004576857422232959, + "step": 2072 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6133, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.576395496591969e-06, + "loss": 0.769, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004576395496591969, + "step": 2073 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6132, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.575933342293681e-06, + "loss": 0.7276, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004575933342293681, + "step": 2074 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6131, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.5754709593889904e-06, + "loss": 0.8636, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00045754709593889896, + "step": 2075 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6129, + "image_masks_loss": 0.0063, + "llm_learning_rate": 4.575008347928814e-06, + "loss": 0.7418, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004575008347928814, + "step": 2076 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6128, + "image_masks_loss": 0.0084, + "llm_learning_rate": 4.574545507964096e-06, + "loss": 0.7395, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.0004574545507964096, + "step": 2077 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6126, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.574082439545807e-06, + "loss": 0.7226, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004574082439545807, + "step": 2078 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6125, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.57361914272494e-06, + "loss": 0.8187, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.00045736191427249396, + "step": 2079 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6123, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.573155617552514e-06, + "loss": 0.6945, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004573155617552513, + "step": 2080 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6122, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.572691864079572e-06, + "loss": 0.7996, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004572691864079572, + "step": 2081 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.612, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.572227882357186e-06, + "loss": 0.7665, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.00045722278823571863, + "step": 2082 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6119, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.571763672436451e-06, + "loss": 0.7006, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004571763672436451, + "step": 2083 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6118, + "image_masks_loss": 0.0234, + "llm_learning_rate": 4.571299234368485e-06, + "loss": 0.7877, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00045712992343684844, + "step": 2084 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6116, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.570834568204434e-06, + "loss": 0.7933, + "output_text_mask_loss": 0.0782, + "predictor_learning_rate": 0.0004570834568204434, + "step": 2085 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6115, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.570369673995467e-06, + "loss": 0.7627, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004570369673995467, + "step": 2086 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6113, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.569904551792781e-06, + "loss": 0.7666, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045699045517927806, + "step": 2087 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6112, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.569439201647594e-06, + "loss": 0.7218, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 0.0004569439201647594, + "step": 2088 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.611, + "image_masks_loss": 0.006, + "llm_learning_rate": 4.568973623611154e-06, + "loss": 0.8525, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.00045689736236111534, + "step": 2089 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6109, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.5685078177347285e-06, + "loss": 0.7225, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 0.00045685078177347285, + "step": 2090 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6107, + "image_masks_loss": 0.0069, + "llm_learning_rate": 4.568041784069616e-06, + "loss": 0.6838, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.0004568041784069615, + "step": 2091 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6106, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.567575522667135e-06, + "loss": 0.783, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045675755226671344, + "step": 2092 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6105, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.567109033578633e-06, + "loss": 0.8189, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00045671090335786327, + "step": 2093 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6103, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.566642316855479e-06, + "loss": 0.7376, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004566642316855479, + "step": 2094 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6102, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.56617537254907e-06, + "loss": 0.7945, + "output_text_mask_loss": 0.0839, + "predictor_learning_rate": 0.00045661753725490694, + "step": 2095 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.61, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.565708200710827e-06, + "loss": 0.6961, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0004565708200710827, + "step": 2096 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6099, + "image_masks_loss": 0.0523, + "llm_learning_rate": 4.565240801392196e-06, + "loss": 0.8684, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004565240801392195, + "step": 2097 + }, + { + "epoch": 0.21, + "gumbel_tau": 0.6097, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.564773174644647e-06, + "loss": 0.8021, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045647731746446476, + "step": 2098 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6096, + "image_masks_loss": 0.0051, + "llm_learning_rate": 4.564305320519677e-06, + "loss": 0.8586, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004564305320519677, + "step": 2099 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6095, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.563837239068808e-06, + "loss": 0.809, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.00045638372390688077, + "step": 2100 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6093, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.5633689303435845e-06, + "loss": 0.6338, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045633689303435843, + "step": 2101 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6092, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.562900394395579e-06, + "loss": 0.7228, + "output_text_mask_loss": 0.0847, + "predictor_learning_rate": 0.0004562900394395579, + "step": 2102 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.609, + "image_masks_loss": 0.0084, + "llm_learning_rate": 4.562431631276387e-06, + "loss": 0.8062, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 0.0004562431631276386, + "step": 2103 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6089, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.5619626410376295e-06, + "loss": 0.8142, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004561962641037629, + "step": 2104 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6087, + "image_masks_loss": 0.0069, + "llm_learning_rate": 4.561493423730953e-06, + "loss": 0.7948, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.00045614934237309525, + "step": 2105 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6086, + "image_masks_loss": 0.0347, + "llm_learning_rate": 4.561023979408029e-06, + "loss": 0.7723, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004561023979408029, + "step": 2106 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6084, + "image_masks_loss": 0.0189, + "llm_learning_rate": 4.560554308120554e-06, + "loss": 0.8695, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.0004560554308120554, + "step": 2107 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6083, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.560084409920248e-06, + "loss": 0.8155, + "output_text_mask_loss": 0.0108, + "predictor_learning_rate": 0.00045600844099202484, + "step": 2108 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6082, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.559614284858859e-06, + "loss": 0.7628, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004559614284858859, + "step": 2109 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.608, + "image_masks_loss": 0.0182, + "llm_learning_rate": 4.559143932988157e-06, + "loss": 0.786, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004559143932988157, + "step": 2110 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6079, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.55867335435994e-06, + "loss": 0.8731, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0004558673354359939, + "step": 2111 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6077, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.558202549026025e-06, + "loss": 0.8054, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004558202549026025, + "step": 2112 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6076, + "image_masks_loss": 0.0071, + "llm_learning_rate": 4.5577315170382625e-06, + "loss": 0.7344, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.00045577315170382623, + "step": 2113 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6074, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.557260258448522e-06, + "loss": 0.8388, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00045572602584485213, + "step": 2114 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6073, + "image_masks_loss": 0.036, + "llm_learning_rate": 4.556788773308699e-06, + "loss": 0.8019, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00045567887733086984, + "step": 2115 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6072, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.556317061670714e-06, + "loss": 0.8226, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00045563170616707145, + "step": 2116 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.607, + "image_masks_loss": 0.0242, + "llm_learning_rate": 4.555845123586515e-06, + "loss": 0.8165, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0004555845123586515, + "step": 2117 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6069, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.555372959108072e-06, + "loss": 0.7365, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0004555372959108072, + "step": 2118 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6067, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.55490056828738e-06, + "loss": 0.8054, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.000455490056828738, + "step": 2119 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6066, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.554427951176461e-06, + "loss": 0.7786, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.00045544279511764606, + "step": 2120 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6064, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.55395510782736e-06, + "loss": 0.8288, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.0004553955107827359, + "step": 2121 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6063, + "image_masks_loss": 0.0519, + "llm_learning_rate": 4.553482038292147e-06, + "loss": 0.7905, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.00045534820382921463, + "step": 2122 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6062, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.553008742622918e-06, + "loss": 0.6404, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004553008742622918, + "step": 2123 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.606, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.552535220871793e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0004552535220871793, + "step": 2124 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6059, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.552061473090918e-06, + "loss": 0.7536, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.00045520614730909184, + "step": 2125 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6057, + "image_masks_loss": 0.0222, + "llm_learning_rate": 4.551587499332463e-06, + "loss": 0.7855, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0004551587499332463, + "step": 2126 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6056, + "image_masks_loss": 0.0406, + "llm_learning_rate": 4.551113299648623e-06, + "loss": 0.7864, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045511132996486225, + "step": 2127 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6054, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.550638874091617e-06, + "loss": 0.7627, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004550638874091617, + "step": 2128 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6053, + "image_masks_loss": 0.024, + "llm_learning_rate": 4.5501642227136914e-06, + "loss": 0.7731, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045501642227136906, + "step": 2129 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6052, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.5496893455671144e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004549689345567114, + "step": 2130 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.605, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.549214242704182e-06, + "loss": 0.7924, + "output_text_mask_loss": 0.0412, + "predictor_learning_rate": 0.00045492142427041817, + "step": 2131 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6049, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.548738914177212e-06, + "loss": 0.7384, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00045487389141772117, + "step": 2132 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6047, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.5482633600385495e-06, + "loss": 0.7766, + "output_text_mask_loss": 0.0303, + "predictor_learning_rate": 0.0004548263360038549, + "step": 2133 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6046, + "image_masks_loss": 0.0097, + "llm_learning_rate": 4.547787580340563e-06, + "loss": 0.7307, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.00045477875803405634, + "step": 2134 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6044, + "image_masks_loss": 0.0241, + "llm_learning_rate": 4.547311575135648e-06, + "loss": 0.7841, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.00045473115751356474, + "step": 2135 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6043, + "image_masks_loss": 0.0079, + "llm_learning_rate": 4.546835344476221e-06, + "loss": 0.7406, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004546835344476221, + "step": 2136 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6042, + "image_masks_loss": 0.0351, + "llm_learning_rate": 4.546358888414728e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00045463588884147275, + "step": 2137 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.604, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.545882207003635e-06, + "loss": 0.7726, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0004545882207003635, + "step": 2138 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6039, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.5454053002954365e-06, + "loss": 0.6998, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00045454053002954366, + "step": 2139 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6037, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.544928168342651e-06, + "loss": 0.7648, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004544928168342651, + "step": 2140 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6036, + "image_masks_loss": 0.0542, + "llm_learning_rate": 4.54445081119782e-06, + "loss": 0.736, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.000454445081119782, + "step": 2141 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6034, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.543973228913512e-06, + "loss": 0.7428, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00045439732289135124, + "step": 2142 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6033, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.54349542154232e-06, + "loss": 0.6845, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.000454349542154232, + "step": 2143 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6032, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.54301738913686e-06, + "loss": 0.72, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.000454301738913686, + "step": 2144 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.603, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.542539131749774e-06, + "loss": 0.8133, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004542539131749773, + "step": 2145 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6029, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.542060649433729e-06, + "loss": 0.8266, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.0004542060649433729, + "step": 2146 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6027, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.541581942241418e-06, + "loss": 0.8977, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 0.00045415819422414173, + "step": 2147 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6026, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.541103010225555e-06, + "loss": 0.7606, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00045411030102255543, + "step": 2148 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6024, + "image_masks_loss": 0.0376, + "llm_learning_rate": 4.540623853438883e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.0004540623853438883, + "step": 2149 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6023, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.540144471934166e-06, + "loss": 0.821, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0004540144471934166, + "step": 2150 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6022, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.539664865764196e-06, + "loss": 0.7819, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004539664865764196, + "step": 2151 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.602, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.539185034981788e-06, + "loss": 0.8314, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00045391850349817874, + "step": 2152 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6019, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.5387049796397816e-06, + "loss": 0.8469, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.00045387049796397815, + "step": 2153 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6017, + "image_masks_loss": 0.0458, + "llm_learning_rate": 4.5382246997910415e-06, + "loss": 0.8337, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00045382246997910413, + "step": 2154 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6016, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.537744195488459e-06, + "loss": 0.8085, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0004537744195488458, + "step": 2155 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6015, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.537263466784945e-06, + "loss": 0.745, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004537263466784945, + "step": 2156 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6013, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.536782513733441e-06, + "loss": 0.7836, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00045367825137334407, + "step": 2157 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6012, + "image_masks_loss": 0.0097, + "llm_learning_rate": 4.536301336386909e-06, + "loss": 0.7801, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.00045363013363869094, + "step": 2158 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.601, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.535819934798339e-06, + "loss": 0.7388, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00045358199347983393, + "step": 2159 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6009, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.535338309020743e-06, + "loss": 0.8531, + "output_text_mask_loss": 0.0523, + "predictor_learning_rate": 0.00045353383090207435, + "step": 2160 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6007, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.53485645910716e-06, + "loss": 0.7522, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 0.00045348564591071597, + "step": 2161 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6006, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.5343743851106495e-06, + "loss": 0.7653, + "output_text_mask_loss": 0.0406, + "predictor_learning_rate": 0.0004534374385110649, + "step": 2162 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6005, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.533892087084301e-06, + "loss": 0.8447, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 0.00045338920870843005, + "step": 2163 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6003, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.533409565081226e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 0.00045334095650812254, + "step": 2164 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6002, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.532926819154559e-06, + "loss": 0.7244, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004532926819154559, + "step": 2165 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.6, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.532443849357464e-06, + "loss": 0.7281, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00045324438493574636, + "step": 2166 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5999, + "image_masks_loss": 0.013, + "llm_learning_rate": 4.531960655743125e-06, + "loss": 0.8622, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.0004531960655743125, + "step": 2167 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5998, + "image_masks_loss": 0.0075, + "llm_learning_rate": 4.531477238364752e-06, + "loss": 0.7779, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.0004531477238364752, + "step": 2168 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5996, + "image_masks_loss": 0.0186, + "llm_learning_rate": 4.530993597275581e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00045309935972755806, + "step": 2169 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5995, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.530509732528871e-06, + "loss": 0.681, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.000453050973252887, + "step": 2170 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5993, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.530025644177906e-06, + "loss": 0.7992, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0004530025644177906, + "step": 2171 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5992, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.529541332275996e-06, + "loss": 0.8162, + "output_text_mask_loss": 0.045, + "predictor_learning_rate": 0.0004529541332275996, + "step": 2172 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.599, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.529056796876474e-06, + "loss": 0.7369, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.00045290567968764734, + "step": 2173 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5989, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.528572038032698e-06, + "loss": 0.7537, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.0004528572038032697, + "step": 2174 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5988, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.52808705579805e-06, + "loss": 0.8122, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045280870557980494, + "step": 2175 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5986, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.527601850225939e-06, + "loss": 0.8558, + "output_text_mask_loss": 0.0328, + "predictor_learning_rate": 0.0004527601850225938, + "step": 2176 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5985, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.527116421369795e-06, + "loss": 0.7572, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00045271164213697945, + "step": 2177 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5983, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.526630769283076e-06, + "loss": 0.6239, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045266307692830754, + "step": 2178 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5982, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.526144894019262e-06, + "loss": 0.865, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.00045261448940192616, + "step": 2179 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5981, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.52565879563186e-06, + "loss": 0.8553, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.00045256587956318594, + "step": 2180 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5979, + "image_masks_loss": 0.0108, + "llm_learning_rate": 4.525172474174398e-06, + "loss": 0.7905, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004525172474174398, + "step": 2181 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5978, + "image_masks_loss": 0.0228, + "llm_learning_rate": 4.5246859297004346e-06, + "loss": 0.8034, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.0004524685929700434, + "step": 2182 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5976, + "image_masks_loss": 0.0034, + "llm_learning_rate": 4.524199162263545e-06, + "loss": 0.7668, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00045241991622635446, + "step": 2183 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5975, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.523712171917335e-06, + "loss": 0.6922, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004523712171917335, + "step": 2184 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5974, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.523224958715434e-06, + "loss": 0.7725, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00045232249587154333, + "step": 2185 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5972, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.5227375227114936e-06, + "loss": 0.7196, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004522737522711493, + "step": 2186 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5971, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.522249863959191e-06, + "loss": 0.7723, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004522249863959191, + "step": 2187 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5969, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.52176198251223e-06, + "loss": 0.7396, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045217619825122294, + "step": 2188 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5968, + "image_masks_loss": 0.0057, + "llm_learning_rate": 4.521273878424335e-06, + "loss": 0.7545, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0004521273878424335, + "step": 2189 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5966, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.52078555174926e-06, + "loss": 0.8278, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.0004520785551749259, + "step": 2190 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5965, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.520297002540778e-06, + "loss": 0.7319, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 0.00045202970025407776, + "step": 2191 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5964, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.519808230852689e-06, + "loss": 0.751, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.0004519808230852689, + "step": 2192 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5962, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.51931923673882e-06, + "loss": 0.6923, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000451931923673882, + "step": 2193 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5961, + "image_masks_loss": 0.0546, + "llm_learning_rate": 4.518830020253018e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004518830020253018, + "step": 2194 + }, + { + "epoch": 0.22, + "gumbel_tau": 0.5959, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.518340581449158e-06, + "loss": 0.7785, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004518340581449158, + "step": 2195 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5958, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.517850920381138e-06, + "loss": 0.7218, + "output_text_mask_loss": 0.1129, + "predictor_learning_rate": 0.00045178509203811374, + "step": 2196 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5957, + "image_masks_loss": 0.0311, + "llm_learning_rate": 4.517361037102878e-06, + "loss": 0.8019, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00045173610371028786, + "step": 2197 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5955, + "image_masks_loss": 0.0072, + "llm_learning_rate": 4.51687093166833e-06, + "loss": 0.7321, + "output_text_mask_loss": 0.0441, + "predictor_learning_rate": 0.0004516870931668329, + "step": 2198 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5954, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.51638060413146e-06, + "loss": 0.7518, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00045163806041314603, + "step": 2199 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5952, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.5158900545462684e-06, + "loss": 0.8303, + "output_text_mask_loss": 0.1968, + "predictor_learning_rate": 0.0004515890054546268, + "step": 2200 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5951, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.515399282966773e-06, + "loss": 0.7797, + "output_text_mask_loss": 0.0128, + "predictor_learning_rate": 0.00045153992829667725, + "step": 2201 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.595, + "image_masks_loss": 0.042, + "llm_learning_rate": 4.51490828944702e-06, + "loss": 0.8304, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004514908289447019, + "step": 2202 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5948, + "image_masks_loss": 0.0694, + "llm_learning_rate": 4.514417074041077e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0004514417074041077, + "step": 2203 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5947, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.513925636803039e-06, + "loss": 0.7247, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.0004513925636803039, + "step": 2204 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5945, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.5134339777870255e-06, + "loss": 0.8306, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0004513433977787025, + "step": 2205 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5944, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.512942097047176e-06, + "loss": 0.7265, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0004512942097047176, + "step": 2206 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5943, + "image_masks_loss": 0.0401, + "llm_learning_rate": 4.51244999463766e-06, + "loss": 0.77, + "output_text_mask_loss": 0.0816, + "predictor_learning_rate": 0.000451244999463766, + "step": 2207 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5941, + "image_masks_loss": 0.0416, + "llm_learning_rate": 4.5119576706126675e-06, + "loss": 0.788, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00045119576706126676, + "step": 2208 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.594, + "image_masks_loss": 0.0725, + "llm_learning_rate": 4.511465125026415e-06, + "loss": 0.8023, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004511465125026415, + "step": 2209 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5938, + "image_masks_loss": 0.0219, + "llm_learning_rate": 4.510972357933143e-06, + "loss": 0.8142, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0004510972357933143, + "step": 2210 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5937, + "image_masks_loss": 0.0147, + "llm_learning_rate": 4.5104793693871155e-06, + "loss": 0.835, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0004510479369387115, + "step": 2211 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5936, + "image_masks_loss": 0.0229, + "llm_learning_rate": 4.509986159442621e-06, + "loss": 0.8014, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00045099861594426205, + "step": 2212 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5934, + "image_masks_loss": 0.0519, + "llm_learning_rate": 4.509492728153975e-06, + "loss": 0.8461, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004509492728153974, + "step": 2213 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5933, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.508999075575512e-06, + "loss": 0.7937, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0004508999075575512, + "step": 2214 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5931, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.508505201761597e-06, + "loss": 0.7817, + "output_text_mask_loss": 0.1251, + "predictor_learning_rate": 0.00045085052017615967, + "step": 2215 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.593, + "image_masks_loss": 0.0477, + "llm_learning_rate": 4.508011106766616e-06, + "loss": 0.781, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00045080111067666154, + "step": 2216 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5929, + "image_masks_loss": 0.0045, + "llm_learning_rate": 4.507516790644978e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00045075167906449777, + "step": 2217 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5927, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.50702225345112e-06, + "loss": 0.8316, + "output_text_mask_loss": 0.0011, + "predictor_learning_rate": 0.00045070222534511196, + "step": 2218 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5926, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.506527495239501e-06, + "loss": 0.7665, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004506527495239501, + "step": 2219 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5924, + "image_masks_loss": 0.0889, + "llm_learning_rate": 4.506032516064604e-06, + "loss": 0.9282, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00045060325160646043, + "step": 2220 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5923, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.50553731598094e-06, + "loss": 0.7499, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004505537315980939, + "step": 2221 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5922, + "image_masks_loss": 0.0652, + "llm_learning_rate": 4.505041895043038e-06, + "loss": 0.7806, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004505041895043037, + "step": 2222 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.592, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.504546253305457e-06, + "loss": 0.7129, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.0004504546253305457, + "step": 2223 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5919, + "image_masks_loss": 0.0546, + "llm_learning_rate": 4.504050390822777e-06, + "loss": 0.7107, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.0004504050390822777, + "step": 2224 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5917, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.503554307649605e-06, + "loss": 0.8096, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004503554307649604, + "step": 2225 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5916, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.503058003840569e-06, + "loss": 0.789, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.0004503058003840569, + "step": 2226 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5915, + "image_masks_loss": 0.0469, + "llm_learning_rate": 4.5025614794503255e-06, + "loss": 0.8311, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004502561479450325, + "step": 2227 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5913, + "image_masks_loss": 0.0209, + "llm_learning_rate": 4.50206473453355e-06, + "loss": 0.6808, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.00045020647345335495, + "step": 2228 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5912, + "image_masks_loss": 0.0435, + "llm_learning_rate": 4.501567769144947e-06, + "loss": 0.7443, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0004501567769144947, + "step": 2229 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.591, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.501070583339243e-06, + "loss": 0.8526, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.00045010705833392427, + "step": 2230 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5909, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.500573177171189e-06, + "loss": 0.7896, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0004500573177171189, + "step": 2231 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5908, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.5000755506955605e-06, + "loss": 0.7318, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00045000755506955606, + "step": 2232 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5906, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.499577703967158e-06, + "loss": 0.7964, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00044995777039671573, + "step": 2233 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5905, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.499079637040804e-06, + "loss": 0.8116, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.0004499079637040804, + "step": 2234 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5903, + "image_masks_loss": 0.0345, + "llm_learning_rate": 4.498581349971348e-06, + "loss": 0.7007, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004498581349971348, + "step": 2235 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5902, + "image_masks_loss": 0.0492, + "llm_learning_rate": 4.498082842813662e-06, + "loss": 0.8179, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.0004498082842813662, + "step": 2236 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5901, + "image_masks_loss": 0.0063, + "llm_learning_rate": 4.497584115622643e-06, + "loss": 0.7563, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004497584115622643, + "step": 2237 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5899, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.497085168453211e-06, + "loss": 0.8242, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0004497085168453211, + "step": 2238 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5898, + "image_masks_loss": 0.0433, + "llm_learning_rate": 4.496586001360312e-06, + "loss": 0.8497, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004496586001360312, + "step": 2239 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5896, + "image_masks_loss": 0.0402, + "llm_learning_rate": 4.496086614398917e-06, + "loss": 0.6689, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.0004496086614398917, + "step": 2240 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5895, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.495587007624016e-06, + "loss": 0.8065, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0004495587007624016, + "step": 2241 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5894, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.4950871810906295e-06, + "loss": 0.774, + "output_text_mask_loss": 0.0591, + "predictor_learning_rate": 0.00044950871810906293, + "step": 2242 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5892, + "image_masks_loss": 0.0576, + "llm_learning_rate": 4.494587134853799e-06, + "loss": 0.7675, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00044945871348537985, + "step": 2243 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5891, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.49408686896859e-06, + "loss": 0.697, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.000449408686896859, + "step": 2244 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.589, + "image_masks_loss": 0.0242, + "llm_learning_rate": 4.493586383490093e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004493586383490093, + "step": 2245 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5888, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.493085678473423e-06, + "loss": 0.6832, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004493085678473423, + "step": 2246 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5887, + "image_masks_loss": 0.0214, + "llm_learning_rate": 4.49258475397372e-06, + "loss": 0.7126, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.000449258475397372, + "step": 2247 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5885, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.492083610046144e-06, + "loss": 0.6875, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 0.00044920836100461437, + "step": 2248 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5884, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.491582246745885e-06, + "loss": 0.75, + "output_text_mask_loss": 0.1305, + "predictor_learning_rate": 0.00044915822467458846, + "step": 2249 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5883, + "image_masks_loss": 0.0467, + "llm_learning_rate": 4.491080664128152e-06, + "loss": 0.738, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00044910806641281517, + "step": 2250 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5881, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.490578862248182e-06, + "loss": 0.8613, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00044905788622481813, + "step": 2251 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.588, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.4900768411612325e-06, + "loss": 0.8933, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.00044900768411612325, + "step": 2252 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5878, + "image_masks_loss": 0.0298, + "llm_learning_rate": 4.4895746009225895e-06, + "loss": 0.774, + "output_text_mask_loss": 0.0923, + "predictor_learning_rate": 0.00044895746009225893, + "step": 2253 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5877, + "image_masks_loss": 0.0399, + "llm_learning_rate": 4.489072141587559e-06, + "loss": 0.7512, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004489072141587559, + "step": 2254 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5876, + "image_masks_loss": 0.0208, + "llm_learning_rate": 4.488569463211475e-06, + "loss": 0.773, + "output_text_mask_loss": 0.1106, + "predictor_learning_rate": 0.0004488569463211475, + "step": 2255 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5874, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.488066565849692e-06, + "loss": 0.7547, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004488066565849692, + "step": 2256 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5873, + "image_masks_loss": 0.0049, + "llm_learning_rate": 4.487563449557591e-06, + "loss": 0.8011, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.000448756344955759, + "step": 2257 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5872, + "image_masks_loss": 0.0029, + "llm_learning_rate": 4.487060114390574e-06, + "loss": 0.7522, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00044870601143905743, + "step": 2258 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.587, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.486556560404073e-06, + "loss": 0.6695, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004486556560404073, + "step": 2259 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5869, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.486052787653538e-06, + "loss": 0.7091, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004486052787653537, + "step": 2260 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5867, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.485548796194447e-06, + "loss": 0.7843, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00044855487961944465, + "step": 2261 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5866, + "image_masks_loss": 0.024, + "llm_learning_rate": 4.4850445860822986e-06, + "loss": 0.7783, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004485044586082298, + "step": 2262 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5865, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.484540157372619e-06, + "loss": 0.824, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004484540157372619, + "step": 2263 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5863, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.484035510120958e-06, + "loss": 0.837, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00044840355101209576, + "step": 2264 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5862, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.483530644382886e-06, + "loss": 0.772, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.0004483530644382886, + "step": 2265 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.586, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.4830255602140025e-06, + "loss": 0.8232, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00044830255602140023, + "step": 2266 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5859, + "image_masks_loss": 0.0064, + "llm_learning_rate": 4.482520257669927e-06, + "loss": 0.755, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00044825202576699266, + "step": 2267 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5858, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.482014736806305e-06, + "loss": 0.7264, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004482014736806305, + "step": 2268 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5856, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.481508997678805e-06, + "loss": 0.7111, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004481508997678805, + "step": 2269 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5855, + "image_masks_loss": 0.0328, + "llm_learning_rate": 4.48100304034312e-06, + "loss": 0.7457, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.000448100304034312, + "step": 2270 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5854, + "image_masks_loss": 0.0192, + "llm_learning_rate": 4.480496864854968e-06, + "loss": 0.8173, + "output_text_mask_loss": 0.0626, + "predictor_learning_rate": 0.0004480496864854968, + "step": 2271 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5852, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.479990471270091e-06, + "loss": 0.7196, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 0.0004479990471270091, + "step": 2272 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5851, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.4794838596442525e-06, + "loss": 0.7869, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.00044794838596442523, + "step": 2273 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5849, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.4789770300332415e-06, + "loss": 0.6881, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004478977030033241, + "step": 2274 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5848, + "image_masks_loss": 0.0378, + "llm_learning_rate": 4.4784699824928725e-06, + "loss": 0.7183, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.0004478469982492872, + "step": 2275 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5847, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.477962717078982e-06, + "loss": 0.6747, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004477962717078982, + "step": 2276 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5845, + "image_masks_loss": 0.0553, + "llm_learning_rate": 4.4774552338474315e-06, + "loss": 0.8612, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004477455233847431, + "step": 2277 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5844, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.476947532854106e-06, + "loss": 0.7044, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.00044769475328541056, + "step": 2278 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5842, + "image_masks_loss": 0.0192, + "llm_learning_rate": 4.476439614154915e-06, + "loss": 0.7277, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00044764396141549145, + "step": 2279 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5841, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.4759314778057904e-06, + "loss": 0.6223, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00044759314778057904, + "step": 2280 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.584, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.475423123862691e-06, + "loss": 0.8403, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004475423123862691, + "step": 2281 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5838, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.474914552381597e-06, + "loss": 0.7578, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004474914552381597, + "step": 2282 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5837, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.474405763418514e-06, + "loss": 0.8089, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0004474405763418513, + "step": 2283 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5836, + "image_masks_loss": 0.0553, + "llm_learning_rate": 4.47389675702947e-06, + "loss": 0.8658, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00044738967570294696, + "step": 2284 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5834, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.4733875332705176e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00044733875332705176, + "step": 2285 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5833, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.472878092197736e-06, + "loss": 0.6864, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0004472878092197736, + "step": 2286 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5831, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.472368433867224e-06, + "loss": 0.7321, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004472368433867224, + "step": 2287 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.583, + "image_masks_loss": 0.0222, + "llm_learning_rate": 4.471858558335107e-06, + "loss": 0.8118, + "output_text_mask_loss": 0.0656, + "predictor_learning_rate": 0.0004471858558335106, + "step": 2288 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5829, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.471348465657532e-06, + "loss": 0.7591, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00044713484656575323, + "step": 2289 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5827, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.470838155890675e-06, + "loss": 0.8273, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004470838155890674, + "step": 2290 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5826, + "image_masks_loss": 0.0097, + "llm_learning_rate": 4.470327629090729e-06, + "loss": 0.7684, + "output_text_mask_loss": 0.0336, + "predictor_learning_rate": 0.00044703276290907287, + "step": 2291 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5825, + "image_masks_loss": 0.0061, + "llm_learning_rate": 4.4698168853139165e-06, + "loss": 0.7795, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0004469816885313916, + "step": 2292 + }, + { + "epoch": 0.23, + "gumbel_tau": 0.5823, + "image_masks_loss": 0.0315, + "llm_learning_rate": 4.46930592461648e-06, + "loss": 0.9366, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.000446930592461648, + "step": 2293 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5822, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.4687947470546895e-06, + "loss": 0.7279, + "output_text_mask_loss": 0.0786, + "predictor_learning_rate": 0.0004468794747054689, + "step": 2294 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.582, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.468283352684836e-06, + "loss": 0.8196, + "output_text_mask_loss": 0.0648, + "predictor_learning_rate": 0.0004468283352684835, + "step": 2295 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5819, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.467771741563235e-06, + "loss": 0.8462, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004467771741563235, + "step": 2296 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5818, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.467259913746228e-06, + "loss": 0.8376, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004467259913746228, + "step": 2297 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5816, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.466747869290176e-06, + "loss": 0.7645, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00044667478692901765, + "step": 2298 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5815, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.466235608251469e-06, + "loss": 0.8121, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0004466235608251469, + "step": 2299 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5814, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.465723130686517e-06, + "loss": 0.7293, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.00044657231306865163, + "step": 2300 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5812, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.465210436651755e-06, + "loss": 0.7912, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00044652104366517544, + "step": 2301 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5811, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.4646975262036415e-06, + "loss": 0.7544, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00044646975262036415, + "step": 2302 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.581, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.464184399398661e-06, + "loss": 0.8331, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00044641843993986613, + "step": 2303 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5808, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.46367105629332e-06, + "loss": 0.8101, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 0.00044636710562933197, + "step": 2304 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5807, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.4631574969441485e-06, + "loss": 0.7274, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.0004463157496944148, + "step": 2305 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5805, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.4626437214077e-06, + "loss": 0.7619, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00044626437214077, + "step": 2306 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5804, + "image_masks_loss": 0.05, + "llm_learning_rate": 4.462129729740553e-06, + "loss": 0.8839, + "output_text_mask_loss": 0.0165, + "predictor_learning_rate": 0.00044621297297405525, + "step": 2307 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5803, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.4616155219993105e-06, + "loss": 0.8056, + "output_text_mask_loss": 0.1183, + "predictor_learning_rate": 0.000446161552199931, + "step": 2308 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5801, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.461101098240597e-06, + "loss": 0.8876, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004461101098240596, + "step": 2309 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.58, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.460586458521062e-06, + "loss": 0.814, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004460586458521062, + "step": 2310 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5799, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.46007160289738e-06, + "loss": 0.7813, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0004460071602897379, + "step": 2311 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5797, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.459556531426246e-06, + "loss": 0.7023, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004459556531426246, + "step": 2312 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5796, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.459041244164383e-06, + "loss": 0.7922, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00044590412441643826, + "step": 2313 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5794, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.458525741168534e-06, + "loss": 0.8642, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0004458525741168534, + "step": 2314 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5793, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.458010022495468e-06, + "loss": 0.87, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.0004458010022495468, + "step": 2315 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5792, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.457494088201978e-06, + "loss": 0.8236, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00044574940882019775, + "step": 2316 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.579, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.456977938344877e-06, + "loss": 0.7835, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0004456977938344877, + "step": 2317 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5789, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.4564615729810085e-06, + "loss": 0.7015, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0004456461572981008, + "step": 2318 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5788, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.455944992167233e-06, + "loss": 0.6944, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00044559449921672325, + "step": 2319 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5786, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.4554281959604376e-06, + "loss": 0.7872, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 0.00044554281959604375, + "step": 2320 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5785, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.454911184417534e-06, + "loss": 0.6653, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0004454911184417534, + "step": 2321 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5784, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.454393957595457e-06, + "loss": 0.7204, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004454393957595457, + "step": 2322 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5782, + "image_masks_loss": 0.0387, + "llm_learning_rate": 4.453876515551164e-06, + "loss": 0.8207, + "output_text_mask_loss": 0.0349, + "predictor_learning_rate": 0.00044538765155511633, + "step": 2323 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5781, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.453358858341636e-06, + "loss": 0.7656, + "output_text_mask_loss": 0.0542, + "predictor_learning_rate": 0.0004453358858341636, + "step": 2324 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5779, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.452840986023881e-06, + "loss": 0.781, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.00044528409860238803, + "step": 2325 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5778, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.452322898654926e-06, + "loss": 0.7183, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00044523228986549255, + "step": 2326 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5777, + "image_masks_loss": 0.0108, + "llm_learning_rate": 4.4518045962918245e-06, + "loss": 0.7594, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 0.0004451804596291824, + "step": 2327 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5775, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.451286078991654e-06, + "loss": 0.7776, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.0004451286078991654, + "step": 2328 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5774, + "image_masks_loss": 0.0079, + "llm_learning_rate": 4.450767346811514e-06, + "loss": 0.7348, + "output_text_mask_loss": 0.0462, + "predictor_learning_rate": 0.0004450767346811514, + "step": 2329 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5773, + "image_masks_loss": 0.022, + "llm_learning_rate": 4.450248399808529e-06, + "loss": 0.7316, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00044502483998085286, + "step": 2330 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5771, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.449729238039846e-06, + "loss": 0.806, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004449729238039845, + "step": 2331 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.577, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.4492098615626355e-06, + "loss": 0.7749, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00044492098615626354, + "step": 2332 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5769, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.448690270434094e-06, + "loss": 0.76, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00044486902704340937, + "step": 2333 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5767, + "image_masks_loss": 0.0038, + "llm_learning_rate": 4.4481704647114395e-06, + "loss": 0.7709, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00044481704647114396, + "step": 2334 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5766, + "image_masks_loss": 0.0242, + "llm_learning_rate": 4.447650444451914e-06, + "loss": 0.8028, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0004447650444451914, + "step": 2335 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5764, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.447130209712783e-06, + "loss": 0.7508, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 0.0004447130209712783, + "step": 2336 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5763, + "image_masks_loss": 0.0383, + "llm_learning_rate": 4.446609760551337e-06, + "loss": 0.7627, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004446609760551337, + "step": 2337 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5762, + "image_masks_loss": 0.0298, + "llm_learning_rate": 4.446089097024889e-06, + "loss": 0.8252, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004446089097024888, + "step": 2338 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.576, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.4455682191907724e-06, + "loss": 0.8138, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004445568219190772, + "step": 2339 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5759, + "image_masks_loss": 0.0045, + "llm_learning_rate": 4.445047127106351e-06, + "loss": 0.6872, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004445047127106351, + "step": 2340 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5758, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.4445258208290085e-06, + "loss": 0.6667, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00044445258208290083, + "step": 2341 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5756, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.444004300416151e-06, + "loss": 0.7613, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000444400430041615, + "step": 2342 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5755, + "image_masks_loss": 0.0105, + "llm_learning_rate": 4.443482565925209e-06, + "loss": 0.7842, + "output_text_mask_loss": 0.1236, + "predictor_learning_rate": 0.0004443482565925209, + "step": 2343 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5754, + "image_masks_loss": 0.0336, + "llm_learning_rate": 4.442960617413639e-06, + "loss": 0.8193, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004442960617413638, + "step": 2344 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5752, + "image_masks_loss": 0.0443, + "llm_learning_rate": 4.4424384549389165e-06, + "loss": 0.8335, + "output_text_mask_loss": 0.0214, + "predictor_learning_rate": 0.0004442438454938916, + "step": 2345 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5751, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.441916078558546e-06, + "loss": 0.7303, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00044419160785585455, + "step": 2346 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.575, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.441393488330051e-06, + "loss": 0.7803, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.0004441393488330051, + "step": 2347 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5748, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.440870684310981e-06, + "loss": 0.7648, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00044408706843109805, + "step": 2348 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5747, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.440347666558908e-06, + "loss": 0.781, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00044403476665589084, + "step": 2349 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5745, + "image_masks_loss": 0.0111, + "llm_learning_rate": 4.439824435131428e-06, + "loss": 0.7572, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00044398244351314276, + "step": 2350 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5744, + "image_masks_loss": 0.0694, + "llm_learning_rate": 4.439300990086161e-06, + "loss": 0.8244, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000443930099008616, + "step": 2351 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5743, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.438777331480748e-06, + "loss": 0.7352, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00044387773314807477, + "step": 2352 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5741, + "image_masks_loss": 0.05, + "llm_learning_rate": 4.438253459372857e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004438253459372857, + "step": 2353 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.574, + "image_masks_loss": 0.0288, + "llm_learning_rate": 4.437729373820179e-06, + "loss": 0.7938, + "output_text_mask_loss": 0.0786, + "predictor_learning_rate": 0.00044377293738201787, + "step": 2354 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5739, + "image_masks_loss": 0.0366, + "llm_learning_rate": 4.4372050748804256e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.00044372050748804257, + "step": 2355 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5737, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.436680562611335e-06, + "loss": 0.7692, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00044366805626113343, + "step": 2356 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5736, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.436155837070666e-06, + "loss": 0.7707, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004436155837070666, + "step": 2357 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5735, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.435630898316205e-06, + "loss": 0.8032, + "output_text_mask_loss": 0.0458, + "predictor_learning_rate": 0.0004435630898316204, + "step": 2358 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5733, + "image_masks_loss": 0.016, + "llm_learning_rate": 4.435105746405756e-06, + "loss": 0.7057, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.0004435105746405756, + "step": 2359 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5732, + "image_masks_loss": 0.0212, + "llm_learning_rate": 4.434580381397153e-06, + "loss": 0.7464, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.0004434580381397153, + "step": 2360 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5731, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.434054803348249e-06, + "loss": 0.7383, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00044340548033482493, + "step": 2361 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5729, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.433529012316923e-06, + "loss": 0.6545, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.00044335290123169224, + "step": 2362 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5728, + "image_masks_loss": 0.0355, + "llm_learning_rate": 4.4330030083610745e-06, + "loss": 0.794, + "output_text_mask_loss": 0.0664, + "predictor_learning_rate": 0.0004433003008361075, + "step": 2363 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5727, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.43247679153863e-06, + "loss": 0.8025, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00044324767915386294, + "step": 2364 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5725, + "image_masks_loss": 0.022, + "llm_learning_rate": 4.431950361907536e-06, + "loss": 0.8431, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 0.0004431950361907536, + "step": 2365 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5724, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.431423719525765e-06, + "loss": 0.7642, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0004431423719525765, + "step": 2366 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5722, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.4308968644513126e-06, + "loss": 0.7232, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0004430896864451312, + "step": 2367 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5721, + "image_masks_loss": 0.0181, + "llm_learning_rate": 4.430369796742196e-06, + "loss": 0.8372, + "output_text_mask_loss": 0.0439, + "predictor_learning_rate": 0.0004430369796742195, + "step": 2368 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.572, + "image_masks_loss": 0.0173, + "llm_learning_rate": 4.4298425164564576e-06, + "loss": 0.8106, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0004429842516456457, + "step": 2369 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5718, + "image_masks_loss": 0.0228, + "llm_learning_rate": 4.429315023652163e-06, + "loss": 0.7376, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00044293150236521624, + "step": 2370 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5717, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.4287873183874e-06, + "loss": 0.8801, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.00044287873183873996, + "step": 2371 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5716, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.428259400720282e-06, + "loss": 0.8127, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 0.00044282594007202813, + "step": 2372 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5714, + "image_masks_loss": 0.046, + "llm_learning_rate": 4.427731270708943e-06, + "loss": 0.7574, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00044277312707089426, + "step": 2373 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5713, + "image_masks_loss": 0.0186, + "llm_learning_rate": 4.427202928411544e-06, + "loss": 0.7978, + "output_text_mask_loss": 0.0475, + "predictor_learning_rate": 0.00044272029284115434, + "step": 2374 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5712, + "image_masks_loss": 0.0664, + "llm_learning_rate": 4.4266743738862646e-06, + "loss": 0.9127, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004426674373886264, + "step": 2375 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.571, + "image_masks_loss": 0.0656, + "llm_learning_rate": 4.4261456071913115e-06, + "loss": 0.7533, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00044261456071913115, + "step": 2376 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5709, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.425616628384915e-06, + "loss": 0.7992, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004425616628384914, + "step": 2377 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5708, + "image_masks_loss": 0.0874, + "llm_learning_rate": 4.425087437525325e-06, + "loss": 0.7961, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004425087437525325, + "step": 2378 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5706, + "image_masks_loss": 0.0462, + "llm_learning_rate": 4.424558034670819e-06, + "loss": 0.8531, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00044245580346708184, + "step": 2379 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5705, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.4240284198796956e-06, + "loss": 0.7231, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004424028419879695, + "step": 2380 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5704, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.423498593210277e-06, + "loss": 0.8294, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00044234985932102767, + "step": 2381 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5702, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.422968554720908e-06, + "loss": 0.7473, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004422968554720908, + "step": 2382 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5701, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.4224383044699595e-06, + "loss": 0.7903, + "output_text_mask_loss": 0.0732, + "predictor_learning_rate": 0.00044224383044699593, + "step": 2383 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.57, + "image_masks_loss": 0.0319, + "llm_learning_rate": 4.421907842515822e-06, + "loss": 0.783, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0004421907842515822, + "step": 2384 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5698, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.421377168916913e-06, + "loss": 0.7992, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004421377168916913, + "step": 2385 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5697, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.42084628373167e-06, + "loss": 0.7865, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.000442084628373167, + "step": 2386 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5696, + "image_masks_loss": 0.0108, + "llm_learning_rate": 4.420315187018556e-06, + "loss": 0.8442, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00044203151870185557, + "step": 2387 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5694, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.419783878836056e-06, + "loss": 0.8156, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00044197838788360554, + "step": 2388 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5693, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.41925235924268e-06, + "loss": 0.7378, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00044192523592426794, + "step": 2389 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.5691, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.4187206282969575e-06, + "loss": 0.8087, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.0004418720628296957, + "step": 2390 + }, + { + "epoch": 0.24, + "gumbel_tau": 0.569, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.418188686057446e-06, + "loss": 0.8109, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004418188686057446, + "step": 2391 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5689, + "image_masks_loss": 0.0595, + "llm_learning_rate": 4.417656532582725e-06, + "loss": 0.7603, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00044176565325827245, + "step": 2392 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5687, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.417124167931395e-06, + "loss": 0.7327, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004417124167931395, + "step": 2393 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5686, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.416591592162081e-06, + "loss": 0.7156, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00044165915921620804, + "step": 2394 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5685, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.416058805333431e-06, + "loss": 0.7539, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00044160588053334304, + "step": 2395 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5683, + "image_masks_loss": 0.0209, + "llm_learning_rate": 4.415525807504119e-06, + "loss": 0.7621, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00044155258075041185, + "step": 2396 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5682, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.4149925987328365e-06, + "loss": 0.7755, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00044149925987328363, + "step": 2397 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5681, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.414459179078305e-06, + "loss": 0.7544, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.00044144591790783047, + "step": 2398 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5679, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.413925548599264e-06, + "loss": 0.7824, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004413925548599264, + "step": 2399 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5678, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.413391707354479e-06, + "loss": 0.8345, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 0.0004413391707354478, + "step": 2400 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5677, + "image_masks_loss": 0.0053, + "llm_learning_rate": 4.412857655402737e-06, + "loss": 0.6965, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00044128576554027367, + "step": 2401 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5675, + "image_masks_loss": 0.0549, + "llm_learning_rate": 4.4123233928028486e-06, + "loss": 0.771, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004412323392802848, + "step": 2402 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5674, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.411788919613649e-06, + "loss": 0.766, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004411788919613649, + "step": 2403 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5673, + "image_masks_loss": 0.0111, + "llm_learning_rate": 4.4112542358939955e-06, + "loss": 0.8695, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.00044112542358939956, + "step": 2404 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5671, + "image_masks_loss": 0.0207, + "llm_learning_rate": 4.4107193417027686e-06, + "loss": 0.8571, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0004410719341702768, + "step": 2405 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.567, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.410184237098872e-06, + "loss": 0.8914, + "output_text_mask_loss": 0.0832, + "predictor_learning_rate": 0.0004410184237098871, + "step": 2406 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5669, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.409648922141231e-06, + "loss": 0.849, + "output_text_mask_loss": 0.0656, + "predictor_learning_rate": 0.00044096489221412313, + "step": 2407 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5667, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.409113396888798e-06, + "loss": 0.7603, + "output_text_mask_loss": 0.0452, + "predictor_learning_rate": 0.00044091133968887983, + "step": 2408 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5666, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.408577661400546e-06, + "loss": 0.8168, + "output_text_mask_loss": 0.0462, + "predictor_learning_rate": 0.00044085776614005457, + "step": 2409 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5665, + "image_masks_loss": 0.0387, + "llm_learning_rate": 4.40804171573547e-06, + "loss": 0.779, + "output_text_mask_loss": 0.0759, + "predictor_learning_rate": 0.00044080417157354697, + "step": 2410 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5663, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.40750555995259e-06, + "loss": 0.7869, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00044075055599525906, + "step": 2411 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5662, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.40696919411095e-06, + "loss": 0.7429, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00044069691941109497, + "step": 2412 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5661, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.406432618269613e-06, + "loss": 0.7042, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004406432618269614, + "step": 2413 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5659, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.405895832487671e-06, + "loss": 0.777, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004405895832487671, + "step": 2414 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5658, + "image_masks_loss": 0.0349, + "llm_learning_rate": 4.4053588368242354e-06, + "loss": 0.7691, + "output_text_mask_loss": 0.0698, + "predictor_learning_rate": 0.0004405358836824235, + "step": 2415 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5657, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.40482163133844e-06, + "loss": 0.8465, + "output_text_mask_loss": 0.0854, + "predictor_learning_rate": 0.0004404821631338439, + "step": 2416 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5655, + "image_masks_loss": 0.0368, + "llm_learning_rate": 4.404284216089443e-06, + "loss": 0.7895, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004404284216089443, + "step": 2417 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5654, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.403746591136427e-06, + "loss": 0.7183, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00044037465911364273, + "step": 2418 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5653, + "image_masks_loss": 0.0105, + "llm_learning_rate": 4.4032087565385965e-06, + "loss": 0.8042, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0004403208756538596, + "step": 2419 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5651, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.402670712355177e-06, + "loss": 0.8308, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.0004402670712355177, + "step": 2420 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.565, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.402132458645422e-06, + "loss": 0.7667, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00044021324586454217, + "step": 2421 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5649, + "image_masks_loss": 0.0519, + "llm_learning_rate": 4.401593995468603e-06, + "loss": 0.756, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004401593995468603, + "step": 2422 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5647, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.401055322884018e-06, + "loss": 0.8359, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00044010553228840177, + "step": 2423 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5646, + "image_masks_loss": 0.0027, + "llm_learning_rate": 4.400516440950987e-06, + "loss": 0.7789, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.00044005164409509866, + "step": 2424 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5645, + "image_masks_loss": 0.0324, + "llm_learning_rate": 4.399977349728851e-06, + "loss": 0.7879, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004399977349728851, + "step": 2425 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5643, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.399438049276978e-06, + "loss": 0.7925, + "output_text_mask_loss": 0.0207, + "predictor_learning_rate": 0.00043994380492769783, + "step": 2426 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5642, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.398898539654756e-06, + "loss": 0.8135, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0004398898539654756, + "step": 2427 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5641, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.3983588209215975e-06, + "loss": 0.7174, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.00043983588209215975, + "step": 2428 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5639, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.397818893136937e-06, + "loss": 0.7443, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004397818893136937, + "step": 2429 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5638, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.397278756360233e-06, + "loss": 0.7897, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004397278756360232, + "step": 2430 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5637, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.396738410650966e-06, + "loss": 0.8815, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.00043967384106509655, + "step": 2431 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5635, + "image_masks_loss": 0.0366, + "llm_learning_rate": 4.39619785606864e-06, + "loss": 0.8608, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000439619785606864, + "step": 2432 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5634, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.395657092672783e-06, + "loss": 0.7839, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.0004395657092672783, + "step": 2433 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5633, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.395116120522944e-06, + "loss": 0.7615, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004395116120522944, + "step": 2434 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5631, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.394574939678697e-06, + "loss": 0.7521, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.00043945749396786973, + "step": 2435 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.563, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.394033550199639e-06, + "loss": 0.7179, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 0.00043940335501996383, + "step": 2436 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5629, + "image_masks_loss": 0.0193, + "llm_learning_rate": 4.3934919521453856e-06, + "loss": 0.7453, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043934919521453856, + "step": 2437 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5627, + "image_masks_loss": 0.0347, + "llm_learning_rate": 4.392950145575582e-06, + "loss": 0.7432, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00043929501455755815, + "step": 2438 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5626, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.392408130549891e-06, + "loss": 0.8733, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004392408130549891, + "step": 2439 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5625, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.391865907128002e-06, + "loss": 0.6907, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00043918659071280023, + "step": 2440 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5623, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.391323475369626e-06, + "loss": 0.8027, + "output_text_mask_loss": 0.2625, + "predictor_learning_rate": 0.0004391323475369626, + "step": 2441 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5622, + "image_masks_loss": 0.0203, + "llm_learning_rate": 4.390780835334495e-06, + "loss": 0.8123, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00043907808353344944, + "step": 2442 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5621, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.390237987082368e-06, + "loss": 0.7018, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00043902379870823673, + "step": 2443 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5619, + "image_masks_loss": 0.0238, + "llm_learning_rate": 4.389694930673023e-06, + "loss": 0.7049, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00043896949306730226, + "step": 2444 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5618, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.3891516661662625e-06, + "loss": 0.7388, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.0004389151666166262, + "step": 2445 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5617, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.388608193621914e-06, + "loss": 0.7031, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00043886081936219133, + "step": 2446 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5615, + "image_masks_loss": 0.0102, + "llm_learning_rate": 4.3880645130998236e-06, + "loss": 0.8056, + "output_text_mask_loss": 0.0166, + "predictor_learning_rate": 0.0004388064513099823, + "step": 2447 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5614, + "image_masks_loss": 0.0223, + "llm_learning_rate": 4.3875206246598634e-06, + "loss": 0.7424, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.0004387520624659863, + "step": 2448 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5613, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.386976528361928e-06, + "loss": 0.8407, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0004386976528361927, + "step": 2449 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5611, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.386432224265934e-06, + "loss": 0.8358, + "output_text_mask_loss": 0.0954, + "predictor_learning_rate": 0.0004386432224265933, + "step": 2450 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.561, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.385887712431821e-06, + "loss": 0.7291, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.00043858877124318207, + "step": 2451 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5609, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.385342992919554e-06, + "loss": 0.7167, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00043853429929195534, + "step": 2452 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5608, + "image_masks_loss": 0.0076, + "llm_learning_rate": 4.384798065789116e-06, + "loss": 0.7954, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00043847980657891154, + "step": 2453 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5606, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.3842529311005175e-06, + "loss": 0.845, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0004384252931100517, + "step": 2454 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5605, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.3837075889137885e-06, + "loss": 0.7519, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00043837075889137877, + "step": 2455 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5604, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.383162039288984e-06, + "loss": 0.7588, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00043831620392889835, + "step": 2456 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5602, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.382616282286182e-06, + "loss": 0.8284, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043826162822861817, + "step": 2457 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5601, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.382070317965481e-06, + "loss": 0.7964, + "output_text_mask_loss": 0.0893, + "predictor_learning_rate": 0.00043820703179654806, + "step": 2458 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.56, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.381524146387005e-06, + "loss": 0.7588, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004381524146387004, + "step": 2459 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5598, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.3809777676108975e-06, + "loss": 0.8576, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004380977767610897, + "step": 2460 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5597, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.380431181697331e-06, + "loss": 0.7995, + "output_text_mask_loss": 0.0267, + "predictor_learning_rate": 0.000438043118169733, + "step": 2461 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5596, + "image_masks_loss": 0.0328, + "llm_learning_rate": 4.379884388706492e-06, + "loss": 0.76, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00043798843887064924, + "step": 2462 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5594, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.379337388698598e-06, + "loss": 0.7309, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00043793373886985977, + "step": 2463 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5593, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.378790181733885e-06, + "loss": 0.7992, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004378790181733884, + "step": 2464 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5592, + "image_masks_loss": 0.021, + "llm_learning_rate": 4.378242767872611e-06, + "loss": 0.7674, + "output_text_mask_loss": 0.0717, + "predictor_learning_rate": 0.0004378242767872611, + "step": 2465 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.559, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.3776951471750604e-06, + "loss": 0.7539, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00043776951471750605, + "step": 2466 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5589, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.377147319701538e-06, + "loss": 0.7992, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00043771473197015383, + "step": 2467 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5588, + "image_masks_loss": 0.0139, + "llm_learning_rate": 4.3765992855123715e-06, + "loss": 0.7822, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.00043765992855123717, + "step": 2468 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5586, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.376051044667913e-06, + "loss": 0.7155, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00043760510446679126, + "step": 2469 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5585, + "image_masks_loss": 0.013, + "llm_learning_rate": 4.375502597228533e-06, + "loss": 0.7534, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004375502597228533, + "step": 2470 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5584, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.37495394325463e-06, + "loss": 0.7943, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00043749539432546295, + "step": 2471 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5582, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.374405082806622e-06, + "loss": 0.7843, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004374405082806622, + "step": 2472 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5581, + "image_masks_loss": 0.0088, + "llm_learning_rate": 4.3738560159449525e-06, + "loss": 0.6848, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004373856015944952, + "step": 2473 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.558, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.373306742730083e-06, + "loss": 0.7079, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00043733067427300835, + "step": 2474 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5578, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.372757263222503e-06, + "loss": 0.7936, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004372757263222503, + "step": 2475 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5577, + "image_masks_loss": 0.0079, + "llm_learning_rate": 4.3722075774827225e-06, + "loss": 0.7492, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004372207577482722, + "step": 2476 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5576, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.3716576855712735e-06, + "loss": 0.7082, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00043716576855712733, + "step": 2477 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5575, + "image_masks_loss": 0.0067, + "llm_learning_rate": 4.37110758754871e-06, + "loss": 0.7752, + "output_text_mask_loss": 0.0225, + "predictor_learning_rate": 0.00043711075875487094, + "step": 2478 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5573, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.370557283475612e-06, + "loss": 0.711, + "output_text_mask_loss": 0.0311, + "predictor_learning_rate": 0.00043705572834756113, + "step": 2479 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5572, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.370006773412579e-06, + "loss": 0.8548, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004370006773412579, + "step": 2480 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5571, + "image_masks_loss": 0.0243, + "llm_learning_rate": 4.369456057420235e-06, + "loss": 0.7408, + "output_text_mask_loss": 0.0183, + "predictor_learning_rate": 0.0004369456057420235, + "step": 2481 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5569, + "image_masks_loss": 0.0481, + "llm_learning_rate": 4.368905135559226e-06, + "loss": 0.7424, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00043689051355592256, + "step": 2482 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5568, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.368354007890221e-06, + "loss": 0.8385, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.0004368354007890221, + "step": 2483 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5567, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.367802674473911e-06, + "loss": 0.7359, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000436780267447391, + "step": 2484 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5565, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.367251135371009e-06, + "loss": 0.7742, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004367251135371009, + "step": 2485 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5564, + "image_masks_loss": 0.0049, + "llm_learning_rate": 4.366699390642254e-06, + "loss": 0.8322, + "output_text_mask_loss": 0.0317, + "predictor_learning_rate": 0.00043666993906422535, + "step": 2486 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5563, + "image_masks_loss": 0.0044, + "llm_learning_rate": 4.366147440348404e-06, + "loss": 0.7886, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.00043661474403484035, + "step": 2487 + }, + { + "epoch": 0.25, + "gumbel_tau": 0.5561, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.365595284550241e-06, + "loss": 0.7831, + "output_text_mask_loss": 0.0408, + "predictor_learning_rate": 0.00043655952845502404, + "step": 2488 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.556, + "image_masks_loss": 0.0546, + "llm_learning_rate": 4.3650429233085695e-06, + "loss": 0.7646, + "output_text_mask_loss": 0.0652, + "predictor_learning_rate": 0.00043650429233085694, + "step": 2489 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5559, + "image_masks_loss": 0.0204, + "llm_learning_rate": 4.364490356684218e-06, + "loss": 0.8071, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.00043644903566842177, + "step": 2490 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5557, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.363937584738034e-06, + "loss": 0.7743, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00043639375847380344, + "step": 2491 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5556, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.3633846075308925e-06, + "loss": 0.7688, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00043633846075308926, + "step": 2492 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5555, + "image_masks_loss": 0.0433, + "llm_learning_rate": 4.362831425123688e-06, + "loss": 0.8633, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043628314251236875, + "step": 2493 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5554, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.362278037577336e-06, + "loss": 0.7754, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.0004362278037577336, + "step": 2494 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5552, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.361724444952779e-06, + "loss": 0.8497, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00043617244449527796, + "step": 2495 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5551, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.36117064731098e-06, + "loss": 0.7893, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.000436117064731098, + "step": 2496 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.555, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.360616644712924e-06, + "loss": 0.7678, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00043606166447129234, + "step": 2497 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5548, + "image_masks_loss": 0.0687, + "llm_learning_rate": 4.360062437219618e-06, + "loss": 0.8307, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00043600624372196173, + "step": 2498 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5547, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.359508024892093e-06, + "loss": 0.8557, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 0.00043595080248920925, + "step": 2499 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5546, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.358953407791403e-06, + "loss": 0.7481, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.00043589534077914026, + "step": 2500 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5544, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.358398585978623e-06, + "loss": 0.8019, + "output_text_mask_loss": 0.0507, + "predictor_learning_rate": 0.0004358398585978622, + "step": 2501 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5543, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.3578435595148505e-06, + "loss": 0.7377, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000435784355951485, + "step": 2502 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5542, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.3572883284612075e-06, + "loss": 0.7419, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00043572883284612076, + "step": 2503 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.554, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.356732892878838e-06, + "loss": 0.7071, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00043567328928788377, + "step": 2504 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5539, + "image_masks_loss": 0.0059, + "llm_learning_rate": 4.3561772528289056e-06, + "loss": 0.7939, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0004356177252828905, + "step": 2505 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5538, + "image_masks_loss": 0.0368, + "llm_learning_rate": 4.3556214083726e-06, + "loss": 0.7759, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00043556214083725995, + "step": 2506 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5537, + "image_masks_loss": 0.0088, + "llm_learning_rate": 4.355065359571131e-06, + "loss": 0.74, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00043550653595711306, + "step": 2507 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5535, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.354509106485733e-06, + "loss": 0.8083, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00043545091064857325, + "step": 2508 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5534, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.353952649177661e-06, + "loss": 0.8672, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00043539526491776604, + "step": 2509 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5533, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.3533959877081946e-06, + "loss": 0.7662, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0004353395987708194, + "step": 2510 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5531, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.3528391221386325e-06, + "loss": 0.7143, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043528391221386327, + "step": 2511 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.553, + "image_masks_loss": 0.0141, + "llm_learning_rate": 4.352282052530301e-06, + "loss": 0.7173, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 0.00043522820525303003, + "step": 2512 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5529, + "image_masks_loss": 0.0038, + "llm_learning_rate": 4.351724778944543e-06, + "loss": 0.7265, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00043517247789445426, + "step": 2513 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5527, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.351167301442728e-06, + "loss": 0.7812, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 0.0004351167301442728, + "step": 2514 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5526, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.350609620086246e-06, + "loss": 0.7496, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.00043506096200862453, + "step": 2515 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5525, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.3500517349365115e-06, + "loss": 0.8082, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.0004350051734936511, + "step": 2516 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5523, + "image_masks_loss": 0.0475, + "llm_learning_rate": 4.349493646054958e-06, + "loss": 0.7763, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004349493646054957, + "step": 2517 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5522, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.348935353503045e-06, + "loss": 0.7442, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00043489353535030445, + "step": 2518 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5521, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.348376857342252e-06, + "loss": 0.7091, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0004348376857342252, + "step": 2519 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.552, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.347818157634084e-06, + "loss": 0.7811, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00043478181576340835, + "step": 2520 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5518, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.347259254440064e-06, + "loss": 0.7713, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004347259254440064, + "step": 2521 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5517, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.34670014782174e-06, + "loss": 0.6992, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004346700147821739, + "step": 2522 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5516, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.3461408378406815e-06, + "loss": 0.7213, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0004346140837840681, + "step": 2523 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5514, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.3455813245584835e-06, + "loss": 0.7317, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004345581324558483, + "step": 2524 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5513, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.345021608036758e-06, + "loss": 0.6921, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0004345021608036758, + "step": 2525 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5512, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.344461688337144e-06, + "loss": 0.8175, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0004344461688337143, + "step": 2526 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.551, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.343901565521299e-06, + "loss": 0.8022, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004343901565521299, + "step": 2527 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5509, + "image_masks_loss": 0.0139, + "llm_learning_rate": 4.343341239650908e-06, + "loss": 0.7442, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00043433412396509083, + "step": 2528 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5508, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.3427807107876744e-06, + "loss": 0.7343, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004342780710787674, + "step": 2529 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5507, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.342219978993323e-06, + "loss": 0.9059, + "output_text_mask_loss": 0.1694, + "predictor_learning_rate": 0.0004342219978993323, + "step": 2530 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5505, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.341659044329605e-06, + "loss": 1.041, + "output_text_mask_loss": 0.1434, + "predictor_learning_rate": 0.00043416590443296046, + "step": 2531 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5504, + "image_masks_loss": 0.0221, + "llm_learning_rate": 4.34109790685829e-06, + "loss": 0.8887, + "output_text_mask_loss": 0.235, + "predictor_learning_rate": 0.00043410979068582895, + "step": 2532 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5503, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.340536566641173e-06, + "loss": 0.8829, + "output_text_mask_loss": 0.1755, + "predictor_learning_rate": 0.0004340536566641173, + "step": 2533 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5501, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.339975023740069e-06, + "loss": 1.0031, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004339975023740069, + "step": 2534 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.55, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.339413278216819e-06, + "loss": 0.8549, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004339413278216818, + "step": 2535 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5499, + "image_masks_loss": 0.0546, + "llm_learning_rate": 4.338851330133279e-06, + "loss": 0.8026, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043388513301332796, + "step": 2536 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5497, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.338289179551337e-06, + "loss": 0.7828, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004338289179551337, + "step": 2537 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5496, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.337726826532894e-06, + "loss": 0.7723, + "output_text_mask_loss": 0.0206, + "predictor_learning_rate": 0.00043377268265328944, + "step": 2538 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5495, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.337164271139881e-06, + "loss": 0.8522, + "output_text_mask_loss": 0.0702, + "predictor_learning_rate": 0.00043371642711398806, + "step": 2539 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5494, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.336601513434246e-06, + "loss": 0.8921, + "output_text_mask_loss": 0.0977, + "predictor_learning_rate": 0.00043366015134342455, + "step": 2540 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5492, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.336038553477961e-06, + "loss": 0.8386, + "output_text_mask_loss": 0.2365, + "predictor_learning_rate": 0.00043360385534779613, + "step": 2541 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5491, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.335475391333022e-06, + "loss": 0.863, + "output_text_mask_loss": 0.2823, + "predictor_learning_rate": 0.00043354753913330213, + "step": 2542 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.549, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.334912027061443e-06, + "loss": 0.8313, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004334912027061443, + "step": 2543 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5488, + "image_masks_loss": 0.0307, + "llm_learning_rate": 4.334348460725265e-06, + "loss": 0.9855, + "output_text_mask_loss": 0.1717, + "predictor_learning_rate": 0.0004334348460725265, + "step": 2544 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5487, + "image_masks_loss": 0.0198, + "llm_learning_rate": 4.333784692386549e-06, + "loss": 0.8405, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004333784692386549, + "step": 2545 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5486, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.333220722107377e-06, + "loss": 0.9644, + "output_text_mask_loss": 0.2258, + "predictor_learning_rate": 0.0004333220722107377, + "step": 2546 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5485, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.3326565499498575e-06, + "loss": 1.0044, + "output_text_mask_loss": 0.1694, + "predictor_learning_rate": 0.0004332656549949857, + "step": 2547 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5483, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.332092175976116e-06, + "loss": 0.8898, + "output_text_mask_loss": 0.267, + "predictor_learning_rate": 0.0004332092175976116, + "step": 2548 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5482, + "image_masks_loss": 0.0364, + "llm_learning_rate": 4.331527600248303e-06, + "loss": 0.8072, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00043315276002483025, + "step": 2549 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5481, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.33096282282859e-06, + "loss": 0.8779, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00043309628228285904, + "step": 2550 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5479, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.330397843779175e-06, + "loss": 0.8663, + "output_text_mask_loss": 0.2197, + "predictor_learning_rate": 0.0004330397843779174, + "step": 2551 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5478, + "image_masks_loss": 0.0079, + "llm_learning_rate": 4.32983266316227e-06, + "loss": 0.7546, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00043298326631622695, + "step": 2552 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5477, + "image_masks_loss": 0.0158, + "llm_learning_rate": 4.3292672810401184e-06, + "loss": 0.7684, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004329267281040118, + "step": 2553 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5475, + "image_masks_loss": 0.008, + "llm_learning_rate": 4.328701697474977e-06, + "loss": 0.7792, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0004328701697474977, + "step": 2554 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5474, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.328135912529133e-06, + "loss": 0.5763, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004328135912529133, + "step": 2555 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5473, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.32756992626489e-06, + "loss": 0.8104, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00043275699262648905, + "step": 2556 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5472, + "image_masks_loss": 0.0181, + "llm_learning_rate": 4.327003738744576e-06, + "loss": 0.5526, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004327003738744576, + "step": 2557 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.547, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.326437350030541e-06, + "loss": 0.7497, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004326437350030541, + "step": 2558 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5469, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.3258707601851565e-06, + "loss": 0.8357, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00043258707601851566, + "step": 2559 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5468, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.325303969270818e-06, + "loss": 0.7579, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00043253039692708173, + "step": 2560 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5466, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.324736977349939e-06, + "loss": 0.8183, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004324736977349939, + "step": 2561 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5465, + "image_masks_loss": 0.0219, + "llm_learning_rate": 4.324169784484961e-06, + "loss": 0.8268, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 0.00043241697844849603, + "step": 2562 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5464, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.323602390738342e-06, + "loss": 0.761, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00043236023907383415, + "step": 2563 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5463, + "image_masks_loss": 0.0072, + "llm_learning_rate": 4.323034796172566e-06, + "loss": 0.7149, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0004323034796172566, + "step": 2564 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5461, + "image_masks_loss": 0.0059, + "llm_learning_rate": 4.322467000850138e-06, + "loss": 0.6931, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004322467000850138, + "step": 2565 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.546, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.321899004833585e-06, + "loss": 0.6903, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043218990048335843, + "step": 2566 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5459, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.321330808185454e-06, + "loss": 0.7312, + "output_text_mask_loss": 0.0395, + "predictor_learning_rate": 0.0004321330808185454, + "step": 2567 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5457, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.320762410968319e-06, + "loss": 0.7596, + "output_text_mask_loss": 0.0458, + "predictor_learning_rate": 0.00043207624109683183, + "step": 2568 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5456, + "image_masks_loss": 0.0292, + "llm_learning_rate": 4.3201938132447706e-06, + "loss": 0.7978, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00043201938132447705, + "step": 2569 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5455, + "image_masks_loss": 0.0232, + "llm_learning_rate": 4.319625015077426e-06, + "loss": 0.7156, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0004319625015077426, + "step": 2570 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5454, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.319056016528922e-06, + "loss": 0.7882, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004319056016528921, + "step": 2571 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5452, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.318486817661917e-06, + "loss": 0.7626, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043184868176619165, + "step": 2572 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5451, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.3179174185390925e-06, + "loss": 0.7362, + "output_text_mask_loss": 0.0523, + "predictor_learning_rate": 0.00043179174185390925, + "step": 2573 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.545, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.3173478192231535e-06, + "loss": 0.8393, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004317347819223153, + "step": 2574 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5448, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.316778019776824e-06, + "loss": 0.892, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 0.00043167780197768234, + "step": 2575 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5447, + "image_masks_loss": 0.034, + "llm_learning_rate": 4.316208020262853e-06, + "loss": 0.7887, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.0004316208020262853, + "step": 2576 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5446, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.3156378207440085e-06, + "loss": 0.7202, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.00043156378207440083, + "step": 2577 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5445, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.315067421283083e-06, + "loss": 0.8912, + "output_text_mask_loss": 0.1221, + "predictor_learning_rate": 0.0004315067421283083, + "step": 2578 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5443, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.314496821942892e-06, + "loss": 0.906, + "output_text_mask_loss": 0.1389, + "predictor_learning_rate": 0.00043144968219428916, + "step": 2579 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5442, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.313926022786268e-06, + "loss": 0.8262, + "output_text_mask_loss": 0.1328, + "predictor_learning_rate": 0.00043139260227862677, + "step": 2580 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5441, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.313355023876069e-06, + "loss": 0.8549, + "output_text_mask_loss": 0.1167, + "predictor_learning_rate": 0.0004313355023876069, + "step": 2581 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5439, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.312783825275178e-06, + "loss": 0.8514, + "output_text_mask_loss": 0.1686, + "predictor_learning_rate": 0.00043127838252751775, + "step": 2582 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5438, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.312212427046492e-06, + "loss": 0.7717, + "output_text_mask_loss": 0.1984, + "predictor_learning_rate": 0.0004312212427046492, + "step": 2583 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5437, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.3116408292529386e-06, + "loss": 0.8958, + "output_text_mask_loss": 0.1717, + "predictor_learning_rate": 0.00043116408292529385, + "step": 2584 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5436, + "image_masks_loss": 0.0395, + "llm_learning_rate": 4.311069031957461e-06, + "loss": 0.9672, + "output_text_mask_loss": 0.1411, + "predictor_learning_rate": 0.00043110690319574606, + "step": 2585 + }, + { + "epoch": 0.26, + "gumbel_tau": 0.5434, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.310497035223028e-06, + "loss": 0.7703, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004310497035223028, + "step": 2586 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5433, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.309924839112628e-06, + "loss": 0.9151, + "output_text_mask_loss": 0.1793, + "predictor_learning_rate": 0.0004309924839112628, + "step": 2587 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5432, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.309352443689274e-06, + "loss": 0.7584, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.0004309352443689274, + "step": 2588 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.543, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.308779849015999e-06, + "loss": 0.8361, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.0004308779849015999, + "step": 2589 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5429, + "image_masks_loss": 0.0341, + "llm_learning_rate": 4.308207055155858e-06, + "loss": 0.9085, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00043082070551558573, + "step": 2590 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5428, + "image_masks_loss": 0.0215, + "llm_learning_rate": 4.307634062171928e-06, + "loss": 0.7426, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.00043076340621719277, + "step": 2591 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5427, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.307060870127309e-06, + "loss": 0.7875, + "output_text_mask_loss": 0.1312, + "predictor_learning_rate": 0.00043070608701273083, + "step": 2592 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5425, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.306487479085122e-06, + "loss": 0.7522, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00043064874790851214, + "step": 2593 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5424, + "image_masks_loss": 0.0422, + "llm_learning_rate": 4.30591388910851e-06, + "loss": 0.8338, + "output_text_mask_loss": 0.0893, + "predictor_learning_rate": 0.0004305913889108509, + "step": 2594 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5423, + "image_masks_loss": 0.0111, + "llm_learning_rate": 4.305340100260637e-06, + "loss": 0.8101, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00043053401002606374, + "step": 2595 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5421, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.304766112604692e-06, + "loss": 0.6273, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.00043047661126046916, + "step": 2596 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.542, + "image_masks_loss": 0.0155, + "llm_learning_rate": 4.3041919262038825e-06, + "loss": 0.7635, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004304191926203882, + "step": 2597 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5419, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.30361754112144e-06, + "loss": 0.7164, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043036175411214397, + "step": 2598 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5418, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.303042957420615e-06, + "loss": 0.7008, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043030429574206154, + "step": 2599 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5416, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.3024681751646845e-06, + "loss": 0.7961, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00043024681751646843, + "step": 2600 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5415, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.301893194416944e-06, + "loss": 0.6916, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004301893194416944, + "step": 2601 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5414, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.301318015240711e-06, + "loss": 0.8292, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00043013180152407107, + "step": 2602 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5413, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.300742637699325e-06, + "loss": 0.7152, + "output_text_mask_loss": 0.09, + "predictor_learning_rate": 0.0004300742637699325, + "step": 2603 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5411, + "image_masks_loss": 0.017, + "llm_learning_rate": 4.300167061856149e-06, + "loss": 0.8069, + "output_text_mask_loss": 0.0492, + "predictor_learning_rate": 0.00043001670618561495, + "step": 2604 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.541, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.299591287774567e-06, + "loss": 0.7541, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00042995912877745674, + "step": 2605 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5409, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.299015315517985e-06, + "loss": 0.8387, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.00042990153155179846, + "step": 2606 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5407, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.298439145149828e-06, + "loss": 0.8723, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.00042984391451498274, + "step": 2607 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5406, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.297862776733547e-06, + "loss": 0.8922, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.00042978627767335467, + "step": 2608 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5405, + "image_masks_loss": 0.0205, + "llm_learning_rate": 4.297286210332612e-06, + "loss": 0.8274, + "output_text_mask_loss": 0.0364, + "predictor_learning_rate": 0.00042972862103326116, + "step": 2609 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5404, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.296709446010516e-06, + "loss": 0.717, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0004296709446010516, + "step": 2610 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5402, + "image_masks_loss": 0.0157, + "llm_learning_rate": 4.296132483830774e-06, + "loss": 0.8037, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00042961324838307737, + "step": 2611 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5401, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.295555323856923e-06, + "loss": 0.6499, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004295555323856922, + "step": 2612 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.54, + "image_masks_loss": 0.0429, + "llm_learning_rate": 4.29497796615252e-06, + "loss": 0.7195, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004294977966152519, + "step": 2613 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5399, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.2944004107811434e-06, + "loss": 0.8137, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004294400410781143, + "step": 2614 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5397, + "image_masks_loss": 0.0139, + "llm_learning_rate": 4.293822657806398e-06, + "loss": 0.8361, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042938226578063976, + "step": 2615 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5396, + "image_masks_loss": 0.008, + "llm_learning_rate": 4.2932447072919065e-06, + "loss": 0.6337, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0004293244707291906, + "step": 2616 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5395, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.292666559301312e-06, + "loss": 0.7763, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00042926665593013126, + "step": 2617 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5393, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.292088213898285e-06, + "loss": 0.8563, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.0004292088213898285, + "step": 2618 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5392, + "image_masks_loss": 0.0217, + "llm_learning_rate": 4.291509671146511e-06, + "loss": 0.8729, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.00042915096711465105, + "step": 2619 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5391, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.290930931109702e-06, + "loss": 0.8085, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0004290930931109702, + "step": 2620 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.539, + "image_masks_loss": 0.0084, + "llm_learning_rate": 4.290351993851589e-06, + "loss": 0.7627, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0004290351993851589, + "step": 2621 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5388, + "image_masks_loss": 0.0315, + "llm_learning_rate": 4.289772859435927e-06, + "loss": 0.8133, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.0004289772859435927, + "step": 2622 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5387, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.289193527926492e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0782, + "predictor_learning_rate": 0.00042891935279264914, + "step": 2623 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5386, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.288613999387079e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042886139993870786, + "step": 2624 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5385, + "image_masks_loss": 0.0307, + "llm_learning_rate": 4.288034273881509e-06, + "loss": 0.8065, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004288034273881509, + "step": 2625 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5383, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.287454351473622e-06, + "loss": 0.7706, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00042874543514736217, + "step": 2626 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5382, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.286874232227281e-06, + "loss": 0.785, + "output_text_mask_loss": 0.1488, + "predictor_learning_rate": 0.0004286874232227281, + "step": 2627 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5381, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.286293916206369e-06, + "loss": 0.6772, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00042862939162063683, + "step": 2628 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5379, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.285713403474792e-06, + "loss": 0.6233, + "output_text_mask_loss": 0.0068, + "predictor_learning_rate": 0.0004285713403474792, + "step": 2629 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5378, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.285132694096478e-06, + "loss": 0.7376, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00042851326940964774, + "step": 2630 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5377, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.284551788135375e-06, + "loss": 0.8157, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00042845517881353744, + "step": 2631 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5376, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.283970685655454e-06, + "loss": 0.7392, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004283970685655454, + "step": 2632 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5374, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.283389386720708e-06, + "loss": 0.7285, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042833893867207074, + "step": 2633 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5373, + "image_masks_loss": 0.0053, + "llm_learning_rate": 4.28280789139515e-06, + "loss": 0.793, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 0.000428280789139515, + "step": 2634 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5372, + "image_masks_loss": 0.0423, + "llm_learning_rate": 4.282226199742816e-06, + "loss": 0.781, + "output_text_mask_loss": 0.0694, + "predictor_learning_rate": 0.00042822261997428157, + "step": 2635 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5371, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.281644311827763e-06, + "loss": 0.7957, + "output_text_mask_loss": 0.1015, + "predictor_learning_rate": 0.0004281644311827763, + "step": 2636 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5369, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.281062227714071e-06, + "loss": 0.8104, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.00042810622277140704, + "step": 2637 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5368, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.280479947465838e-06, + "loss": 0.6599, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00042804799474658375, + "step": 2638 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5367, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.279897471147188e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00042798974711471876, + "step": 2639 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5366, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.2793147988222635e-06, + "loss": 0.7916, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00042793147988222634, + "step": 2640 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5364, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.27873193055523e-06, + "loss": 0.7053, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00042787319305552296, + "step": 2641 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5363, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.2781488664102746e-06, + "loss": 0.8525, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042781488664102746, + "step": 2642 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5362, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.277565606451606e-06, + "loss": 0.6717, + "output_text_mask_loss": 0.0317, + "predictor_learning_rate": 0.00042775656064516056, + "step": 2643 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.536, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.276982150743452e-06, + "loss": 0.7749, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004276982150743453, + "step": 2644 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5359, + "image_masks_loss": 0.0109, + "llm_learning_rate": 4.276398499350067e-06, + "loss": 0.723, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042763984993500666, + "step": 2645 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5358, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.2758146523357215e-06, + "loss": 0.6887, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004275814652335721, + "step": 2646 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5357, + "image_masks_loss": 0.0068, + "llm_learning_rate": 4.275230609764711e-06, + "loss": 0.7745, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 0.00042752306097647107, + "step": 2647 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5355, + "image_masks_loss": 0.0152, + "llm_learning_rate": 4.274646371701352e-06, + "loss": 0.7534, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004274646371701352, + "step": 2648 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5354, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.274061938209982e-06, + "loss": 0.7145, + "output_text_mask_loss": 0.0214, + "predictor_learning_rate": 0.00042740619382099817, + "step": 2649 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5353, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.273477309354959e-06, + "loss": 0.7621, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004273477309354959, + "step": 2650 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5352, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.272892485200665e-06, + "loss": 0.7968, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.00042728924852006647, + "step": 2651 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.535, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.272307465811502e-06, + "loss": 0.8034, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.0004272307465811501, + "step": 2652 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5349, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.271722251251892e-06, + "loss": 0.8513, + "output_text_mask_loss": 0.1389, + "predictor_learning_rate": 0.0004271722251251892, + "step": 2653 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5348, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.271136841586282e-06, + "loss": 0.7666, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00042711368415862815, + "step": 2654 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5347, + "image_masks_loss": 0.0311, + "llm_learning_rate": 4.270551236879137e-06, + "loss": 0.8579, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004270551236879137, + "step": 2655 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5345, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.269965437194948e-06, + "loss": 0.7721, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.0004269965437194947, + "step": 2656 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5344, + "image_masks_loss": 0.0458, + "llm_learning_rate": 4.26937944259822e-06, + "loss": 0.7768, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.00042693794425982203, + "step": 2657 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5343, + "image_masks_loss": 0.005, + "llm_learning_rate": 4.26879325315349e-06, + "loss": 0.6837, + "output_text_mask_loss": 0.0622, + "predictor_learning_rate": 0.0004268793253153489, + "step": 2658 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5342, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.268206868925304e-06, + "loss": 0.7987, + "output_text_mask_loss": 0.0782, + "predictor_learning_rate": 0.0004268206868925304, + "step": 2659 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.534, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.267620289978241e-06, + "loss": 0.8025, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000426762028997824, + "step": 2660 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5339, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.267033516376893e-06, + "loss": 0.7047, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0004267033516376893, + "step": 2661 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5338, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.26644654818588e-06, + "loss": 0.8158, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.0004266446548185879, + "step": 2662 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5336, + "image_masks_loss": 0.0045, + "llm_learning_rate": 4.265859385469837e-06, + "loss": 0.7492, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00042658593854698367, + "step": 2663 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5335, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.265272028293426e-06, + "loss": 0.8354, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.0004265272028293425, + "step": 2664 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5334, + "image_masks_loss": 0.0406, + "llm_learning_rate": 4.2646844767213265e-06, + "loss": 0.8599, + "output_text_mask_loss": 0.0664, + "predictor_learning_rate": 0.00042646844767213265, + "step": 2665 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5333, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.2640967308182425e-06, + "loss": 0.8356, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004264096730818242, + "step": 2666 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5331, + "image_masks_loss": 0.0219, + "llm_learning_rate": 4.263508790648897e-06, + "loss": 0.776, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00042635087906488966, + "step": 2667 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.533, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.262920656278034e-06, + "loss": 0.7938, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.0004262920656278034, + "step": 2668 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5329, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.262332327770424e-06, + "loss": 0.7398, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00042623323277704236, + "step": 2669 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5328, + "image_masks_loss": 0.0155, + "llm_learning_rate": 4.2617438051908515e-06, + "loss": 0.8417, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00042617438051908513, + "step": 2670 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5326, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.261155088604127e-06, + "loss": 0.6814, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00042611550886041265, + "step": 2671 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5325, + "image_masks_loss": 0.0441, + "llm_learning_rate": 4.260566178075081e-06, + "loss": 0.777, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004260566178075081, + "step": 2672 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5324, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.2599770736685665e-06, + "loss": 0.7834, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004259977073668566, + "step": 2673 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5323, + "image_masks_loss": 0.0213, + "llm_learning_rate": 4.2593877754494565e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0801, + "predictor_learning_rate": 0.0004259387775449456, + "step": 2674 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5321, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.258798283482645e-06, + "loss": 0.8004, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004258798283482645, + "step": 2675 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.532, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.25820859783305e-06, + "loss": 0.8199, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000425820859783305, + "step": 2676 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5319, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.257618718565607e-06, + "loss": 0.7455, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00042576187185656067, + "step": 2677 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5318, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.257028645745276e-06, + "loss": 0.5997, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0004257028645745276, + "step": 2678 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5316, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.256438379437038e-06, + "loss": 0.8281, + "output_text_mask_loss": 0.1068, + "predictor_learning_rate": 0.00042564383794370375, + "step": 2679 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5315, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.2558479197058925e-06, + "loss": 0.7981, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004255847919705892, + "step": 2680 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5314, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.255257266616862e-06, + "loss": 0.7656, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.00042552572666168623, + "step": 2681 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5313, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.254666420234993e-06, + "loss": 0.8086, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.00042546664202349925, + "step": 2682 + }, + { + "epoch": 0.27, + "gumbel_tau": 0.5311, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.25407538062535e-06, + "loss": 0.6811, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004254075380625349, + "step": 2683 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.531, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.2534841478530175e-06, + "loss": 0.8132, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00042534841478530173, + "step": 2684 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5309, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.252892721983105e-06, + "loss": 0.8182, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00042528927219831047, + "step": 2685 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5308, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.252301103080743e-06, + "loss": 0.8233, + "output_text_mask_loss": 0.0173, + "predictor_learning_rate": 0.0004252301103080742, + "step": 2686 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5306, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.251709291211079e-06, + "loss": 0.8114, + "output_text_mask_loss": 0.1297, + "predictor_learning_rate": 0.0004251709291211079, + "step": 2687 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5305, + "image_masks_loss": 0.0034, + "llm_learning_rate": 4.251117286439287e-06, + "loss": 0.8305, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00042511172864392866, + "step": 2688 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5304, + "image_masks_loss": 0.0305, + "llm_learning_rate": 4.250525088830558e-06, + "loss": 0.7819, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004250525088830558, + "step": 2689 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5303, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.249932698450108e-06, + "loss": 0.7582, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042499326984501073, + "step": 2690 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5301, + "image_masks_loss": 0.0211, + "llm_learning_rate": 4.2493401153631705e-06, + "loss": 0.7398, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.00042493401153631706, + "step": 2691 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.53, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.248747339635004e-06, + "loss": 0.7665, + "output_text_mask_loss": 0.1167, + "predictor_learning_rate": 0.0004248747339635003, + "step": 2692 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5299, + "image_masks_loss": 0.0355, + "llm_learning_rate": 4.248154371330885e-06, + "loss": 0.7613, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00042481543713308845, + "step": 2693 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5298, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.247561210516113e-06, + "loss": 0.779, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0004247561210516112, + "step": 2694 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5296, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.246967857256007e-06, + "loss": 0.8553, + "output_text_mask_loss": 0.0904, + "predictor_learning_rate": 0.00042469678572560065, + "step": 2695 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5295, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.246374311615909e-06, + "loss": 0.751, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004246374311615909, + "step": 2696 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5294, + "image_masks_loss": 0.041, + "llm_learning_rate": 4.2457805736611836e-06, + "loss": 0.8122, + "output_text_mask_loss": 0.0675, + "predictor_learning_rate": 0.0004245780573661183, + "step": 2697 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5293, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.245186643457212e-06, + "loss": 0.7919, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042451866434572115, + "step": 2698 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5291, + "image_masks_loss": 0.0492, + "llm_learning_rate": 4.2445925210694e-06, + "loss": 0.6999, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00042445925210693995, + "step": 2699 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.529, + "image_masks_loss": 0.0189, + "llm_learning_rate": 4.2439982065631736e-06, + "loss": 0.7215, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004243998206563173, + "step": 2700 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5289, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.24340370000398e-06, + "loss": 0.8383, + "output_text_mask_loss": 0.0973, + "predictor_learning_rate": 0.000424340370000398, + "step": 2701 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5288, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.242809001457288e-06, + "loss": 0.7833, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00042428090014572876, + "step": 2702 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5286, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.242214110988586e-06, + "loss": 0.8243, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004242214110988586, + "step": 2703 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5285, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.241619028663386e-06, + "loss": 0.6905, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004241619028663386, + "step": 2704 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5284, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.24102375454722e-06, + "loss": 0.7554, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000424102375454722, + "step": 2705 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5283, + "image_masks_loss": 0.0067, + "llm_learning_rate": 4.240428288705639e-06, + "loss": 0.7758, + "output_text_mask_loss": 0.0851, + "predictor_learning_rate": 0.0004240428288705639, + "step": 2706 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5281, + "image_masks_loss": 0.037, + "llm_learning_rate": 4.23983263120422e-06, + "loss": 0.7661, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0004239832631204219, + "step": 2707 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.528, + "image_masks_loss": 0.0265, + "llm_learning_rate": 4.239236782108555e-06, + "loss": 0.7824, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0004239236782108554, + "step": 2708 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5279, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.238640741484261e-06, + "loss": 0.8305, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004238640741484261, + "step": 2709 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5278, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.238044509396977e-06, + "loss": 0.8352, + "output_text_mask_loss": 0.0645, + "predictor_learning_rate": 0.0004238044509396976, + "step": 2710 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5276, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.237448085912359e-06, + "loss": 0.7011, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004237448085912359, + "step": 2711 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5275, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.236851471096088e-06, + "loss": 0.7324, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004236851471096088, + "step": 2712 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5274, + "image_masks_loss": 0.0233, + "llm_learning_rate": 4.236254665013866e-06, + "loss": 0.7911, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004236254665013865, + "step": 2713 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5273, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.235657667731411e-06, + "loss": 0.7545, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.00042356576677314106, + "step": 2714 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5271, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.235060479314468e-06, + "loss": 0.7545, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004235060479314468, + "step": 2715 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.527, + "image_masks_loss": 0.0441, + "llm_learning_rate": 4.234463099828801e-06, + "loss": 0.8572, + "output_text_mask_loss": 0.1152, + "predictor_learning_rate": 0.00042344630998288014, + "step": 2716 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5269, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.233865529340194e-06, + "loss": 0.6845, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004233865529340194, + "step": 2717 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5268, + "image_masks_loss": 0.008, + "llm_learning_rate": 4.233267767914453e-06, + "loss": 0.6664, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042332677679144526, + "step": 2718 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5266, + "image_masks_loss": 0.0336, + "llm_learning_rate": 4.232669815617404e-06, + "loss": 0.8201, + "output_text_mask_loss": 0.0477, + "predictor_learning_rate": 0.00042326698156174036, + "step": 2719 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5265, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.2320716725148955e-06, + "loss": 0.7851, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.00042320716725148956, + "step": 2720 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5264, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.231473338672797e-06, + "loss": 0.7768, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004231473338672797, + "step": 2721 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5263, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.2308748141569975e-06, + "loss": 0.745, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00042308748141569973, + "step": 2722 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5261, + "image_masks_loss": 0.0077, + "llm_learning_rate": 4.2302760990334084e-06, + "loss": 0.7062, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004230276099033408, + "step": 2723 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.526, + "image_masks_loss": 0.0186, + "llm_learning_rate": 4.2296771933679605e-06, + "loss": 0.8187, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004229677193367961, + "step": 2724 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5259, + "image_masks_loss": 0.0389, + "llm_learning_rate": 4.229078097226609e-06, + "loss": 0.779, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00042290780972266086, + "step": 2725 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5258, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.228478810675324e-06, + "loss": 0.6457, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004228478810675324, + "step": 2726 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5257, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.227879333780103e-06, + "loss": 0.8346, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004227879333780103, + "step": 2727 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5255, + "image_masks_loss": 0.0159, + "llm_learning_rate": 4.227279666606962e-06, + "loss": 0.7183, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004227279666606961, + "step": 2728 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5254, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.226679809221936e-06, + "loss": 0.7651, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00042266798092219357, + "step": 2729 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5253, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.226079761691083e-06, + "loss": 0.8003, + "output_text_mask_loss": 0.0824, + "predictor_learning_rate": 0.0004226079761691083, + "step": 2730 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5252, + "image_masks_loss": 0.0338, + "llm_learning_rate": 4.225479524080483e-06, + "loss": 0.8298, + "output_text_mask_loss": 0.0599, + "predictor_learning_rate": 0.0004225479524080482, + "step": 2731 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.525, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.2248790964562335e-06, + "loss": 0.7641, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042248790964562335, + "step": 2732 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5249, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.224278478884456e-06, + "loss": 0.7336, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042242784788844556, + "step": 2733 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5248, + "image_masks_loss": 0.0063, + "llm_learning_rate": 4.223677671431292e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004223677671431292, + "step": 2734 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5247, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.223076674162903e-06, + "loss": 0.8812, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004223076674162903, + "step": 2735 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5245, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.222475487145473e-06, + "loss": 0.5929, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00042224754871454727, + "step": 2736 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5244, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.221874110445206e-06, + "loss": 0.7863, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.00042218741104452055, + "step": 2737 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5243, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.221272544128326e-06, + "loss": 0.8062, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004221272544128326, + "step": 2738 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5242, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.22067078826108e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00042206707882610795, + "step": 2739 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.524, + "image_masks_loss": 0.0088, + "llm_learning_rate": 4.2200688429097345e-06, + "loss": 0.7293, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004220068842909734, + "step": 2740 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5239, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.219466708140576e-06, + "loss": 0.7454, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.0004219466708140576, + "step": 2741 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5238, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.218864384019914e-06, + "loss": 0.7729, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.0004218864384019913, + "step": 2742 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5237, + "image_masks_loss": 0.0395, + "llm_learning_rate": 4.218261870614077e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.00042182618706140763, + "step": 2743 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5235, + "image_masks_loss": 0.0193, + "llm_learning_rate": 4.217659167989416e-06, + "loss": 0.7595, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00042176591679894156, + "step": 2744 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5234, + "image_masks_loss": 0.0263, + "llm_learning_rate": 4.2170562762123e-06, + "loss": 0.74, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00042170562762123, + "step": 2745 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5233, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.216453195349124e-06, + "loss": 0.7493, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0004216453195349124, + "step": 2746 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5232, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.2158499254662985e-06, + "loss": 0.809, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00042158499254662987, + "step": 2747 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5231, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.215246466630257e-06, + "loss": 0.6335, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0004215246466630257, + "step": 2748 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5229, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.214642818907455e-06, + "loss": 0.8814, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 0.0004214642818907454, + "step": 2749 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5228, + "image_masks_loss": 0.0326, + "llm_learning_rate": 4.214038982364365e-06, + "loss": 0.8012, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0004214038982364365, + "step": 2750 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5227, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.2134349570674855e-06, + "loss": 0.8327, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004213434957067485, + "step": 2751 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5226, + "image_masks_loss": 0.0181, + "llm_learning_rate": 4.212830743083332e-06, + "loss": 0.8339, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00042128307430833314, + "step": 2752 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5224, + "image_masks_loss": 0.0603, + "llm_learning_rate": 4.212226340478441e-06, + "loss": 0.7954, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042122263404784414, + "step": 2753 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5223, + "image_masks_loss": 0.032, + "llm_learning_rate": 4.2116217493193726e-06, + "loss": 0.7363, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.0004211621749319372, + "step": 2754 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5222, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.211016969672704e-06, + "loss": 0.8294, + "output_text_mask_loss": 0.1106, + "predictor_learning_rate": 0.0004211016969672704, + "step": 2755 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5221, + "image_masks_loss": 0.029, + "llm_learning_rate": 4.210412001605035e-06, + "loss": 0.6313, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004210412001605035, + "step": 2756 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5219, + "image_masks_loss": 0.0075, + "llm_learning_rate": 4.209806845182988e-06, + "loss": 0.7584, + "output_text_mask_loss": 0.0591, + "predictor_learning_rate": 0.00042098068451829876, + "step": 2757 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5218, + "image_masks_loss": 0.0019, + "llm_learning_rate": 4.209201500473202e-06, + "loss": 0.7811, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0004209201500473202, + "step": 2758 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5217, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.20859596754234e-06, + "loss": 0.7194, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00042085959675423396, + "step": 2759 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5216, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.207990246457084e-06, + "loss": 0.8282, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.00042079902464570835, + "step": 2760 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5215, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.207384337284138e-06, + "loss": 0.8073, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.00042073843372841374, + "step": 2761 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5213, + "image_masks_loss": 0.0224, + "llm_learning_rate": 4.206778240090226e-06, + "loss": 0.8124, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004206778240090225, + "step": 2762 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5212, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.2061719549420906e-06, + "loss": 0.7778, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00042061719549420907, + "step": 2763 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5211, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.205565481906501e-06, + "loss": 0.7538, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004205565481906501, + "step": 2764 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.521, + "image_masks_loss": 0.0064, + "llm_learning_rate": 4.204958821050241e-06, + "loss": 0.8074, + "output_text_mask_loss": 0.1099, + "predictor_learning_rate": 0.0004204958821050241, + "step": 2765 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5208, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.204351972440118e-06, + "loss": 0.8712, + "output_text_mask_loss": 0.1312, + "predictor_learning_rate": 0.0004204351972440118, + "step": 2766 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5207, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.203744936142959e-06, + "loss": 0.7375, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0004203744936142959, + "step": 2767 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5206, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.203137712225614e-06, + "loss": 0.736, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00042031377122256136, + "step": 2768 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5205, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.20253030075495e-06, + "loss": 0.8735, + "output_text_mask_loss": 0.1289, + "predictor_learning_rate": 0.000420253030075495, + "step": 2769 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5203, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.201922701797856e-06, + "loss": 0.7164, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004201922701797856, + "step": 2770 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5202, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.2013149154212455e-06, + "loss": 0.6592, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004201314915421245, + "step": 2771 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5201, + "image_masks_loss": 0.0111, + "llm_learning_rate": 4.2007069416920445e-06, + "loss": 0.7696, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00042007069416920447, + "step": 2772 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.52, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.200098780677208e-06, + "loss": 0.8141, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004200098780677208, + "step": 2773 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5199, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.1994904324437066e-06, + "loss": 0.711, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004199490432443707, + "step": 2774 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5197, + "image_masks_loss": 0.0155, + "llm_learning_rate": 4.198881897058535e-06, + "loss": 0.7172, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004198881897058534, + "step": 2775 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5196, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.198273174588703e-06, + "loss": 0.7424, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.00041982731745887023, + "step": 2776 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5195, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.197664265101246e-06, + "loss": 0.7079, + "output_text_mask_loss": 0.1366, + "predictor_learning_rate": 0.00041976642651012455, + "step": 2777 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5194, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.197055168663219e-06, + "loss": 0.72, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00041970551686632186, + "step": 2778 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5192, + "image_masks_loss": 0.0234, + "llm_learning_rate": 4.196445885341697e-06, + "loss": 0.7521, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.0004196445885341697, + "step": 2779 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.5191, + "image_masks_loss": 0.0146, + "llm_learning_rate": 4.195836415203776e-06, + "loss": 0.8167, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004195836415203775, + "step": 2780 + }, + { + "epoch": 0.28, + "gumbel_tau": 0.519, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.195226758316571e-06, + "loss": 0.7032, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004195226758316571, + "step": 2781 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5189, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.19461691474722e-06, + "loss": 0.6845, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000419461691474722, + "step": 2782 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5188, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.194006884562879e-06, + "loss": 0.7591, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00041940068845628786, + "step": 2783 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5186, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.193396667830727e-06, + "loss": 0.8058, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0004193396667830727, + "step": 2784 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5185, + "image_masks_loss": 0.0031, + "llm_learning_rate": 4.192786264617961e-06, + "loss": 0.752, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.00041927862646179615, + "step": 2785 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5184, + "image_masks_loss": 0.0234, + "llm_learning_rate": 4.192175674991803e-06, + "loss": 0.8123, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.0004192175674991803, + "step": 2786 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5183, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.191564899019489e-06, + "loss": 0.7311, + "output_text_mask_loss": 0.0683, + "predictor_learning_rate": 0.0004191564899019489, + "step": 2787 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5181, + "image_masks_loss": 0.0406, + "llm_learning_rate": 4.190953936768282e-06, + "loss": 0.6822, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00041909539367682814, + "step": 2788 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.518, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.190342788305459e-06, + "loss": 0.7814, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004190342788305459, + "step": 2789 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5179, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.189731453698325e-06, + "loss": 0.7637, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00041897314536983244, + "step": 2790 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5178, + "image_masks_loss": 0.0273, + "llm_learning_rate": 4.1891199330141984e-06, + "loss": 0.8707, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00041891199330141985, + "step": 2791 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5177, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.188508226320422e-06, + "loss": 0.7202, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.0004188508226320422, + "step": 2792 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5175, + "image_masks_loss": 0.0236, + "llm_learning_rate": 4.187896333684361e-06, + "loss": 0.8939, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.000418789633368436, + "step": 2793 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5174, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.1872842551733934e-06, + "loss": 0.7807, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004187284255173393, + "step": 2794 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5173, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.186671990854927e-06, + "loss": 0.7232, + "output_text_mask_loss": 0.0439, + "predictor_learning_rate": 0.0004186671990854926, + "step": 2795 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5172, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.186059540796383e-06, + "loss": 0.7326, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00041860595407963827, + "step": 2796 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.517, + "image_masks_loss": 0.0246, + "llm_learning_rate": 4.185446905065207e-06, + "loss": 0.7083, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00041854469050652064, + "step": 2797 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5169, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.184834083728864e-06, + "loss": 0.7084, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0004184834083728863, + "step": 2798 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5168, + "image_masks_loss": 0.0086, + "llm_learning_rate": 4.1842210768548395e-06, + "loss": 0.739, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 0.0004184221076854839, + "step": 2799 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5167, + "image_masks_loss": 0.0038, + "llm_learning_rate": 4.1836078845106375e-06, + "loss": 0.8299, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.0004183607884510637, + "step": 2800 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5166, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.1829945067637855e-06, + "loss": 0.6598, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00041829945067637855, + "step": 2801 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5164, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.182380943681829e-06, + "loss": 0.8442, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.00041823809436818295, + "step": 2802 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5163, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.181767195332337e-06, + "loss": 0.7624, + "output_text_mask_loss": 0.0877, + "predictor_learning_rate": 0.00041817671953323364, + "step": 2803 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5162, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.181153261782894e-06, + "loss": 0.7971, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004181153261782894, + "step": 2804 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5161, + "image_masks_loss": 0.0284, + "llm_learning_rate": 4.1805391431011114e-06, + "loss": 0.7152, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00041805391431011106, + "step": 2805 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5159, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.179924839354613e-06, + "loss": 0.7481, + "output_text_mask_loss": 0.029, + "predictor_learning_rate": 0.0004179924839354613, + "step": 2806 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5158, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.17931035061105e-06, + "loss": 0.7804, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000417931035061105, + "step": 2807 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5157, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.178695676938091e-06, + "loss": 0.7765, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004178695676938091, + "step": 2808 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5156, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.178080818403425e-06, + "loss": 0.8477, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0004178080818403425, + "step": 2809 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5155, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.17746577507476e-06, + "loss": 0.7609, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000417746577507476, + "step": 2810 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5153, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.176850547019829e-06, + "loss": 0.7108, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.0004176850547019829, + "step": 2811 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5152, + "image_masks_loss": 0.0416, + "llm_learning_rate": 4.176235134306381e-06, + "loss": 0.7922, + "output_text_mask_loss": 0.066, + "predictor_learning_rate": 0.0004176235134306381, + "step": 2812 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5151, + "image_masks_loss": 0.0416, + "llm_learning_rate": 4.175619537002186e-06, + "loss": 0.7829, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00041756195370021857, + "step": 2813 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.515, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.175003755175034e-06, + "loss": 0.6792, + "output_text_mask_loss": 0.0406, + "predictor_learning_rate": 0.00041750037551750344, + "step": 2814 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5148, + "image_masks_loss": 0.0012, + "llm_learning_rate": 4.174387788892738e-06, + "loss": 0.7314, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00041743877888927384, + "step": 2815 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5147, + "image_masks_loss": 0.0286, + "llm_learning_rate": 4.173771638223131e-06, + "loss": 0.7405, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00041737716382231305, + "step": 2816 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5146, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.173155303234061e-06, + "loss": 0.763, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004173155303234061, + "step": 2817 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5145, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.172538783993404e-06, + "loss": 0.7929, + "output_text_mask_loss": 0.0683, + "predictor_learning_rate": 0.0004172538783993403, + "step": 2818 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5144, + "image_masks_loss": 0.0051, + "llm_learning_rate": 4.171922080569049e-06, + "loss": 0.6593, + "output_text_mask_loss": 0.0584, + "predictor_learning_rate": 0.00041719220805690486, + "step": 2819 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5142, + "image_masks_loss": 0.0332, + "llm_learning_rate": 4.171305193028911e-06, + "loss": 0.8455, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004171305193028911, + "step": 2820 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5141, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.170688121440922e-06, + "loss": 0.7763, + "output_text_mask_loss": 0.0225, + "predictor_learning_rate": 0.0004170688121440922, + "step": 2821 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.514, + "image_masks_loss": 0.0159, + "llm_learning_rate": 4.170070865873037e-06, + "loss": 0.7239, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.0004170070865873036, + "step": 2822 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5139, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.169453426393227e-06, + "loss": 0.7981, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00041694534263932273, + "step": 2823 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5138, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.168835803069487e-06, + "loss": 0.8019, + "output_text_mask_loss": 0.1068, + "predictor_learning_rate": 0.0004168835803069487, + "step": 2824 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5136, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.168217995969832e-06, + "loss": 0.7189, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0004168217995969832, + "step": 2825 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5135, + "image_masks_loss": 0.0439, + "llm_learning_rate": 4.167600005162295e-06, + "loss": 0.8403, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00041676000051622954, + "step": 2826 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5134, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.166981830714933e-06, + "loss": 0.7172, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0004166981830714932, + "step": 2827 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5133, + "image_masks_loss": 0.0336, + "llm_learning_rate": 4.1663634726958165e-06, + "loss": 0.8469, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004166363472695816, + "step": 2828 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5132, + "image_masks_loss": 0.0047, + "llm_learning_rate": 4.165744931173044e-06, + "loss": 0.7463, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.00041657449311730433, + "step": 2829 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.513, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.165126206214728e-06, + "loss": 0.8128, + "output_text_mask_loss": 0.1167, + "predictor_learning_rate": 0.0004165126206214728, + "step": 2830 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5129, + "image_masks_loss": 0.0088, + "llm_learning_rate": 4.164507297889005e-06, + "loss": 0.716, + "output_text_mask_loss": 0.0469, + "predictor_learning_rate": 0.00041645072978890055, + "step": 2831 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5128, + "image_masks_loss": 0.0216, + "llm_learning_rate": 4.163888206264033e-06, + "loss": 0.8344, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004163888206264032, + "step": 2832 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5127, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.163268931407983e-06, + "loss": 0.6796, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.0004163268931407983, + "step": 2833 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5125, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.162649473389056e-06, + "loss": 0.855, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.0004162649473389055, + "step": 2834 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5124, + "image_masks_loss": 0.0086, + "llm_learning_rate": 4.1620298322754635e-06, + "loss": 0.7722, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004162029832275464, + "step": 2835 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5123, + "image_masks_loss": 0.0063, + "llm_learning_rate": 4.161410008135444e-06, + "loss": 0.7173, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00041614100081354446, + "step": 2836 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5122, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.1607900010372556e-06, + "loss": 0.8731, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00041607900010372554, + "step": 2837 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5121, + "image_masks_loss": 0.0256, + "llm_learning_rate": 4.160169811049171e-06, + "loss": 0.8076, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00041601698110491715, + "step": 2838 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5119, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.15954943823949e-06, + "loss": 0.7759, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.000415954943823949, + "step": 2839 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5118, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.158928882676528e-06, + "loss": 0.7983, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0004158928882676528, + "step": 2840 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5117, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.158308144428623e-06, + "loss": 0.7962, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004158308144428622, + "step": 2841 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5116, + "image_masks_loss": 0.0182, + "llm_learning_rate": 4.157687223564131e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.00041576872235641303, + "step": 2842 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5115, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.157066120151429e-06, + "loss": 0.7252, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004157066120151429, + "step": 2843 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5113, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.1564448342589155e-06, + "loss": 0.7536, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004156444834258915, + "step": 2844 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5112, + "image_masks_loss": 0.036, + "llm_learning_rate": 4.155823365955007e-06, + "loss": 0.7816, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00041558233659550066, + "step": 2845 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5111, + "image_masks_loss": 0.0049, + "llm_learning_rate": 4.155201715308141e-06, + "loss": 0.6879, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00041552017153081414, + "step": 2846 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.511, + "image_masks_loss": 0.0189, + "llm_learning_rate": 4.154579882386775e-06, + "loss": 0.7616, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0004154579882386775, + "step": 2847 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5109, + "image_masks_loss": 0.0185, + "llm_learning_rate": 4.153957867259389e-06, + "loss": 0.7813, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.00041539578672593883, + "step": 2848 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5107, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.153335669994476e-06, + "loss": 0.8448, + "output_text_mask_loss": 0.1038, + "predictor_learning_rate": 0.00041533356699944765, + "step": 2849 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5106, + "image_masks_loss": 0.0213, + "llm_learning_rate": 4.1527132906605584e-06, + "loss": 0.8151, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004152713290660558, + "step": 2850 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5105, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.1520907293261705e-06, + "loss": 0.8386, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.000415209072932617, + "step": 2851 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5104, + "image_masks_loss": 0.0077, + "llm_learning_rate": 4.151467986059872e-06, + "loss": 0.7781, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004151467986059872, + "step": 2852 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5103, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.150845060930241e-06, + "loss": 0.7134, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00041508450609302407, + "step": 2853 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5101, + "image_masks_loss": 0.0063, + "llm_learning_rate": 4.1502219540058745e-06, + "loss": 0.8216, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004150221954005874, + "step": 2854 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.51, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.14959866535539e-06, + "loss": 0.7717, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00041495986653553897, + "step": 2855 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5099, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.1489751950474266e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.069, + "predictor_learning_rate": 0.0004148975195047426, + "step": 2856 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5098, + "image_masks_loss": 0.0341, + "llm_learning_rate": 4.148351543150642e-06, + "loss": 0.718, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004148351543150642, + "step": 2857 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5097, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.147727709733714e-06, + "loss": 0.8062, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00041477277097337135, + "step": 2858 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5095, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.14710369486534e-06, + "loss": 0.7165, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00041471036948653397, + "step": 2859 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5094, + "image_masks_loss": 0.0301, + "llm_learning_rate": 4.146479498614239e-06, + "loss": 0.6959, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00041464794986142387, + "step": 2860 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5093, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.145855121049147e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.0004145855121049147, + "step": 2861 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5092, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.145230562238824e-06, + "loss": 0.8794, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.0004145230562238823, + "step": 2862 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5091, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.144605822252047e-06, + "loss": 0.7607, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00041446058222520464, + "step": 2863 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5089, + "image_masks_loss": 0.0181, + "llm_learning_rate": 4.143980901157613e-06, + "loss": 0.7534, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0004143980901157613, + "step": 2864 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5088, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.143355799024341e-06, + "loss": 0.7361, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.000414335579902434, + "step": 2865 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5087, + "image_masks_loss": 0.0706, + "llm_learning_rate": 4.142730515921067e-06, + "loss": 0.7764, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0004142730515921067, + "step": 2866 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5086, + "image_masks_loss": 0.0444, + "llm_learning_rate": 4.142105051916651e-06, + "loss": 0.7466, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.000414210505191665, + "step": 2867 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5085, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.141479407079968e-06, + "loss": 0.7488, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 0.00041414794070799676, + "step": 2868 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5083, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.140853581479916e-06, + "loss": 0.7015, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.0004140853581479916, + "step": 2869 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5082, + "image_masks_loss": 0.0105, + "llm_learning_rate": 4.140227575185413e-06, + "loss": 0.7482, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00041402275751854133, + "step": 2870 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5081, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.139601388265397e-06, + "loss": 0.7449, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 0.00041396013882653973, + "step": 2871 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.508, + "image_masks_loss": 0.0212, + "llm_learning_rate": 4.138975020788823e-06, + "loss": 0.7295, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.0004138975020788823, + "step": 2872 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5079, + "image_masks_loss": 0.0425, + "llm_learning_rate": 4.138348472824671e-06, + "loss": 0.6973, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00041383484728246703, + "step": 2873 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5077, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.137721744441934e-06, + "loss": 0.7613, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0004137721744441933, + "step": 2874 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5076, + "image_masks_loss": 0.0032, + "llm_learning_rate": 4.13709483570963e-06, + "loss": 0.7109, + "output_text_mask_loss": 0.0435, + "predictor_learning_rate": 0.000413709483570963, + "step": 2875 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5075, + "image_masks_loss": 0.0372, + "llm_learning_rate": 4.136467746696797e-06, + "loss": 0.8971, + "output_text_mask_loss": 0.0028, + "predictor_learning_rate": 0.0004136467746696797, + "step": 2876 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5074, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.1358404774724915e-06, + "loss": 0.8332, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0004135840477472491, + "step": 2877 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5073, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.135213028105788e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004135213028105788, + "step": 2878 + }, + { + "epoch": 0.29, + "gumbel_tau": 0.5071, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.134585398665784e-06, + "loss": 0.8689, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.0004134585398665783, + "step": 2879 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.507, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.133957589221595e-06, + "loss": 0.8007, + "output_text_mask_loss": 0.0652, + "predictor_learning_rate": 0.00041339575892215943, + "step": 2880 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5069, + "image_masks_loss": 0.0359, + "llm_learning_rate": 4.133329599842356e-06, + "loss": 0.701, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.0004133329599842356, + "step": 2881 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5068, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.132701430597223e-06, + "loss": 0.7707, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004132701430597223, + "step": 2882 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5067, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.132073081555374e-06, + "loss": 0.8136, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.00041320730815553735, + "step": 2883 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5065, + "image_masks_loss": 0.0317, + "llm_learning_rate": 4.131444552786001e-06, + "loss": 0.7707, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00041314445527860003, + "step": 2884 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5064, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.130815844358319e-06, + "loss": 0.8275, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0004130815844358319, + "step": 2885 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5063, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.130186956341564e-06, + "loss": 0.7602, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004130186956341564, + "step": 2886 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5062, + "image_masks_loss": 0.0366, + "llm_learning_rate": 4.129557888804991e-06, + "loss": 0.873, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.0004129557888804991, + "step": 2887 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5061, + "image_masks_loss": 0.0408, + "llm_learning_rate": 4.128928641817873e-06, + "loss": 0.7833, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00041289286418178726, + "step": 2888 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5059, + "image_masks_loss": 0.0162, + "llm_learning_rate": 4.128299215449505e-06, + "loss": 0.8311, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00041282992154495046, + "step": 2889 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5058, + "image_masks_loss": 0.0222, + "llm_learning_rate": 4.1276696097692004e-06, + "loss": 0.8136, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00041276696097692, + "step": 2890 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5057, + "image_masks_loss": 0.0435, + "llm_learning_rate": 4.127039824846293e-06, + "loss": 0.7718, + "output_text_mask_loss": 0.0458, + "predictor_learning_rate": 0.00041270398248462924, + "step": 2891 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5056, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.1264098607501355e-06, + "loss": 0.8238, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00041264098607501356, + "step": 2892 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5055, + "image_masks_loss": 0.0366, + "llm_learning_rate": 4.1257797175501025e-06, + "loss": 0.7712, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00041257797175501024, + "step": 2893 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5053, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.125149395315584e-06, + "loss": 0.7744, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 0.00041251493953155846, + "step": 2894 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5052, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.1245188941159954e-06, + "loss": 0.7647, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00041245188941159953, + "step": 2895 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5051, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.123888214020768e-06, + "loss": 0.8077, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0004123888214020768, + "step": 2896 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.505, + "image_masks_loss": 0.0462, + "llm_learning_rate": 4.123257355099352e-06, + "loss": 0.8073, + "output_text_mask_loss": 0.0965, + "predictor_learning_rate": 0.00041232573550993524, + "step": 2897 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5049, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.1226263174212225e-06, + "loss": 0.7983, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.0004122626317421222, + "step": 2898 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5047, + "image_masks_loss": 0.0141, + "llm_learning_rate": 4.1219951010558665e-06, + "loss": 0.9071, + "output_text_mask_loss": 0.0486, + "predictor_learning_rate": 0.00041219951010558665, + "step": 2899 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5046, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.1213637060727986e-06, + "loss": 0.8286, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004121363706072798, + "step": 2900 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5045, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.120732132541547e-06, + "loss": 0.7765, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0004120732132541547, + "step": 2901 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5044, + "image_masks_loss": 0.0217, + "llm_learning_rate": 4.120100380531664e-06, + "loss": 0.907, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0004120100380531664, + "step": 2902 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5043, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.119468450112718e-06, + "loss": 0.7475, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0004119468450112718, + "step": 2903 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5042, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.1188363413542985e-06, + "loss": 0.7749, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00041188363413542986, + "step": 2904 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.504, + "image_masks_loss": 0.0277, + "llm_learning_rate": 4.118204054326016e-06, + "loss": 0.7695, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0004118204054326016, + "step": 2905 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5039, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.117571589097497e-06, + "loss": 0.7999, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.00041175715890974973, + "step": 2906 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5038, + "image_masks_loss": 0.0139, + "llm_learning_rate": 4.116938945738394e-06, + "loss": 0.776, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004116938945738393, + "step": 2907 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5037, + "image_masks_loss": 0.016, + "llm_learning_rate": 4.116306124318371e-06, + "loss": 0.7447, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004116306124318371, + "step": 2908 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5036, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.115673124907119e-06, + "loss": 0.7901, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.00041156731249071185, + "step": 2909 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5034, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.115039947574342e-06, + "loss": 0.7224, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.0004115039947574342, + "step": 2910 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5033, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.11440659238977e-06, + "loss": 0.7665, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00041144065923897694, + "step": 2911 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5032, + "image_masks_loss": 0.0207, + "llm_learning_rate": 4.113773059423147e-06, + "loss": 0.7203, + "output_text_mask_loss": 0.0885, + "predictor_learning_rate": 0.0004113773059423147, + "step": 2912 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5031, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.113139348744241e-06, + "loss": 0.8283, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00041131393487442403, + "step": 2913 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.503, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.112505460422836e-06, + "loss": 0.7513, + "output_text_mask_loss": 0.0507, + "predictor_learning_rate": 0.0004112505460422836, + "step": 2914 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5028, + "image_masks_loss": 0.0077, + "llm_learning_rate": 4.111871394528738e-06, + "loss": 0.7353, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00041118713945287376, + "step": 2915 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5027, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.111237151131772e-06, + "loss": 0.748, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004111237151131772, + "step": 2916 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5026, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.110602730301782e-06, + "loss": 0.7675, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00041106027303017826, + "step": 2917 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5025, + "image_masks_loss": 0.022, + "llm_learning_rate": 4.1099681321086325e-06, + "loss": 0.7108, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00041099681321086324, + "step": 2918 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5024, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.109333356622205e-06, + "loss": 0.77, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.0004109333356622205, + "step": 2919 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5023, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.108698403912405e-06, + "loss": 0.8789, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004108698403912405, + "step": 2920 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5021, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.108063274049153e-06, + "loss": 0.7639, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0004108063274049153, + "step": 2921 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.502, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.107427967102391e-06, + "loss": 0.7877, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004107427967102391, + "step": 2922 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5019, + "image_masks_loss": 0.0343, + "llm_learning_rate": 4.1067924831420815e-06, + "loss": 0.8412, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.0004106792483142081, + "step": 2923 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5018, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.106156822238204e-06, + "loss": 0.7681, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0004106156822238204, + "step": 2924 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5017, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.10552098446076e-06, + "loss": 0.7316, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000410552098446076, + "step": 2925 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5015, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.104884969879769e-06, + "loss": 0.7375, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0004104884969879769, + "step": 2926 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5014, + "image_masks_loss": 0.0334, + "llm_learning_rate": 4.104248778565271e-06, + "loss": 0.7938, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00041042487785652705, + "step": 2927 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5013, + "image_masks_loss": 0.013, + "llm_learning_rate": 4.103612410587323e-06, + "loss": 0.8398, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0004103612410587323, + "step": 2928 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5012, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.102975866016005e-06, + "loss": 0.7789, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0004102975866016005, + "step": 2929 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5011, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.102339144921415e-06, + "loss": 0.7458, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004102339144921415, + "step": 2930 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.501, + "image_masks_loss": 0.0267, + "llm_learning_rate": 4.101702247373669e-06, + "loss": 0.7565, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.00041017022473736686, + "step": 2931 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5008, + "image_masks_loss": 0.0204, + "llm_learning_rate": 4.101065173442903e-06, + "loss": 0.7296, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.00041010651734429027, + "step": 2932 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5007, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.100427923199274e-06, + "loss": 0.7172, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00041004279231992733, + "step": 2933 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5006, + "image_masks_loss": 0.0231, + "llm_learning_rate": 4.099790496712957e-06, + "loss": 0.7491, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004099790496712957, + "step": 2934 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5005, + "image_masks_loss": 0.0241, + "llm_learning_rate": 4.099152894054148e-06, + "loss": 0.6918, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004099152894054148, + "step": 2935 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5004, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.09851511529306e-06, + "loss": 0.761, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.000409851511529306, + "step": 2936 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5002, + "image_masks_loss": 0.0427, + "llm_learning_rate": 4.097877160499927e-06, + "loss": 0.7284, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0004097877160499926, + "step": 2937 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5001, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.097239029745002e-06, + "loss": 0.8045, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00040972390297450026, + "step": 2938 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.5, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.0966007230985574e-06, + "loss": 0.6666, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00040966007230985575, + "step": 2939 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4999, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.095962240630885e-06, + "loss": 0.6561, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0004095962240630885, + "step": 2940 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4998, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.095323582412296e-06, + "loss": 0.7337, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.00040953235824122954, + "step": 2941 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4997, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.094684748513119e-06, + "loss": 0.818, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004094684748513119, + "step": 2942 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4995, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.094045739003707e-06, + "loss": 0.7201, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0004094045739003707, + "step": 2943 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4994, + "image_masks_loss": 0.0077, + "llm_learning_rate": 4.093406553954427e-06, + "loss": 0.7821, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040934065539544273, + "step": 2944 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4993, + "image_masks_loss": 0.0223, + "llm_learning_rate": 4.092767193435668e-06, + "loss": 0.8725, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004092767193435668, + "step": 2945 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4992, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.0921276575178395e-06, + "loss": 0.6145, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0004092127657517839, + "step": 2946 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4991, + "image_masks_loss": 0.033, + "llm_learning_rate": 4.0914879462713655e-06, + "loss": 0.7352, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004091487946271365, + "step": 2947 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4989, + "image_masks_loss": 0.0261, + "llm_learning_rate": 4.090848059766694e-06, + "loss": 0.787, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 0.0004090848059766694, + "step": 2948 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4988, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.090207998074291e-06, + "loss": 0.8162, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.0004090207998074291, + "step": 2949 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4987, + "image_masks_loss": 0.0241, + "llm_learning_rate": 4.089567761264641e-06, + "loss": 0.7586, + "output_text_mask_loss": 0.0824, + "predictor_learning_rate": 0.0004089567761264641, + "step": 2950 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4986, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.088927349408249e-06, + "loss": 1.0179, + "output_text_mask_loss": 0.3021, + "predictor_learning_rate": 0.00040889273494082483, + "step": 2951 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4985, + "image_masks_loss": 0.0172, + "llm_learning_rate": 4.088286762575637e-06, + "loss": 1.0733, + "output_text_mask_loss": 0.1831, + "predictor_learning_rate": 0.0004088286762575637, + "step": 2952 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4984, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.08764600083735e-06, + "loss": 1.1992, + "output_text_mask_loss": 0.5798, + "predictor_learning_rate": 0.00040876460008373503, + "step": 2953 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4982, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.087005064263949e-06, + "loss": 1.3972, + "output_text_mask_loss": 0.7324, + "predictor_learning_rate": 0.00040870050642639487, + "step": 2954 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4981, + "image_masks_loss": 0.0205, + "llm_learning_rate": 4.0863639529260154e-06, + "loss": 1.4318, + "output_text_mask_loss": 0.9216, + "predictor_learning_rate": 0.0004086363952926015, + "step": 2955 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.498, + "image_masks_loss": 0.0271, + "llm_learning_rate": 4.0857226668941495e-06, + "loss": 1.223, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004085722666894149, + "step": 2956 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4979, + "image_masks_loss": 0.0119, + "llm_learning_rate": 4.085081206238972e-06, + "loss": 1.1252, + "output_text_mask_loss": 0.2884, + "predictor_learning_rate": 0.00040850812062389713, + "step": 2957 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4978, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.08443957103112e-06, + "loss": 0.868, + "output_text_mask_loss": 0.1953, + "predictor_learning_rate": 0.000408443957103112, + "step": 2958 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4977, + "image_masks_loss": 0.0254, + "llm_learning_rate": 4.083797761341255e-06, + "loss": 0.8464, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00040837977613412545, + "step": 2959 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4975, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.083155777240052e-06, + "loss": 0.9128, + "output_text_mask_loss": 0.119, + "predictor_learning_rate": 0.0004083155777240051, + "step": 2960 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4974, + "image_masks_loss": 0.0244, + "llm_learning_rate": 4.082513618798208e-06, + "loss": 0.7675, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.00040825136187982075, + "step": 2961 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4973, + "image_masks_loss": 0.0278, + "llm_learning_rate": 4.08187128608644e-06, + "loss": 0.7429, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00040818712860864393, + "step": 2962 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4972, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.081228779175482e-06, + "loss": 0.7693, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00040812287791754814, + "step": 2963 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4971, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.080586098136089e-06, + "loss": 0.8422, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004080586098136089, + "step": 2964 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4969, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.079943243039034e-06, + "loss": 0.882, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00040799432430390335, + "step": 2965 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4968, + "image_masks_loss": 0.0299, + "llm_learning_rate": 4.079300213955111e-06, + "loss": 0.8923, + "output_text_mask_loss": 0.0349, + "predictor_learning_rate": 0.00040793002139551105, + "step": 2966 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4967, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.07865701095513e-06, + "loss": 0.8217, + "output_text_mask_loss": 0.0938, + "predictor_learning_rate": 0.0004078657010955129, + "step": 2967 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4966, + "image_masks_loss": 0.0204, + "llm_learning_rate": 4.078013634109922e-06, + "loss": 0.8351, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 0.00040780136341099217, + "step": 2968 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4965, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.077370083490338e-06, + "loss": 0.7805, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00040773700834903384, + "step": 2969 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4964, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.076726359167248e-06, + "loss": 0.6241, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00040767263591672475, + "step": 2970 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4962, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.076082461211538e-06, + "loss": 0.8048, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 0.0004076082461211538, + "step": 2971 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4961, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.0754383896941175e-06, + "loss": 0.7999, + "output_text_mask_loss": 0.0828, + "predictor_learning_rate": 0.0004075438389694117, + "step": 2972 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.496, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.0747941446859115e-06, + "loss": 0.8224, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00040747941446859115, + "step": 2973 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4959, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.074149726257868e-06, + "loss": 0.8457, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00040741497262578675, + "step": 2974 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4958, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.07350513448095e-06, + "loss": 0.8509, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0004073505134480949, + "step": 2975 + }, + { + "epoch": 0.3, + "gumbel_tau": 0.4957, + "image_masks_loss": 0.0058, + "llm_learning_rate": 4.0728603694261416e-06, + "loss": 0.7822, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00040728603694261415, + "step": 2976 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4955, + "image_masks_loss": 0.0155, + "llm_learning_rate": 4.072215431164445e-06, + "loss": 0.8262, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004072215431164445, + "step": 2977 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4954, + "image_masks_loss": 0.0057, + "llm_learning_rate": 4.071570319766885e-06, + "loss": 0.7689, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.0004071570319766884, + "step": 2978 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4953, + "image_masks_loss": 0.023, + "llm_learning_rate": 4.070925035304498e-06, + "loss": 0.7892, + "output_text_mask_loss": 0.0288, + "predictor_learning_rate": 0.0004070925035304498, + "step": 2979 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4952, + "image_masks_loss": 0.0211, + "llm_learning_rate": 4.07027957784835e-06, + "loss": 0.7356, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004070279577848349, + "step": 2980 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4951, + "image_masks_loss": 0.0088, + "llm_learning_rate": 4.069633947469515e-06, + "loss": 0.7631, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004069633947469515, + "step": 2981 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.495, + "image_masks_loss": 0.0108, + "llm_learning_rate": 4.068988144239094e-06, + "loss": 0.8154, + "output_text_mask_loss": 0.0515, + "predictor_learning_rate": 0.00040689881442390944, + "step": 2982 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4948, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.068342168228204e-06, + "loss": 0.7679, + "output_text_mask_loss": 0.0969, + "predictor_learning_rate": 0.0004068342168228204, + "step": 2983 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4947, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.067696019507983e-06, + "loss": 0.7376, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.0004067696019507982, + "step": 2984 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4946, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.0670496981495825e-06, + "loss": 0.7085, + "output_text_mask_loss": 0.116, + "predictor_learning_rate": 0.0004067049698149582, + "step": 2985 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4945, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.066403204224178e-06, + "loss": 0.8139, + "output_text_mask_loss": 0.116, + "predictor_learning_rate": 0.0004066403204224178, + "step": 2986 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4944, + "image_masks_loss": 0.0038, + "llm_learning_rate": 4.065756537802965e-06, + "loss": 0.7606, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 0.0004065756537802965, + "step": 2987 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4943, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.065109698957154e-06, + "loss": 0.8079, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.00040651096989571535, + "step": 2988 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4941, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.064462687757976e-06, + "loss": 0.6593, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0004064462687757976, + "step": 2989 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.494, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.063815504276684e-06, + "loss": 0.7638, + "output_text_mask_loss": 0.1251, + "predictor_learning_rate": 0.0004063815504276683, + "step": 2990 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4939, + "image_masks_loss": 0.0114, + "llm_learning_rate": 4.063168148584545e-06, + "loss": 0.7273, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0004063168148584544, + "step": 2991 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4938, + "image_masks_loss": 0.0088, + "llm_learning_rate": 4.062520620752846e-06, + "loss": 0.8113, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00040625206207528454, + "step": 2992 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4937, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.061872920852897e-06, + "loss": 0.8413, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00040618729208528973, + "step": 2993 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4936, + "image_masks_loss": 0.0047, + "llm_learning_rate": 4.061225048956024e-06, + "loss": 0.7975, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0004061225048956023, + "step": 2994 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4934, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.0605770051335695e-06, + "loss": 0.7746, + "output_text_mask_loss": 0.0315, + "predictor_learning_rate": 0.0004060577005133569, + "step": 2995 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4933, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.0599287894569e-06, + "loss": 0.8365, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.0004059928789456899, + "step": 2996 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4932, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.059280401997396e-06, + "loss": 0.6607, + "output_text_mask_loss": 0.0725, + "predictor_learning_rate": 0.0004059280401997396, + "step": 2997 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4931, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.058631842826462e-06, + "loss": 0.7515, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00040586318428264623, + "step": 2998 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.493, + "image_masks_loss": 0.0187, + "llm_learning_rate": 4.057983112015519e-06, + "loss": 0.8684, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00040579831120155185, + "step": 2999 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4929, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.057334209636004e-06, + "loss": 0.7604, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.0004057334209636004, + "step": 3000 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4927, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.056685135759377e-06, + "loss": 0.7398, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004056685135759377, + "step": 3001 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4926, + "image_masks_loss": 0.0176, + "llm_learning_rate": 4.056035890457116e-06, + "loss": 0.7541, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004056035890457116, + "step": 3002 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4925, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.055386473800718e-06, + "loss": 0.8098, + "output_text_mask_loss": 0.0237, + "predictor_learning_rate": 0.00040553864738007174, + "step": 3003 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4924, + "image_masks_loss": 0.0607, + "llm_learning_rate": 4.054736885861695e-06, + "loss": 0.6839, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040547368858616953, + "step": 3004 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4923, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.054087126711585e-06, + "loss": 0.698, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0004054087126711584, + "step": 3005 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4922, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.053437196421937e-06, + "loss": 0.746, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.0004053437196421936, + "step": 3006 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.492, + "image_masks_loss": 0.0313, + "llm_learning_rate": 4.052787095064325e-06, + "loss": 0.7173, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004052787095064325, + "step": 3007 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4919, + "image_masks_loss": 0.0275, + "llm_learning_rate": 4.052136822710341e-06, + "loss": 0.7036, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004052136822710341, + "step": 3008 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4918, + "image_masks_loss": 0.0553, + "llm_learning_rate": 4.051486379431592e-06, + "loss": 0.8114, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00040514863794315916, + "step": 3009 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4917, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.050835765299707e-06, + "loss": 0.7776, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0004050835765299706, + "step": 3010 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4916, + "image_masks_loss": 0.0158, + "llm_learning_rate": 4.0501849803863325e-06, + "loss": 0.741, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0004050184980386332, + "step": 3011 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4915, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.049534024763136e-06, + "loss": 0.8568, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00040495340247631356, + "step": 3012 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4914, + "image_masks_loss": 0.036, + "llm_learning_rate": 4.048882898501801e-06, + "loss": 0.8647, + "output_text_mask_loss": 0.1495, + "predictor_learning_rate": 0.00040488828985018005, + "step": 3013 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4912, + "image_masks_loss": 0.0242, + "llm_learning_rate": 4.048231601674031e-06, + "loss": 0.8323, + "output_text_mask_loss": 0.1129, + "predictor_learning_rate": 0.00040482316016740306, + "step": 3014 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4911, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.0475801343515485e-06, + "loss": 0.7294, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00040475801343515486, + "step": 3015 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.491, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.046928496606095e-06, + "loss": 0.7221, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.00040469284966060946, + "step": 3016 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4909, + "image_masks_loss": 0.0374, + "llm_learning_rate": 4.046276688509429e-06, + "loss": 0.8928, + "output_text_mask_loss": 0.0584, + "predictor_learning_rate": 0.00040462766885094285, + "step": 3017 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4908, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.0456247101333305e-06, + "loss": 0.7116, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000404562471013333, + "step": 3018 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4907, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.044972561549596e-06, + "loss": 0.7228, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0004044972561549596, + "step": 3019 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4905, + "image_masks_loss": 0.0085, + "llm_learning_rate": 4.044320242830041e-06, + "loss": 0.774, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0004044320242830041, + "step": 3020 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4904, + "image_masks_loss": 0.0191, + "llm_learning_rate": 4.043667754046502e-06, + "loss": 0.8234, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004043667754046502, + "step": 3021 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4903, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.0430150952708315e-06, + "loss": 0.6838, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0004043015095270831, + "step": 3022 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4902, + "image_masks_loss": 0.0193, + "llm_learning_rate": 4.042362266574902e-06, + "loss": 0.787, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 0.0004042362266574902, + "step": 3023 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4901, + "image_masks_loss": 0.0259, + "llm_learning_rate": 4.041709268030603e-06, + "loss": 0.7802, + "output_text_mask_loss": 0.0702, + "predictor_learning_rate": 0.0004041709268030603, + "step": 3024 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.49, + "image_masks_loss": 0.0198, + "llm_learning_rate": 4.041056099709847e-06, + "loss": 0.7831, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00040410560997098467, + "step": 3025 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4898, + "image_masks_loss": 0.0444, + "llm_learning_rate": 4.040402761684559e-06, + "loss": 0.788, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0004040402761684559, + "step": 3026 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4897, + "image_masks_loss": 0.0142, + "llm_learning_rate": 4.039749254026689e-06, + "loss": 0.6978, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004039749254026689, + "step": 3027 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4896, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.039095576808201e-06, + "loss": 0.7768, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 0.0004039095576808201, + "step": 3028 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4895, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.03844173010108e-06, + "loss": 0.7683, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000403844173010108, + "step": 3029 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4894, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.037787713977329e-06, + "loss": 0.7796, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0004037787713977329, + "step": 3030 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4893, + "image_masks_loss": 0.0201, + "llm_learning_rate": 4.03713352850897e-06, + "loss": 0.8386, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004037133528508969, + "step": 3031 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4892, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.036479173768043e-06, + "loss": 0.789, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0004036479173768042, + "step": 3032 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.489, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.035824649826606e-06, + "loss": 0.7447, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004035824649826606, + "step": 3033 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4889, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.035169956756738e-06, + "loss": 0.733, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.0004035169956756738, + "step": 3034 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4888, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.0345150946305356e-06, + "loss": 0.8075, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00040345150946305354, + "step": 3035 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4887, + "image_masks_loss": 0.024, + "llm_learning_rate": 4.033860063520112e-06, + "loss": 0.7278, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00040338600635201125, + "step": 3036 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4886, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.033204863497603e-06, + "loss": 0.7638, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040332048634976025, + "step": 3037 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4885, + "image_masks_loss": 0.024, + "llm_learning_rate": 4.032549494635159e-06, + "loss": 0.8132, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0004032549494635158, + "step": 3038 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4883, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.0318939570049495e-06, + "loss": 0.7195, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00040318939570049497, + "step": 3039 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4882, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.031238250679167e-06, + "loss": 0.7879, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.0004031238250679167, + "step": 3040 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4881, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.030582375730016e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00040305823757300167, + "step": 3041 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.488, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.029926332229727e-06, + "loss": 0.761, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0004029926332229726, + "step": 3042 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4879, + "image_masks_loss": 0.0085, + "llm_learning_rate": 4.029270120250541e-06, + "loss": 0.7352, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040292701202505405, + "step": 3043 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4878, + "image_masks_loss": 0.0225, + "llm_learning_rate": 4.028613739864723e-06, + "loss": 0.822, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0004028613739864723, + "step": 3044 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4877, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.027957191144555e-06, + "loss": 0.8546, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.00040279571911445544, + "step": 3045 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4875, + "image_masks_loss": 0.0171, + "llm_learning_rate": 4.0273004741623375e-06, + "loss": 0.8538, + "output_text_mask_loss": 0.1167, + "predictor_learning_rate": 0.00040273004741623377, + "step": 3046 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4874, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.0266435889903906e-06, + "loss": 0.7419, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004026643588990391, + "step": 3047 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4873, + "image_masks_loss": 0.0235, + "llm_learning_rate": 4.025986535701051e-06, + "loss": 0.7928, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00040259865357010504, + "step": 3048 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4872, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.025329314366675e-06, + "loss": 0.7904, + "output_text_mask_loss": 0.0338, + "predictor_learning_rate": 0.00040253293143666753, + "step": 3049 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4871, + "image_masks_loss": 0.0209, + "llm_learning_rate": 4.024671925059637e-06, + "loss": 0.7656, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.00040246719250596375, + "step": 3050 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.487, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.024014367852331e-06, + "loss": 0.7561, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.00040240143678523313, + "step": 3051 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4869, + "image_masks_loss": 0.0075, + "llm_learning_rate": 4.023356642817169e-06, + "loss": 0.6975, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00040233566428171685, + "step": 3052 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4867, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.0226987500265794e-06, + "loss": 0.7676, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00040226987500265786, + "step": 3053 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4866, + "image_masks_loss": 0.038, + "llm_learning_rate": 4.022040689553012e-06, + "loss": 0.853, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00040220406895530113, + "step": 3054 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4865, + "image_masks_loss": 0.0196, + "llm_learning_rate": 4.0213824614689335e-06, + "loss": 0.6732, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0004021382461468933, + "step": 3055 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4864, + "image_masks_loss": 0.0147, + "llm_learning_rate": 4.0207240658468295e-06, + "loss": 0.8548, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0004020724065846829, + "step": 3056 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4863, + "image_masks_loss": 0.0242, + "llm_learning_rate": 4.0200655027592035e-06, + "loss": 0.7781, + "output_text_mask_loss": 0.0483, + "predictor_learning_rate": 0.00040200655027592037, + "step": 3057 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4862, + "image_masks_loss": 0.0211, + "llm_learning_rate": 4.019406772278579e-06, + "loss": 0.7703, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.00040194067722785793, + "step": 3058 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.486, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.018747874477498e-06, + "loss": 0.7632, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00040187478744774973, + "step": 3059 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4859, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.018088809428517e-06, + "loss": 0.6639, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.0004018088809428516, + "step": 3060 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4858, + "image_masks_loss": 0.005, + "llm_learning_rate": 4.017429577204214e-06, + "loss": 0.8486, + "output_text_mask_loss": 0.1411, + "predictor_learning_rate": 0.0004017429577204214, + "step": 3061 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4857, + "image_masks_loss": 0.0031, + "llm_learning_rate": 4.016770177877186e-06, + "loss": 0.7538, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040167701778771864, + "step": 3062 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4856, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.016110611520049e-06, + "loss": 0.682, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040161106115200486, + "step": 3063 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4855, + "image_masks_loss": 0.0412, + "llm_learning_rate": 4.015450878205433e-06, + "loss": 0.823, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0004015450878205433, + "step": 3064 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4854, + "image_masks_loss": 0.0083, + "llm_learning_rate": 4.014790978005992e-06, + "loss": 0.7918, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00040147909780059916, + "step": 3065 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4852, + "image_masks_loss": 0.0068, + "llm_learning_rate": 4.014130910994394e-06, + "loss": 0.8187, + "output_text_mask_loss": 0.1335, + "predictor_learning_rate": 0.00040141309109943934, + "step": 3066 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4851, + "image_masks_loss": 0.0233, + "llm_learning_rate": 4.013470677243327e-06, + "loss": 0.6967, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00040134706772433274, + "step": 3067 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.485, + "image_masks_loss": 0.028, + "llm_learning_rate": 4.012810276825498e-06, + "loss": 0.81, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0004012810276825498, + "step": 3068 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4849, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.012149709813633e-06, + "loss": 0.7597, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0004012149709813633, + "step": 3069 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4848, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.011488976280472e-06, + "loss": 0.7995, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0004011488976280472, + "step": 3070 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4847, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.010828076298779e-06, + "loss": 0.8094, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040108280762987794, + "step": 3071 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4846, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.010167009941333e-06, + "loss": 0.7742, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.00040101670099413326, + "step": 3072 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4844, + "image_masks_loss": 0.0055, + "llm_learning_rate": 4.0095057772809324e-06, + "loss": 0.6807, + "output_text_mask_loss": 0.0854, + "predictor_learning_rate": 0.0004009505777280932, + "step": 3073 + }, + { + "epoch": 0.31, + "gumbel_tau": 0.4843, + "image_masks_loss": 0.0542, + "llm_learning_rate": 4.008844378390393e-06, + "loss": 0.7523, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0004008844378390393, + "step": 3074 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4842, + "image_masks_loss": 0.006, + "llm_learning_rate": 4.00818281334255e-06, + "loss": 0.8316, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00040081828133425493, + "step": 3075 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4841, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.007521082210256e-06, + "loss": 0.8149, + "output_text_mask_loss": 0.0414, + "predictor_learning_rate": 0.00040075210822102555, + "step": 3076 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.484, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.006859185066382e-06, + "loss": 0.7895, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0004006859185066382, + "step": 3077 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4839, + "image_masks_loss": 0.0222, + "llm_learning_rate": 4.006197121983819e-06, + "loss": 0.8762, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.0004006197121983819, + "step": 3078 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4838, + "image_masks_loss": 0.0418, + "llm_learning_rate": 4.005534893035474e-06, + "loss": 0.7557, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040055348930354733, + "step": 3079 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4836, + "image_masks_loss": 0.0043, + "llm_learning_rate": 4.004872498294272e-06, + "loss": 0.7079, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00040048724982942726, + "step": 3080 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4835, + "image_masks_loss": 0.0181, + "llm_learning_rate": 4.00420993783316e-06, + "loss": 0.7916, + "output_text_mask_loss": 0.0237, + "predictor_learning_rate": 0.00040042099378331597, + "step": 3081 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4834, + "image_masks_loss": 0.0131, + "llm_learning_rate": 4.0035472117250985e-06, + "loss": 0.8149, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.0004003547211725098, + "step": 3082 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4833, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.002884320043069e-06, + "loss": 0.7577, + "output_text_mask_loss": 0.095, + "predictor_learning_rate": 0.0004002884320043069, + "step": 3083 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4832, + "image_masks_loss": 0.0058, + "llm_learning_rate": 4.002221262860071e-06, + "loss": 0.7625, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0004002221262860071, + "step": 3084 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4831, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.00155804024912e-06, + "loss": 0.7433, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00040015580402491204, + "step": 3085 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.483, + "image_masks_loss": 0.0064, + "llm_learning_rate": 4.000894652283255e-06, + "loss": 0.7571, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0004000894652283255, + "step": 3086 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4828, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.000231099035527e-06, + "loss": 0.835, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0004000231099035527, + "step": 3087 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4827, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.999567380579009e-06, + "loss": 0.7765, + "output_text_mask_loss": 0.0839, + "predictor_learning_rate": 0.0003999567380579009, + "step": 3088 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4826, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.99890349698679e-06, + "loss": 0.8467, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.000399890349698679, + "step": 3089 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4825, + "image_masks_loss": 0.0193, + "llm_learning_rate": 3.99823944833198e-06, + "loss": 0.7106, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.000399823944833198, + "step": 3090 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4824, + "image_masks_loss": 0.0146, + "llm_learning_rate": 3.9975752346877054e-06, + "loss": 0.6698, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003997575234687705, + "step": 3091 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4823, + "image_masks_loss": 0.0256, + "llm_learning_rate": 3.9969108561271085e-06, + "loss": 0.8614, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.00039969108561271086, + "step": 3092 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4822, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.996246312723355e-06, + "loss": 0.8958, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0003996246312723355, + "step": 3093 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4821, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.9955816045496245e-06, + "loss": 0.723, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0003995581604549624, + "step": 3094 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4819, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.994916731679117e-06, + "loss": 0.8243, + "output_text_mask_loss": 0.1289, + "predictor_learning_rate": 0.00039949167316791163, + "step": 3095 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4818, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.994251694185048e-06, + "loss": 0.6927, + "output_text_mask_loss": 0.0178, + "predictor_learning_rate": 0.0003994251694185048, + "step": 3096 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4817, + "image_masks_loss": 0.0188, + "llm_learning_rate": 3.993586492140656e-06, + "loss": 0.6686, + "output_text_mask_loss": 0.0931, + "predictor_learning_rate": 0.00039935864921406553, + "step": 3097 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4816, + "image_masks_loss": 0.0216, + "llm_learning_rate": 3.992921125619191e-06, + "loss": 0.7922, + "output_text_mask_loss": 0.0931, + "predictor_learning_rate": 0.00039929211256191913, + "step": 3098 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4815, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.992255594693927e-06, + "loss": 0.8405, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00039922555946939265, + "step": 3099 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4814, + "image_masks_loss": 0.0222, + "llm_learning_rate": 3.991589899438152e-06, + "loss": 0.784, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003991589899438152, + "step": 3100 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4813, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.990924039925176e-06, + "loss": 0.7104, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003990924039925176, + "step": 3101 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4811, + "image_masks_loss": 0.0043, + "llm_learning_rate": 3.9902580162283236e-06, + "loss": 0.7456, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.0003990258016228323, + "step": 3102 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.481, + "image_masks_loss": 0.02, + "llm_learning_rate": 3.989591828420939e-06, + "loss": 0.7738, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0003989591828420939, + "step": 3103 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4809, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.9889254765763845e-06, + "loss": 0.8026, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0003988925476576384, + "step": 3104 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4808, + "image_masks_loss": 0.0357, + "llm_learning_rate": 3.98825896076804e-06, + "loss": 0.766, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00039882589607680396, + "step": 3105 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4807, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.987592281069303e-06, + "loss": 0.7194, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003987592281069303, + "step": 3106 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4806, + "image_masks_loss": 0.0239, + "llm_learning_rate": 3.986925437553592e-06, + "loss": 0.7484, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003986925437553591, + "step": 3107 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4805, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.9862584302943385e-06, + "loss": 0.841, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00039862584302943386, + "step": 3108 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4803, + "image_masks_loss": 0.0111, + "llm_learning_rate": 3.985591259364997e-06, + "loss": 0.8207, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00039855912593649966, + "step": 3109 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4802, + "image_masks_loss": 0.0201, + "llm_learning_rate": 3.984923924839036e-06, + "loss": 0.7208, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0003984923924839036, + "step": 3110 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4801, + "image_masks_loss": 0.0042, + "llm_learning_rate": 3.984256426789946e-06, + "loss": 0.7689, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.00039842564267899454, + "step": 3111 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.48, + "image_masks_loss": 0.0199, + "llm_learning_rate": 3.983588765291231e-06, + "loss": 0.7806, + "output_text_mask_loss": 0.0477, + "predictor_learning_rate": 0.0003983588765291231, + "step": 3112 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4799, + "image_masks_loss": 0.0148, + "llm_learning_rate": 3.982920940416418e-06, + "loss": 0.7537, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0003982920940416418, + "step": 3113 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4798, + "image_masks_loss": 0.0067, + "llm_learning_rate": 3.9822529522390475e-06, + "loss": 0.5976, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003982252952239047, + "step": 3114 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4797, + "image_masks_loss": 0.0236, + "llm_learning_rate": 3.98158480083268e-06, + "loss": 0.7962, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003981584800832679, + "step": 3115 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4796, + "image_masks_loss": 0.0259, + "llm_learning_rate": 3.980916486270894e-06, + "loss": 0.7776, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003980916486270894, + "step": 3116 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4794, + "image_masks_loss": 0.0239, + "llm_learning_rate": 3.9802480086272865e-06, + "loss": 0.7674, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 0.00039802480086272864, + "step": 3117 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4793, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.9795793679754704e-06, + "loss": 0.7596, + "output_text_mask_loss": 0.1297, + "predictor_learning_rate": 0.000397957936797547, + "step": 3118 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4792, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.978910564389079e-06, + "loss": 0.7703, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.0003978910564389079, + "step": 3119 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4791, + "image_masks_loss": 0.0081, + "llm_learning_rate": 3.978241597941762e-06, + "loss": 0.7137, + "output_text_mask_loss": 0.1129, + "predictor_learning_rate": 0.0003978241597941762, + "step": 3120 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.479, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.977572468707188e-06, + "loss": 0.7889, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003977572468707188, + "step": 3121 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4789, + "image_masks_loss": 0.0095, + "llm_learning_rate": 3.976903176759043e-06, + "loss": 0.8116, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 0.00039769031767590427, + "step": 3122 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4788, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.97623372217103e-06, + "loss": 0.7898, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00039762337221710295, + "step": 3123 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4787, + "image_masks_loss": 0.033, + "llm_learning_rate": 3.975564105016871e-06, + "loss": 0.6234, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003975564105016871, + "step": 3124 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4785, + "image_masks_loss": 0.0269, + "llm_learning_rate": 3.974894325370306e-06, + "loss": 0.7093, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00039748943253703055, + "step": 3125 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4784, + "image_masks_loss": 0.006, + "llm_learning_rate": 3.974224383305093e-06, + "loss": 0.7824, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0003974224383305093, + "step": 3126 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4783, + "image_masks_loss": 0.0248, + "llm_learning_rate": 3.973554278895006e-06, + "loss": 0.8083, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00039735542788950064, + "step": 3127 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4782, + "image_masks_loss": 0.0383, + "llm_learning_rate": 3.972884012213842e-06, + "loss": 0.7506, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003972884012213841, + "step": 3128 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4781, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.972213583335408e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.0003972213583335408, + "step": 3129 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.478, + "image_masks_loss": 0.012, + "llm_learning_rate": 3.971542992333535e-06, + "loss": 0.7092, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0003971542992333536, + "step": 3130 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4779, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.970872239282071e-06, + "loss": 0.731, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00039708722392820705, + "step": 3131 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4778, + "image_masks_loss": 0.0198, + "llm_learning_rate": 3.970201324254879e-06, + "loss": 0.714, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.0003970201324254879, + "step": 3132 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4776, + "image_masks_loss": 0.017, + "llm_learning_rate": 3.969530247325843e-06, + "loss": 0.7624, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0003969530247325843, + "step": 3133 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4775, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.968859008568863e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00039688590085688625, + "step": 3134 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4774, + "image_masks_loss": 0.0138, + "llm_learning_rate": 3.968187608057857e-06, + "loss": 0.7815, + "output_text_mask_loss": 0.0135, + "predictor_learning_rate": 0.0003968187608057857, + "step": 3135 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4773, + "image_masks_loss": 0.011, + "llm_learning_rate": 3.9675160458667606e-06, + "loss": 0.8053, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.000396751604586676, + "step": 3136 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4772, + "image_masks_loss": 0.0368, + "llm_learning_rate": 3.96684432206953e-06, + "loss": 0.7513, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00039668443220695297, + "step": 3137 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4771, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.966172436740134e-06, + "loss": 0.733, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00039661724367401335, + "step": 3138 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.477, + "image_masks_loss": 0.0132, + "llm_learning_rate": 3.965500389952563e-06, + "loss": 0.5588, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0003965500389952563, + "step": 3139 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4768, + "image_masks_loss": 0.0184, + "llm_learning_rate": 3.9648281817808265e-06, + "loss": 0.7566, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.0003964828181780826, + "step": 3140 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4767, + "image_masks_loss": 0.0248, + "llm_learning_rate": 3.964155812298947e-06, + "loss": 0.809, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003964155812298947, + "step": 3141 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4766, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.963483281580969e-06, + "loss": 0.6745, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00039634832815809687, + "step": 3142 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4765, + "image_masks_loss": 0.0119, + "llm_learning_rate": 3.962810589700951e-06, + "loss": 0.7656, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00039628105897009504, + "step": 3143 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4764, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.962137736732973e-06, + "loss": 0.7099, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.00039621377367329727, + "step": 3144 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4763, + "image_masks_loss": 0.0184, + "llm_learning_rate": 3.961464722751131e-06, + "loss": 0.7533, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.000396146472275113, + "step": 3145 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4762, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.960791547829538e-06, + "loss": 0.8018, + "output_text_mask_loss": 0.1778, + "predictor_learning_rate": 0.00039607915478295373, + "step": 3146 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4761, + "image_masks_loss": 0.015, + "llm_learning_rate": 3.960118212042326e-06, + "loss": 0.7733, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003960118212042326, + "step": 3147 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.476, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.959444715463644e-06, + "loss": 0.9242, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.00039594447154636436, + "step": 3148 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4758, + "image_masks_loss": 0.0275, + "llm_learning_rate": 3.95877105816766e-06, + "loss": 0.756, + "output_text_mask_loss": 0.0311, + "predictor_learning_rate": 0.00039587710581676593, + "step": 3149 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4757, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.958097240228557e-06, + "loss": 0.7335, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0003958097240228556, + "step": 3150 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4756, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.957423261720539e-06, + "loss": 0.6857, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00039574232617205386, + "step": 3151 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4755, + "image_masks_loss": 0.0235, + "llm_learning_rate": 3.956749122717824e-06, + "loss": 0.7678, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00039567491227178236, + "step": 3152 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4754, + "image_masks_loss": 0.0322, + "llm_learning_rate": 3.956074823294651e-06, + "loss": 0.8525, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0003956074823294651, + "step": 3153 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4753, + "image_masks_loss": 0.0087, + "llm_learning_rate": 3.955400363525276e-06, + "loss": 0.7553, + "output_text_mask_loss": 0.0328, + "predictor_learning_rate": 0.00039554003635252754, + "step": 3154 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4752, + "image_masks_loss": 0.0111, + "llm_learning_rate": 3.95472574348397e-06, + "loss": 0.7135, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00039547257434839703, + "step": 3155 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4751, + "image_masks_loss": 0.02, + "llm_learning_rate": 3.954050963245026e-06, + "loss": 0.6838, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003954050963245026, + "step": 3156 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4749, + "image_masks_loss": 0.0345, + "llm_learning_rate": 3.953376022882751e-06, + "loss": 0.7586, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003953376022882751, + "step": 3157 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4748, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.9527009224714704e-06, + "loss": 0.8063, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00039527009224714707, + "step": 3158 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4747, + "image_masks_loss": 0.0343, + "llm_learning_rate": 3.952025662085529e-06, + "loss": 0.6713, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.0003952025662085529, + "step": 3159 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4746, + "image_masks_loss": 0.0052, + "llm_learning_rate": 3.951350241799288e-06, + "loss": 0.7171, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00039513502417992875, + "step": 3160 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4745, + "image_masks_loss": 0.0073, + "llm_learning_rate": 3.950674661687125e-06, + "loss": 0.7715, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00039506746616871244, + "step": 3161 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4744, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.949998921823436e-06, + "loss": 0.9248, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003949998921823436, + "step": 3162 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4743, + "image_masks_loss": 0.0105, + "llm_learning_rate": 3.9493230222826374e-06, + "loss": 0.738, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003949323022282637, + "step": 3163 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4742, + "image_masks_loss": 0.0441, + "llm_learning_rate": 3.948646963139159e-06, + "loss": 0.8358, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00039486469631391585, + "step": 3164 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.474, + "image_masks_loss": 0.0231, + "llm_learning_rate": 3.9479707444674494e-06, + "loss": 0.6352, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00039479707444674495, + "step": 3165 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4739, + "image_masks_loss": 0.0158, + "llm_learning_rate": 3.947294366341977e-06, + "loss": 0.7765, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0003947294366341977, + "step": 3166 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4738, + "image_masks_loss": 0.0177, + "llm_learning_rate": 3.9466178288372255e-06, + "loss": 0.7088, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003946617828837225, + "step": 3167 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4737, + "image_masks_loss": 0.0048, + "llm_learning_rate": 3.945941132027697e-06, + "loss": 0.7223, + "output_text_mask_loss": 0.0097, + "predictor_learning_rate": 0.0003945941132027696, + "step": 3168 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4736, + "image_masks_loss": 0.0064, + "llm_learning_rate": 3.945264275987908e-06, + "loss": 0.7314, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0003945264275987908, + "step": 3169 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4735, + "image_masks_loss": 0.0366, + "llm_learning_rate": 3.9445872607924e-06, + "loss": 0.729, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00039445872607923996, + "step": 3170 + }, + { + "epoch": 0.32, + "gumbel_tau": 0.4734, + "image_masks_loss": 0.0168, + "llm_learning_rate": 3.9439100865157245e-06, + "loss": 0.7761, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0003943910086515724, + "step": 3171 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4733, + "image_masks_loss": 0.005, + "llm_learning_rate": 3.943232753232454e-06, + "loss": 0.757, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00039432327532324535, + "step": 3172 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4732, + "image_masks_loss": 0.019, + "llm_learning_rate": 3.942555261017178e-06, + "loss": 0.8314, + "output_text_mask_loss": 0.0614, + "predictor_learning_rate": 0.0003942555261017178, + "step": 3173 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.473, + "image_masks_loss": 0.0271, + "llm_learning_rate": 3.9418776099445035e-06, + "loss": 0.7555, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0003941877609944503, + "step": 3174 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4729, + "image_masks_loss": 0.0067, + "llm_learning_rate": 3.941199800089054e-06, + "loss": 0.7815, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0003941199800089054, + "step": 3175 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4728, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.940521831525474e-06, + "loss": 0.7399, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00039405218315254737, + "step": 3176 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4727, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.93984370432842e-06, + "loss": 0.8393, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.000393984370432842, + "step": 3177 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4726, + "image_masks_loss": 0.0126, + "llm_learning_rate": 3.9391654185725714e-06, + "loss": 0.655, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0003939165418572571, + "step": 3178 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4725, + "image_masks_loss": 0.01, + "llm_learning_rate": 3.9384869743326206e-06, + "loss": 0.7909, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000393848697433262, + "step": 3179 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4724, + "image_masks_loss": 0.0173, + "llm_learning_rate": 3.93780837168328e-06, + "loss": 0.6907, + "output_text_mask_loss": 0.0645, + "predictor_learning_rate": 0.00039378083716832803, + "step": 3180 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4723, + "image_masks_loss": 0.012, + "llm_learning_rate": 3.937129610699279e-06, + "loss": 0.8001, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.00039371296106992785, + "step": 3181 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4721, + "image_masks_loss": 0.0217, + "llm_learning_rate": 3.936450691455364e-06, + "loss": 0.6726, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00039364506914553644, + "step": 3182 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.472, + "image_masks_loss": 0.0166, + "llm_learning_rate": 3.9357716140263e-06, + "loss": 0.8091, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00039357716140262997, + "step": 3183 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4719, + "image_masks_loss": 0.0048, + "llm_learning_rate": 3.935092378486867e-06, + "loss": 0.7683, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00039350923784868666, + "step": 3184 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4718, + "image_masks_loss": 0.0157, + "llm_learning_rate": 3.934412984911865e-06, + "loss": 0.8142, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0003934412984911865, + "step": 3185 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4717, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.93373343337611e-06, + "loss": 0.7685, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.000393373343337611, + "step": 3186 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4716, + "image_masks_loss": 0.0452, + "llm_learning_rate": 3.9330537239544355e-06, + "loss": 0.8291, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00039330537239544357, + "step": 3187 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4715, + "image_masks_loss": 0.0056, + "llm_learning_rate": 3.932373856721694e-06, + "loss": 0.8697, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003932373856721694, + "step": 3188 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4714, + "image_masks_loss": 0.0227, + "llm_learning_rate": 3.931693831752752e-06, + "loss": 0.7525, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00039316938317527516, + "step": 3189 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4713, + "image_masks_loss": 0.0057, + "llm_learning_rate": 3.931013649122497e-06, + "loss": 0.7274, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0003931013649122497, + "step": 3190 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4711, + "image_masks_loss": 0.0076, + "llm_learning_rate": 3.93033330890583e-06, + "loss": 0.752, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00039303333089058304, + "step": 3191 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.471, + "image_masks_loss": 0.0259, + "llm_learning_rate": 3.929652811177674e-06, + "loss": 0.7447, + "output_text_mask_loss": 0.0832, + "predictor_learning_rate": 0.0003929652811177674, + "step": 3192 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4709, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.928972156012967e-06, + "loss": 0.7693, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00039289721560129665, + "step": 3193 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4708, + "image_masks_loss": 0.0112, + "llm_learning_rate": 3.928291343486662e-06, + "loss": 0.6966, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003928291343486662, + "step": 3194 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4707, + "image_masks_loss": 0.0026, + "llm_learning_rate": 3.927610373673734e-06, + "loss": 0.7241, + "output_text_mask_loss": 0.0011, + "predictor_learning_rate": 0.00039276103736737334, + "step": 3195 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4706, + "image_masks_loss": 0.0075, + "llm_learning_rate": 3.92692924664917e-06, + "loss": 0.8373, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.000392692924664917, + "step": 3196 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4705, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.92624796248798e-06, + "loss": 0.7692, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00039262479624879796, + "step": 3197 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4704, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.925566521265188e-06, + "loss": 0.7927, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003925566521265188, + "step": 3198 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4703, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.924884923055835e-06, + "loss": 0.7215, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.0003924884923055835, + "step": 3199 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4701, + "image_masks_loss": 0.0214, + "llm_learning_rate": 3.92420316793498e-06, + "loss": 0.784, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.000392420316793498, + "step": 3200 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.47, + "image_masks_loss": 0.0222, + "llm_learning_rate": 3.9235212559777e-06, + "loss": 0.7842, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00039235212559777, + "step": 3201 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4699, + "image_masks_loss": 0.0183, + "llm_learning_rate": 3.922839187259089e-06, + "loss": 0.742, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.0003922839187259088, + "step": 3202 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4698, + "image_masks_loss": 0.008, + "llm_learning_rate": 3.922156961854257e-06, + "loss": 0.718, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00039221569618542567, + "step": 3203 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4697, + "image_masks_loss": 0.0211, + "llm_learning_rate": 3.921474579838333e-06, + "loss": 0.6349, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003921474579838333, + "step": 3204 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4696, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.920792041286462e-06, + "loss": 0.7619, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00039207920412864615, + "step": 3205 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4695, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.920109346273806e-06, + "loss": 0.8762, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.00039201093462738057, + "step": 3206 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4694, + "image_masks_loss": 0.0305, + "llm_learning_rate": 3.9194264948755465e-06, + "loss": 0.6979, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00039194264948755466, + "step": 3207 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4693, + "image_masks_loss": 0.0145, + "llm_learning_rate": 3.918743487166879e-06, + "loss": 0.6927, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.0003918743487166879, + "step": 3208 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4691, + "image_masks_loss": 0.0309, + "llm_learning_rate": 3.9180603232230195e-06, + "loss": 0.7815, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003918060323223019, + "step": 3209 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.469, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.917377003119198e-06, + "loss": 0.7927, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0003917377003119198, + "step": 3210 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4689, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.916693526930664e-06, + "loss": 0.732, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 0.00039166935269306636, + "step": 3211 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4688, + "image_masks_loss": 0.007, + "llm_learning_rate": 3.9160098947326835e-06, + "loss": 0.786, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.0003916009894732683, + "step": 3212 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4687, + "image_masks_loss": 0.0286, + "llm_learning_rate": 3.915326106600539e-06, + "loss": 0.7346, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003915326106600539, + "step": 3213 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4686, + "image_masks_loss": 0.0186, + "llm_learning_rate": 3.914642162609531e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.00039146421626095305, + "step": 3214 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4685, + "image_masks_loss": 0.017, + "llm_learning_rate": 3.913958062834977e-06, + "loss": 0.6445, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003913958062834977, + "step": 3215 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4684, + "image_masks_loss": 0.0184, + "llm_learning_rate": 3.913273807352212e-06, + "loss": 0.8388, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.00039132738073522125, + "step": 3216 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4683, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.912589396236588e-06, + "loss": 0.7245, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003912589396236588, + "step": 3217 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4682, + "image_masks_loss": 0.0163, + "llm_learning_rate": 3.911904829563472e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0003911904829563472, + "step": 3218 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.468, + "image_masks_loss": 0.0137, + "llm_learning_rate": 3.911220107408252e-06, + "loss": 0.8144, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003911220107408252, + "step": 3219 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4679, + "image_masks_loss": 0.0084, + "llm_learning_rate": 3.910535229846331e-06, + "loss": 0.8016, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000391053522984633, + "step": 3220 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4678, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.909850196953127e-06, + "loss": 0.6685, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00039098501969531265, + "step": 3221 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4677, + "image_masks_loss": 0.0286, + "llm_learning_rate": 3.90916500880408e-06, + "loss": 0.7106, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.000390916500880408, + "step": 3222 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4676, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.9084796654746436e-06, + "loss": 0.8388, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003908479665474643, + "step": 3223 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4675, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.907794167040289e-06, + "loss": 0.742, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.0003907794167040288, + "step": 3224 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4674, + "image_masks_loss": 0.0443, + "llm_learning_rate": 3.907108513576504e-06, + "loss": 0.7829, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003907108513576504, + "step": 3225 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4673, + "image_masks_loss": 0.0073, + "llm_learning_rate": 3.906422705158797e-06, + "loss": 0.7766, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003906422705158796, + "step": 3226 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4672, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.905736741862688e-06, + "loss": 0.8181, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.0003905736741862688, + "step": 3227 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4671, + "image_masks_loss": 0.0129, + "llm_learning_rate": 3.905050623763717e-06, + "loss": 0.775, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.00039050506237637173, + "step": 3228 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4669, + "image_masks_loss": 0.0079, + "llm_learning_rate": 3.904364350937443e-06, + "loss": 0.6821, + "output_text_mask_loss": 0.0462, + "predictor_learning_rate": 0.00039043643509374426, + "step": 3229 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4668, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.903677923459438e-06, + "loss": 0.8428, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.00039036779234594376, + "step": 3230 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4667, + "image_masks_loss": 0.0219, + "llm_learning_rate": 3.902991341405295e-06, + "loss": 0.817, + "output_text_mask_loss": 0.0387, + "predictor_learning_rate": 0.0003902991341405294, + "step": 3231 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4666, + "image_masks_loss": 0.0171, + "llm_learning_rate": 3.9023046048506185e-06, + "loss": 0.6785, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.00039023046048506185, + "step": 3232 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4665, + "image_masks_loss": 0.0142, + "llm_learning_rate": 3.9016177138710356e-06, + "loss": 0.7524, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003901617713871036, + "step": 3233 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4664, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.900930668542188e-06, + "loss": 0.7893, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.0003900930668542189, + "step": 3234 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4663, + "image_masks_loss": 0.0131, + "llm_learning_rate": 3.9002434689397365e-06, + "loss": 0.6535, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0003900243468939736, + "step": 3235 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4662, + "image_masks_loss": 0.0448, + "llm_learning_rate": 3.899556115139354e-06, + "loss": 0.7383, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0003899556115139354, + "step": 3236 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4661, + "image_masks_loss": 0.0146, + "llm_learning_rate": 3.898868607216735e-06, + "loss": 0.7602, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.00038988686072167345, + "step": 3237 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.466, + "image_masks_loss": 0.0119, + "llm_learning_rate": 3.898180945247589e-06, + "loss": 0.7218, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003898180945247588, + "step": 3238 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4658, + "image_masks_loss": 0.0433, + "llm_learning_rate": 3.897493129307643e-06, + "loss": 0.7847, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.0003897493129307642, + "step": 3239 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4657, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.89680515947264e-06, + "loss": 0.7335, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.00038968051594726397, + "step": 3240 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4656, + "image_masks_loss": 0.0383, + "llm_learning_rate": 3.896117035818342e-06, + "loss": 0.7757, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038961170358183417, + "step": 3241 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4655, + "image_masks_loss": 0.0098, + "llm_learning_rate": 3.895428758420526e-06, + "loss": 0.7453, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00038954287584205263, + "step": 3242 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4654, + "image_masks_loss": 0.0185, + "llm_learning_rate": 3.8947403273549886e-06, + "loss": 0.8952, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038947403273549883, + "step": 3243 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4653, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.894051742697539e-06, + "loss": 0.704, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00038940517426975385, + "step": 3244 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4652, + "image_masks_loss": 0.0507, + "llm_learning_rate": 3.893363004524006e-06, + "loss": 0.709, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038933630045240054, + "step": 3245 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4651, + "image_masks_loss": 0.0307, + "llm_learning_rate": 3.8926741129102344e-06, + "loss": 0.774, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0003892674112910234, + "step": 3246 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.465, + "image_masks_loss": 0.0248, + "llm_learning_rate": 3.891985067932089e-06, + "loss": 0.8542, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00038919850679320885, + "step": 3247 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4649, + "image_masks_loss": 0.0238, + "llm_learning_rate": 3.891295869665446e-06, + "loss": 0.8374, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.00038912958696654454, + "step": 3248 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4647, + "image_masks_loss": 0.0298, + "llm_learning_rate": 3.890606518186204e-06, + "loss": 0.8076, + "output_text_mask_loss": 0.0622, + "predictor_learning_rate": 0.0003890606518186204, + "step": 3249 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4646, + "image_masks_loss": 0.0309, + "llm_learning_rate": 3.889917013570274e-06, + "loss": 0.8175, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038899170135702735, + "step": 3250 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4645, + "image_masks_loss": 0.0284, + "llm_learning_rate": 3.889227355893587e-06, + "loss": 0.7376, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00038892273558935867, + "step": 3251 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4644, + "image_masks_loss": 0.0326, + "llm_learning_rate": 3.888537545232088e-06, + "loss": 0.7722, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0003888537545232088, + "step": 3252 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4643, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.887847581661744e-06, + "loss": 0.8317, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0003887847581661743, + "step": 3253 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4642, + "image_masks_loss": 0.0475, + "llm_learning_rate": 3.887157465258531e-06, + "loss": 0.8334, + "output_text_mask_loss": 0.0481, + "predictor_learning_rate": 0.0003887157465258531, + "step": 3254 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4641, + "image_masks_loss": 0.0259, + "llm_learning_rate": 3.886467196098449e-06, + "loss": 0.7782, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0003886467196098449, + "step": 3255 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.464, + "image_masks_loss": 0.0191, + "llm_learning_rate": 3.8857767742575115e-06, + "loss": 0.7285, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0003885776774257511, + "step": 3256 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4639, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.885086199811748e-06, + "loss": 0.7811, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003885086199811748, + "step": 3257 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4638, + "image_masks_loss": 0.0177, + "llm_learning_rate": 3.8843954728372075e-06, + "loss": 0.682, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0003884395472837207, + "step": 3258 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4636, + "image_masks_loss": 0.0614, + "llm_learning_rate": 3.883704593409954e-06, + "loss": 0.7403, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.00038837045934099533, + "step": 3259 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4635, + "image_masks_loss": 0.0546, + "llm_learning_rate": 3.8830135616060685e-06, + "loss": 0.8104, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003883013561606068, + "step": 3260 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4634, + "image_masks_loss": 0.0151, + "llm_learning_rate": 3.882322377501648e-06, + "loss": 0.7616, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.0003882322377501648, + "step": 3261 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4633, + "image_masks_loss": 0.0374, + "llm_learning_rate": 3.8816310411728105e-06, + "loss": 0.8375, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.000388163104117281, + "step": 3262 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4632, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.880939552695684e-06, + "loss": 0.8071, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00038809395526956836, + "step": 3263 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4631, + "image_masks_loss": 0.0161, + "llm_learning_rate": 3.880247912146418e-06, + "loss": 0.6716, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00038802479121464176, + "step": 3264 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.463, + "image_masks_loss": 0.0197, + "llm_learning_rate": 3.879556119601178e-06, + "loss": 0.8187, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.0003879556119601178, + "step": 3265 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4629, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.878864175136144e-06, + "loss": 0.7819, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038788641751361444, + "step": 3266 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4628, + "image_masks_loss": 0.037, + "llm_learning_rate": 3.878172078827517e-06, + "loss": 0.772, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00038781720788275164, + "step": 3267 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4627, + "image_masks_loss": 0.0492, + "llm_learning_rate": 3.87747983075151e-06, + "loss": 0.7787, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038774798307515104, + "step": 3268 + }, + { + "epoch": 0.33, + "gumbel_tau": 0.4626, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.876787430984357e-06, + "loss": 0.7834, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003876787430984357, + "step": 3269 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4624, + "image_masks_loss": 0.0216, + "llm_learning_rate": 3.876094879602305e-06, + "loss": 0.798, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.00038760948796023046, + "step": 3270 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4623, + "image_masks_loss": 0.0332, + "llm_learning_rate": 3.875402176681619e-06, + "loss": 0.7915, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00038754021766816184, + "step": 3271 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4622, + "image_masks_loss": 0.0273, + "llm_learning_rate": 3.874709322298582e-06, + "loss": 0.7643, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0003874709322298582, + "step": 3272 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4621, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.874016316529493e-06, + "loss": 0.792, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003874016316529493, + "step": 3273 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.462, + "image_masks_loss": 0.0203, + "llm_learning_rate": 3.873323159450666e-06, + "loss": 0.7811, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00038733231594506655, + "step": 3274 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4619, + "image_masks_loss": 0.0326, + "llm_learning_rate": 3.872629851138434e-06, + "loss": 0.755, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0003872629851138434, + "step": 3275 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4618, + "image_masks_loss": 0.0142, + "llm_learning_rate": 3.8719363916691456e-06, + "loss": 0.769, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003871936391669145, + "step": 3276 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4617, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.871242781119166e-06, + "loss": 0.7361, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003871242781119166, + "step": 3277 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4616, + "image_masks_loss": 0.029, + "llm_learning_rate": 3.870549019564877e-06, + "loss": 0.8107, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038705490195648764, + "step": 3278 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4615, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.869855107082676e-06, + "loss": 0.7958, + "output_text_mask_loss": 0.0296, + "predictor_learning_rate": 0.0003869855107082676, + "step": 3279 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4614, + "image_masks_loss": 0.0095, + "llm_learning_rate": 3.869161043748981e-06, + "loss": 0.8462, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.00038691610437489805, + "step": 3280 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4612, + "image_masks_loss": 0.0355, + "llm_learning_rate": 3.8684668296402215e-06, + "loss": 0.7575, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00038684668296402213, + "step": 3281 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4611, + "image_masks_loss": 0.0111, + "llm_learning_rate": 3.8677724648328465e-06, + "loss": 0.7639, + "output_text_mask_loss": 0.0469, + "predictor_learning_rate": 0.0003867772464832846, + "step": 3282 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.461, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.867077949403322e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0452, + "predictor_learning_rate": 0.00038670779494033217, + "step": 3283 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4609, + "image_masks_loss": 0.0515, + "llm_learning_rate": 3.8663832834281264e-06, + "loss": 0.793, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.00038663832834281265, + "step": 3284 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4608, + "image_masks_loss": 0.0359, + "llm_learning_rate": 3.865688466983762e-06, + "loss": 0.7721, + "output_text_mask_loss": 0.0683, + "predictor_learning_rate": 0.00038656884669837624, + "step": 3285 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4607, + "image_masks_loss": 0.0166, + "llm_learning_rate": 3.864993500146741e-06, + "loss": 0.8239, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003864993500146741, + "step": 3286 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4606, + "image_masks_loss": 0.014, + "llm_learning_rate": 3.864298382993596e-06, + "loss": 0.7678, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003864298382993595, + "step": 3287 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4605, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.8636031156008734e-06, + "loss": 0.6941, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00038636031156008733, + "step": 3288 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4604, + "image_masks_loss": 0.0043, + "llm_learning_rate": 3.862907698045139e-06, + "loss": 0.8346, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0003862907698045138, + "step": 3289 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4603, + "image_masks_loss": 0.0313, + "llm_learning_rate": 3.862212130402973e-06, + "loss": 0.7873, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003862212130402973, + "step": 3290 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4602, + "image_masks_loss": 0.0131, + "llm_learning_rate": 3.861516412750973e-06, + "loss": 0.8727, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 0.00038615164127509726, + "step": 3291 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4601, + "image_masks_loss": 0.0185, + "llm_learning_rate": 3.860820545165752e-06, + "loss": 0.706, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003860820545165752, + "step": 3292 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4599, + "image_masks_loss": 0.0353, + "llm_learning_rate": 3.860124527723943e-06, + "loss": 0.7918, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.00038601245277239426, + "step": 3293 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4598, + "image_masks_loss": 0.0242, + "llm_learning_rate": 3.859428360502191e-06, + "loss": 0.7552, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.0003859428360502191, + "step": 3294 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4597, + "image_masks_loss": 0.0173, + "llm_learning_rate": 3.858732043577159e-06, + "loss": 0.7958, + "output_text_mask_loss": 0.0372, + "predictor_learning_rate": 0.00038587320435771587, + "step": 3295 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4596, + "image_masks_loss": 0.013, + "llm_learning_rate": 3.858035577025529e-06, + "loss": 0.7937, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00038580355770255286, + "step": 3296 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4595, + "image_masks_loss": 0.0098, + "llm_learning_rate": 3.857338960923996e-06, + "loss": 0.731, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003857338960923996, + "step": 3297 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4594, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.8566421953492735e-06, + "loss": 0.8017, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003856642195349273, + "step": 3298 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4593, + "image_masks_loss": 0.0652, + "llm_learning_rate": 3.855945280378091e-06, + "loss": 0.7481, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.000385594528037809, + "step": 3299 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4592, + "image_masks_loss": 0.0089, + "llm_learning_rate": 3.855248216087193e-06, + "loss": 0.7917, + "output_text_mask_loss": 0.0183, + "predictor_learning_rate": 0.00038552482160871925, + "step": 3300 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4591, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.854551002553343e-06, + "loss": 0.7835, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00038545510025533424, + "step": 3301 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.459, + "image_masks_loss": 0.0488, + "llm_learning_rate": 3.853853639853319e-06, + "loss": 0.7654, + "output_text_mask_loss": 0.0366, + "predictor_learning_rate": 0.0003853853639853319, + "step": 3302 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4589, + "image_masks_loss": 0.0181, + "llm_learning_rate": 3.853156128063917e-06, + "loss": 0.7652, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00038531561280639174, + "step": 3303 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4588, + "image_masks_loss": 0.0185, + "llm_learning_rate": 3.8524584672619485e-06, + "loss": 0.6812, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00038524584672619484, + "step": 3304 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4586, + "image_masks_loss": 0.0456, + "llm_learning_rate": 3.8517606575242406e-06, + "loss": 0.7948, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00038517606575242405, + "step": 3305 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4585, + "image_masks_loss": 0.0127, + "llm_learning_rate": 3.8510626989276386e-06, + "loss": 0.7855, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038510626989276387, + "step": 3306 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4584, + "image_masks_loss": 0.0031, + "llm_learning_rate": 3.850364591549003e-06, + "loss": 0.7208, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.00038503645915490025, + "step": 3307 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4583, + "image_masks_loss": 0.025, + "llm_learning_rate": 3.84966633546521e-06, + "loss": 0.8093, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.000384966633546521, + "step": 3308 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4582, + "image_masks_loss": 0.0209, + "llm_learning_rate": 3.848967930753155e-06, + "loss": 0.8231, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00038489679307531546, + "step": 3309 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4581, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.848269377489746e-06, + "loss": 0.6976, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003848269377489746, + "step": 3310 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.458, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.84757067575191e-06, + "loss": 0.7009, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.000384757067575191, + "step": 3311 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4579, + "image_masks_loss": 0.0101, + "llm_learning_rate": 3.846871825616589e-06, + "loss": 0.7015, + "output_text_mask_loss": 0.0094, + "predictor_learning_rate": 0.0003846871825616589, + "step": 3312 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4578, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.8461728271607435e-06, + "loss": 0.7227, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003846172827160743, + "step": 3313 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4577, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.845473680461348e-06, + "loss": 0.7834, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038454736804613475, + "step": 3314 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4576, + "image_masks_loss": 0.012, + "llm_learning_rate": 3.844774385595394e-06, + "loss": 0.7766, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0003844774385595393, + "step": 3315 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4575, + "image_masks_loss": 0.0315, + "llm_learning_rate": 3.844074942639887e-06, + "loss": 0.776, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003844074942639887, + "step": 3316 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4573, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.843375351671855e-06, + "loss": 0.7158, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0003843375351671855, + "step": 3317 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4572, + "image_masks_loss": 0.0189, + "llm_learning_rate": 3.842675612768338e-06, + "loss": 0.8294, + "output_text_mask_loss": 0.042, + "predictor_learning_rate": 0.0003842675612768338, + "step": 3318 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4571, + "image_masks_loss": 0.0067, + "llm_learning_rate": 3.841975726006391e-06, + "loss": 0.724, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00038419757260063907, + "step": 3319 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.457, + "image_masks_loss": 0.0186, + "llm_learning_rate": 3.841275691463088e-06, + "loss": 0.7372, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 0.00038412756914630875, + "step": 3320 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4569, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.840575509215518e-06, + "loss": 0.6705, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003840575509215518, + "step": 3321 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4568, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.839875179340788e-06, + "loss": 0.8183, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00038398751793407873, + "step": 3322 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4567, + "image_masks_loss": 0.0189, + "llm_learning_rate": 3.839174701916017e-06, + "loss": 0.7273, + "output_text_mask_loss": 0.0694, + "predictor_learning_rate": 0.0003839174701916017, + "step": 3323 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4566, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.838474077018346e-06, + "loss": 0.7751, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00038384740770183457, + "step": 3324 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4565, + "image_masks_loss": 0.0288, + "llm_learning_rate": 3.8377733047249285e-06, + "loss": 0.6772, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003837773304724928, + "step": 3325 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4564, + "image_masks_loss": 0.0047, + "llm_learning_rate": 3.837072385112933e-06, + "loss": 0.7903, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.00038370723851129335, + "step": 3326 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4563, + "image_masks_loss": 0.0177, + "llm_learning_rate": 3.836371318259551e-06, + "loss": 0.6755, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003836371318259551, + "step": 3327 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4562, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.835670104241982e-06, + "loss": 0.7907, + "output_text_mask_loss": 0.0067, + "predictor_learning_rate": 0.0003835670104241981, + "step": 3328 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4561, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.834968743137445e-06, + "loss": 0.6656, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 0.0003834968743137445, + "step": 3329 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4559, + "image_masks_loss": 0.0067, + "llm_learning_rate": 3.834267235023177e-06, + "loss": 0.5927, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003834267235023177, + "step": 3330 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4558, + "image_masks_loss": 0.0057, + "llm_learning_rate": 3.83356557997643e-06, + "loss": 0.6995, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003833565579976429, + "step": 3331 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4557, + "image_masks_loss": 0.0135, + "llm_learning_rate": 3.832863778074469e-06, + "loss": 0.797, + "output_text_mask_loss": 0.1442, + "predictor_learning_rate": 0.0003832863778074469, + "step": 3332 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4556, + "image_masks_loss": 0.0107, + "llm_learning_rate": 3.832161829394582e-06, + "loss": 0.7807, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00038321618293945815, + "step": 3333 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4555, + "image_masks_loss": 0.0343, + "llm_learning_rate": 3.831459734014066e-06, + "loss": 0.7533, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.00038314597340140656, + "step": 3334 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4554, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.8307574920102374e-06, + "loss": 0.781, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.00038307574920102373, + "step": 3335 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4553, + "image_masks_loss": 0.023, + "llm_learning_rate": 3.830055103460431e-06, + "loss": 0.6965, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.00038300551034604316, + "step": 3336 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4552, + "image_masks_loss": 0.0144, + "llm_learning_rate": 3.8293525684419945e-06, + "loss": 0.8064, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038293525684419944, + "step": 3337 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4551, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.828649887032291e-06, + "loss": 0.7773, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00038286498870322906, + "step": 3338 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.455, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.827947059308703e-06, + "loss": 0.7168, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0003827947059308703, + "step": 3339 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4549, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.827244085348627e-06, + "loss": 0.6419, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.00038272440853486265, + "step": 3340 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4548, + "image_masks_loss": 0.0032, + "llm_learning_rate": 3.8265409652294764e-06, + "loss": 0.7544, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0003826540965229476, + "step": 3341 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4547, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.825837699028678e-06, + "loss": 0.7278, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0003825837699028678, + "step": 3342 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4546, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.825134286823681e-06, + "loss": 0.7308, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003825134286823681, + "step": 3343 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4544, + "image_masks_loss": 0.0282, + "llm_learning_rate": 3.8244307286919444e-06, + "loss": 0.8074, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00038244307286919445, + "step": 3344 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4543, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.8237270247109445e-06, + "loss": 0.736, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00038237270247109444, + "step": 3345 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4542, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.823023174958178e-06, + "loss": 0.7676, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003823023174958178, + "step": 3346 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4541, + "image_masks_loss": 0.007, + "llm_learning_rate": 3.822319179511152e-06, + "loss": 0.7289, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003822319179511152, + "step": 3347 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.454, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.821615038447392e-06, + "loss": 0.7838, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.00038216150384473926, + "step": 3348 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4539, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.820910751844441e-06, + "loss": 0.7862, + "output_text_mask_loss": 0.1015, + "predictor_learning_rate": 0.0003820910751844441, + "step": 3349 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4538, + "image_masks_loss": 0.0355, + "llm_learning_rate": 3.820206319779856e-06, + "loss": 0.6676, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.0003820206319779855, + "step": 3350 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4537, + "image_masks_loss": 0.0216, + "llm_learning_rate": 3.8195017423312096e-06, + "loss": 0.6863, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0003819501742331209, + "step": 3351 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4536, + "image_masks_loss": 0.0437, + "llm_learning_rate": 3.818797019576092e-06, + "loss": 0.8495, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003818797019576092, + "step": 3352 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4535, + "image_masks_loss": 0.028, + "llm_learning_rate": 3.81809215159211e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.00038180921515921097, + "step": 3353 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4534, + "image_masks_loss": 0.029, + "llm_learning_rate": 3.817387138456884e-06, + "loss": 0.8796, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00038173871384568835, + "step": 3354 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4533, + "image_masks_loss": 0.0319, + "llm_learning_rate": 3.8166819802480516e-06, + "loss": 0.7439, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00038166819802480515, + "step": 3355 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4532, + "image_masks_loss": 0.0282, + "llm_learning_rate": 3.815976677043267e-06, + "loss": 0.8158, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0003815976677043267, + "step": 3356 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4531, + "image_masks_loss": 0.0136, + "llm_learning_rate": 3.8152712289202e-06, + "loss": 0.7425, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00038152712289201996, + "step": 3357 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4529, + "image_masks_loss": 0.0036, + "llm_learning_rate": 3.8145656359565342e-06, + "loss": 0.7561, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003814565635956534, + "step": 3358 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4528, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.8138598982299733e-06, + "loss": 0.7989, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003813859898229973, + "step": 3359 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4527, + "image_masks_loss": 0.0128, + "llm_learning_rate": 3.8131540158182346e-06, + "loss": 0.631, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003813154015818234, + "step": 3360 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4526, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.8124479887990495e-06, + "loss": 0.7391, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003812447988799049, + "step": 3361 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4525, + "image_masks_loss": 0.0146, + "llm_learning_rate": 3.8117418172501697e-06, + "loss": 0.7413, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00038117418172501694, + "step": 3362 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4524, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.8110355012493584e-06, + "loss": 0.8635, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0003811035501249358, + "step": 3363 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4523, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.8103290408743977e-06, + "loss": 0.7988, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00038103290408743974, + "step": 3364 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4522, + "image_masks_loss": 0.015, + "llm_learning_rate": 3.8096224362030846e-06, + "loss": 0.7335, + "output_text_mask_loss": 0.0133, + "predictor_learning_rate": 0.00038096224362030843, + "step": 3365 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.4521, + "image_masks_loss": 0.0144, + "llm_learning_rate": 3.8089156873132314e-06, + "loss": 0.7394, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 0.0003808915687313231, + "step": 3366 + }, + { + "epoch": 0.34, + "gumbel_tau": 0.452, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.8082087942826673e-06, + "loss": 0.7147, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.00038082087942826673, + "step": 3367 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4519, + "image_masks_loss": 0.0185, + "llm_learning_rate": 3.8075017571892368e-06, + "loss": 0.8733, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.0003807501757189237, + "step": 3368 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4518, + "image_masks_loss": 0.053, + "llm_learning_rate": 3.806794576110801e-06, + "loss": 0.7806, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0003806794576110801, + "step": 3369 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4517, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.8060872511252354e-06, + "loss": 0.8325, + "output_text_mask_loss": 0.0265, + "predictor_learning_rate": 0.0003806087251125235, + "step": 3370 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4516, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.8053797823104322e-06, + "loss": 0.7661, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00038053797823104317, + "step": 3371 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4515, + "image_masks_loss": 0.0189, + "llm_learning_rate": 3.8046721697443e-06, + "loss": 0.6721, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00038046721697442996, + "step": 3372 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4513, + "image_masks_loss": 0.0138, + "llm_learning_rate": 3.8039644135047625e-06, + "loss": 0.7779, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00038039644135047626, + "step": 3373 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4512, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.8032565136697593e-06, + "loss": 0.7318, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003803256513669759, + "step": 3374 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4511, + "image_masks_loss": 0.0102, + "llm_learning_rate": 3.8025484703172455e-06, + "loss": 0.7275, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003802548470317245, + "step": 3375 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.451, + "image_masks_loss": 0.0058, + "llm_learning_rate": 3.8018402835251933e-06, + "loss": 0.7467, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0003801840283525193, + "step": 3376 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4509, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.8011319533715892e-06, + "loss": 0.7473, + "output_text_mask_loss": 0.0108, + "predictor_learning_rate": 0.0003801131953371589, + "step": 3377 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4508, + "image_masks_loss": 0.0127, + "llm_learning_rate": 3.8004234799344364e-06, + "loss": 0.8216, + "output_text_mask_loss": 0.0138, + "predictor_learning_rate": 0.0003800423479934436, + "step": 3378 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4507, + "image_masks_loss": 0.0145, + "llm_learning_rate": 3.7997148632917533e-06, + "loss": 0.6773, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00037997148632917533, + "step": 3379 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4506, + "image_masks_loss": 0.0064, + "llm_learning_rate": 3.7990061035215743e-06, + "loss": 0.7855, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003799006103521574, + "step": 3380 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4505, + "image_masks_loss": 0.014, + "llm_learning_rate": 3.798297200701949e-06, + "loss": 0.7232, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003798297200701949, + "step": 3381 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4504, + "image_masks_loss": 0.034, + "llm_learning_rate": 3.7975881549109457e-06, + "loss": 0.7506, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003797588154910945, + "step": 3382 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4503, + "image_masks_loss": 0.011, + "llm_learning_rate": 3.7968789662266437e-06, + "loss": 0.6723, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0003796878966226643, + "step": 3383 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4502, + "image_masks_loss": 0.0492, + "llm_learning_rate": 3.7961696347271417e-06, + "loss": 0.6875, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037961696347271415, + "step": 3384 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4501, + "image_masks_loss": 0.024, + "llm_learning_rate": 3.7954601604905517e-06, + "loss": 0.6209, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0003795460160490551, + "step": 3385 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.45, + "image_masks_loss": 0.0178, + "llm_learning_rate": 3.7947505435950037e-06, + "loss": 0.7783, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.00037947505435950035, + "step": 3386 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4499, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.7940407841186424e-06, + "loss": 0.7363, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003794040784118642, + "step": 3387 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4498, + "image_masks_loss": 0.0225, + "llm_learning_rate": 3.7933308821396264e-06, + "loss": 0.7398, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003793330882139626, + "step": 3388 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4496, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.792620837736134e-06, + "loss": 0.7633, + "output_text_mask_loss": 0.0317, + "predictor_learning_rate": 0.0003792620837736134, + "step": 3389 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4495, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.7919106509863546e-06, + "loss": 0.7689, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00037919106509863543, + "step": 3390 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4494, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.791200321968498e-06, + "loss": 0.7293, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003791200321968498, + "step": 3391 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4493, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.7904898507607857e-06, + "loss": 0.7252, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00037904898507607854, + "step": 3392 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4492, + "image_masks_loss": 0.0208, + "llm_learning_rate": 3.7897792374414554e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00037897792374414553, + "step": 3393 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4491, + "image_masks_loss": 0.0296, + "llm_learning_rate": 3.7890684820887636e-06, + "loss": 0.706, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0003789068482088763, + "step": 3394 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.449, + "image_masks_loss": 0.0309, + "llm_learning_rate": 3.7883575847809795e-06, + "loss": 0.7264, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00037883575847809793, + "step": 3395 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4489, + "image_masks_loss": 0.0587, + "llm_learning_rate": 3.787646545596387e-06, + "loss": 0.8194, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.0003787646545596387, + "step": 3396 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4488, + "image_masks_loss": 0.0454, + "llm_learning_rate": 3.78693536461329e-06, + "loss": 0.6897, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00037869353646132896, + "step": 3397 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4487, + "image_masks_loss": 0.0307, + "llm_learning_rate": 3.7862240419100036e-06, + "loss": 0.717, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037862240419100034, + "step": 3398 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4486, + "image_masks_loss": 0.0223, + "llm_learning_rate": 3.7855125775648615e-06, + "loss": 0.7192, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.00037855125775648616, + "step": 3399 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4485, + "image_masks_loss": 0.0345, + "llm_learning_rate": 3.784800971656211e-06, + "loss": 0.7897, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 0.00037848009716562103, + "step": 3400 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4484, + "image_masks_loss": 0.0338, + "llm_learning_rate": 3.784089224262415e-06, + "loss": 0.7871, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00037840892242624145, + "step": 3401 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4483, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.783377335461854e-06, + "loss": 0.7562, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003783377335461854, + "step": 3402 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4482, + "image_masks_loss": 0.0109, + "llm_learning_rate": 3.782665305332921e-06, + "loss": 0.7605, + "output_text_mask_loss": 0.0847, + "predictor_learning_rate": 0.0003782665305332921, + "step": 3403 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4481, + "image_masks_loss": 0.0089, + "llm_learning_rate": 3.781953133954029e-06, + "loss": 0.6621, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037819531339540284, + "step": 3404 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4479, + "image_masks_loss": 0.028, + "llm_learning_rate": 3.7812408214036015e-06, + "loss": 0.7953, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.00037812408214036013, + "step": 3405 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4478, + "image_masks_loss": 0.0063, + "llm_learning_rate": 3.7805283677600814e-06, + "loss": 0.723, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0003780528367760081, + "step": 3406 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4477, + "image_masks_loss": 0.0126, + "llm_learning_rate": 3.779815773101925e-06, + "loss": 0.7853, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.00037798157731019243, + "step": 3407 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4476, + "image_masks_loss": 0.0317, + "llm_learning_rate": 3.7791030375076055e-06, + "loss": 0.772, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003779103037507605, + "step": 3408 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4475, + "image_masks_loss": 0.0068, + "llm_learning_rate": 3.7783901610556097e-06, + "loss": 0.7334, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037783901610556093, + "step": 3409 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4474, + "image_masks_loss": 0.0043, + "llm_learning_rate": 3.7776771438244418e-06, + "loss": 0.7386, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003777677143824442, + "step": 3410 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4473, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.776963985892622e-06, + "loss": 0.7513, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003776963985892622, + "step": 3411 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4472, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.776250687338683e-06, + "loss": 0.8032, + "output_text_mask_loss": 0.0538, + "predictor_learning_rate": 0.00037762506873386824, + "step": 3412 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4471, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.775537248241175e-06, + "loss": 0.7453, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0003775537248241175, + "step": 3413 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.447, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.7748236686786645e-06, + "loss": 0.78, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00037748236686786644, + "step": 3414 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4469, + "image_masks_loss": 0.0216, + "llm_learning_rate": 3.774109948729733e-06, + "loss": 0.6831, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00037741099487297324, + "step": 3415 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4468, + "image_masks_loss": 0.0236, + "llm_learning_rate": 3.7733960884729742e-06, + "loss": 0.8147, + "output_text_mask_loss": 0.0774, + "predictor_learning_rate": 0.0003773396088472974, + "step": 3416 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4467, + "image_masks_loss": 0.0303, + "llm_learning_rate": 3.772682087987003e-06, + "loss": 0.7587, + "output_text_mask_loss": 0.0296, + "predictor_learning_rate": 0.0003772682087987003, + "step": 3417 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4466, + "image_masks_loss": 0.0217, + "llm_learning_rate": 3.771967947350445e-06, + "loss": 0.7679, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003771967947350445, + "step": 3418 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4465, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.7712536666419443e-06, + "loss": 0.7687, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0003771253666641944, + "step": 3419 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4464, + "image_masks_loss": 0.0288, + "llm_learning_rate": 3.7705392459401567e-06, + "loss": 0.7487, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0003770539245940157, + "step": 3420 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4463, + "image_masks_loss": 0.0153, + "llm_learning_rate": 3.7698246853237584e-06, + "loss": 0.7113, + "output_text_mask_loss": 0.0759, + "predictor_learning_rate": 0.0003769824685323758, + "step": 3421 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4462, + "image_masks_loss": 0.0271, + "llm_learning_rate": 3.7691099848714375e-06, + "loss": 0.7653, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0003769109984871437, + "step": 3422 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4461, + "image_masks_loss": 0.0203, + "llm_learning_rate": 3.7683951446618973e-06, + "loss": 0.7372, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003768395144661897, + "step": 3423 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4459, + "image_masks_loss": 0.0243, + "llm_learning_rate": 3.767680164773859e-06, + "loss": 0.7726, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 0.0003767680164773859, + "step": 3424 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4458, + "image_masks_loss": 0.032, + "llm_learning_rate": 3.7669650452860577e-06, + "loss": 0.789, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.0003766965045286058, + "step": 3425 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4457, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.7662497862772428e-06, + "loss": 0.8103, + "output_text_mask_loss": 0.0675, + "predictor_learning_rate": 0.00037662497862772426, + "step": 3426 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4456, + "image_masks_loss": 0.0471, + "llm_learning_rate": 3.765534387826182e-06, + "loss": 0.8728, + "output_text_mask_loss": 0.0725, + "predictor_learning_rate": 0.00037655343878261816, + "step": 3427 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4455, + "image_masks_loss": 0.0263, + "llm_learning_rate": 3.7648188500116547e-06, + "loss": 0.8281, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0003764818850011654, + "step": 3428 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4454, + "image_masks_loss": 0.0326, + "llm_learning_rate": 3.7641031729124588e-06, + "loss": 0.751, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0003764103172912459, + "step": 3429 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4453, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.7633873566074058e-06, + "loss": 0.7296, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.00037633873566074054, + "step": 3430 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4452, + "image_masks_loss": 0.0236, + "llm_learning_rate": 3.762671401175323e-06, + "loss": 0.7397, + "output_text_mask_loss": 0.1938, + "predictor_learning_rate": 0.0003762671401175323, + "step": 3431 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4451, + "image_masks_loss": 0.0362, + "llm_learning_rate": 3.761955306695052e-06, + "loss": 0.7179, + "output_text_mask_loss": 0.0507, + "predictor_learning_rate": 0.0003761955306695052, + "step": 3432 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.445, + "image_masks_loss": 0.0159, + "llm_learning_rate": 3.761239073245453e-06, + "loss": 0.798, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00037612390732454527, + "step": 3433 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4449, + "image_masks_loss": 0.0603, + "llm_learning_rate": 3.7605227009053978e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037605227009053976, + "step": 3434 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4448, + "image_masks_loss": 0.0229, + "llm_learning_rate": 3.7598061897537753e-06, + "loss": 0.7539, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003759806189753775, + "step": 3435 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4447, + "image_masks_loss": 0.0063, + "llm_learning_rate": 3.7590895398694878e-06, + "loss": 0.836, + "output_text_mask_loss": 0.0626, + "predictor_learning_rate": 0.00037590895398694873, + "step": 3436 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4446, + "image_masks_loss": 0.0307, + "llm_learning_rate": 3.758372751331456e-06, + "loss": 0.7558, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.0003758372751331456, + "step": 3437 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4445, + "image_masks_loss": 0.038, + "llm_learning_rate": 3.7576558242186157e-06, + "loss": 0.9015, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0003757655824218615, + "step": 3438 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4444, + "image_masks_loss": 0.0206, + "llm_learning_rate": 3.7569387586099126e-06, + "loss": 0.786, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00037569387586099123, + "step": 3439 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4443, + "image_masks_loss": 0.0217, + "llm_learning_rate": 3.756221554584315e-06, + "loss": 0.6784, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003756221554584315, + "step": 3440 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4442, + "image_masks_loss": 0.0296, + "llm_learning_rate": 3.7555042122208006e-06, + "loss": 0.9097, + "output_text_mask_loss": 0.0166, + "predictor_learning_rate": 0.00037555042122208005, + "step": 3441 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4441, + "image_masks_loss": 0.0706, + "llm_learning_rate": 3.754786731598367e-06, + "loss": 0.863, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00037547867315983663, + "step": 3442 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.444, + "image_masks_loss": 0.0456, + "llm_learning_rate": 3.754069112796023e-06, + "loss": 0.7639, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.00037540691127960224, + "step": 3443 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4438, + "image_masks_loss": 0.0102, + "llm_learning_rate": 3.753351355892794e-06, + "loss": 0.7978, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0003753351355892794, + "step": 3444 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4437, + "image_masks_loss": 0.0345, + "llm_learning_rate": 3.7526334609677234e-06, + "loss": 0.8692, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.0003752633460967723, + "step": 3445 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4436, + "image_masks_loss": 0.0671, + "llm_learning_rate": 3.751915428099865e-06, + "loss": 0.7588, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037519154280998645, + "step": 3446 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4435, + "image_masks_loss": 0.0071, + "llm_learning_rate": 3.751197257368292e-06, + "loss": 0.6977, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003751197257368292, + "step": 3447 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4434, + "image_masks_loss": 0.0359, + "llm_learning_rate": 3.750478948852089e-06, + "loss": 0.8869, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 0.0003750478948852089, + "step": 3448 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4433, + "image_masks_loss": 0.0219, + "llm_learning_rate": 3.7497605026303595e-06, + "loss": 0.8176, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00037497605026303593, + "step": 3449 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4432, + "image_masks_loss": 0.0393, + "llm_learning_rate": 3.7490419187822204e-06, + "loss": 0.6869, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000374904191878222, + "step": 3450 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4431, + "image_masks_loss": 0.0385, + "llm_learning_rate": 3.748323197386802e-06, + "loss": 0.7762, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003748323197386802, + "step": 3451 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.443, + "image_masks_loss": 0.0243, + "llm_learning_rate": 3.747604338523253e-06, + "loss": 0.7595, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0003747604338523253, + "step": 3452 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4429, + "image_masks_loss": 0.0357, + "llm_learning_rate": 3.746885342270735e-06, + "loss": 0.7858, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003746885342270735, + "step": 3453 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4428, + "image_masks_loss": 0.0084, + "llm_learning_rate": 3.7461662087084262e-06, + "loss": 0.7072, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00037461662087084256, + "step": 3454 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4427, + "image_masks_loss": 0.0299, + "llm_learning_rate": 3.745446937915518e-06, + "loss": 0.7415, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0003745446937915518, + "step": 3455 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4426, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.74472752997122e-06, + "loss": 0.7224, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037447275299712195, + "step": 3456 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4425, + "image_masks_loss": 0.0157, + "llm_learning_rate": 3.7440079849547537e-06, + "loss": 0.8022, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.00037440079849547535, + "step": 3457 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4424, + "image_masks_loss": 0.0086, + "llm_learning_rate": 3.7432883029453565e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037432883029453565, + "step": 3458 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4423, + "image_masks_loss": 0.0349, + "llm_learning_rate": 3.742568484022282e-06, + "loss": 0.8211, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.0003742568484022282, + "step": 3459 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4422, + "image_masks_loss": 0.0173, + "llm_learning_rate": 3.741848528264799e-06, + "loss": 0.6807, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.0003741848528264799, + "step": 3460 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4421, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.74112843575219e-06, + "loss": 0.7298, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037411284357521897, + "step": 3461 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.442, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.7404082065637526e-06, + "loss": 0.6935, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00037404082065637523, + "step": 3462 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4419, + "image_masks_loss": 0.028, + "llm_learning_rate": 3.7396878407788005e-06, + "loss": 0.7209, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.00037396878407788006, + "step": 3463 + }, + { + "epoch": 0.35, + "gumbel_tau": 0.4418, + "image_masks_loss": 0.0307, + "llm_learning_rate": 3.7389673384766623e-06, + "loss": 0.7195, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.0003738967338476662, + "step": 3464 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4417, + "image_masks_loss": 0.0345, + "llm_learning_rate": 3.7382466997366812e-06, + "loss": 0.7465, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0003738246699736681, + "step": 3465 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4415, + "image_masks_loss": 0.0045, + "llm_learning_rate": 3.7375259246382154e-06, + "loss": 0.703, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00037375259246382153, + "step": 3466 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4414, + "image_masks_loss": 0.0074, + "llm_learning_rate": 3.7368050132606386e-06, + "loss": 0.7889, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0003736805013260639, + "step": 3467 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4413, + "image_masks_loss": 0.0225, + "llm_learning_rate": 3.7360839656833386e-06, + "loss": 0.7735, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003736083965683338, + "step": 3468 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4412, + "image_masks_loss": 0.0139, + "llm_learning_rate": 3.735362781985719e-06, + "loss": 0.6541, + "output_text_mask_loss": 0.0629, + "predictor_learning_rate": 0.00037353627819857187, + "step": 3469 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4411, + "image_masks_loss": 0.0406, + "llm_learning_rate": 3.734641462247198e-06, + "loss": 0.8288, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0003734641462247198, + "step": 3470 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.441, + "image_masks_loss": 0.0169, + "llm_learning_rate": 3.733920006547209e-06, + "loss": 0.6493, + "output_text_mask_loss": 0.0215, + "predictor_learning_rate": 0.0003733920006547209, + "step": 3471 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4409, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.7331984149652013e-06, + "loss": 0.7241, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003733198414965201, + "step": 3472 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4408, + "image_masks_loss": 0.0096, + "llm_learning_rate": 3.732476687580637e-06, + "loss": 0.8619, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.00037324766875806363, + "step": 3473 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4407, + "image_masks_loss": 0.0205, + "llm_learning_rate": 3.731754824472994e-06, + "loss": 0.7626, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003731754824472994, + "step": 3474 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4406, + "image_masks_loss": 0.0158, + "llm_learning_rate": 3.7310328257217664e-06, + "loss": 0.8191, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0003731032825721766, + "step": 3475 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4405, + "image_masks_loss": 0.0224, + "llm_learning_rate": 3.7303106914064623e-06, + "loss": 0.7081, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00037303106914064624, + "step": 3476 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4404, + "image_masks_loss": 0.0278, + "llm_learning_rate": 3.729588421606605e-06, + "loss": 0.8143, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00037295884216066044, + "step": 3477 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4403, + "image_masks_loss": 0.0179, + "llm_learning_rate": 3.7288660164017303e-06, + "loss": 0.7694, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.000372886601640173, + "step": 3478 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4402, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.7281434758713934e-06, + "loss": 0.7509, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.0003728143475871393, + "step": 3479 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4401, + "image_masks_loss": 0.0041, + "llm_learning_rate": 3.7274208000951618e-06, + "loss": 0.786, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003727420800095162, + "step": 3480 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.44, + "image_masks_loss": 0.029, + "llm_learning_rate": 3.7266979891526168e-06, + "loss": 0.7319, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00037266979891526164, + "step": 3481 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4399, + "image_masks_loss": 0.038, + "llm_learning_rate": 3.725975043123358e-06, + "loss": 0.7664, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0003725975043123357, + "step": 3482 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4398, + "image_masks_loss": 0.0161, + "llm_learning_rate": 3.725251962086996e-06, + "loss": 0.818, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003725251962086996, + "step": 3483 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4397, + "image_masks_loss": 0.0129, + "llm_learning_rate": 3.7245287461231594e-06, + "loss": 0.7627, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003724528746123159, + "step": 3484 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4396, + "image_masks_loss": 0.0422, + "llm_learning_rate": 3.7238053953114896e-06, + "loss": 0.6725, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037238053953114893, + "step": 3485 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4395, + "image_masks_loss": 0.0222, + "llm_learning_rate": 3.723081909731643e-06, + "loss": 0.7983, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 0.00037230819097316427, + "step": 3486 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4394, + "image_masks_loss": 0.0393, + "llm_learning_rate": 3.7223582894632926e-06, + "loss": 0.7997, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 0.0003722358289463292, + "step": 3487 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4393, + "image_masks_loss": 0.0166, + "llm_learning_rate": 3.721634534586125e-06, + "loss": 0.7187, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00037216345345861246, + "step": 3488 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4392, + "image_masks_loss": 0.0138, + "llm_learning_rate": 3.7209106451798405e-06, + "loss": 0.7341, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.000372091064517984, + "step": 3489 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4391, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.720186621324157e-06, + "loss": 0.7852, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.0003720186621324157, + "step": 3490 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.439, + "image_masks_loss": 0.0119, + "llm_learning_rate": 3.7194624630988047e-06, + "loss": 0.7658, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00037194624630988043, + "step": 3491 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4388, + "image_masks_loss": 0.0223, + "llm_learning_rate": 3.71873817058353e-06, + "loss": 0.7644, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.000371873817058353, + "step": 3492 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4387, + "image_masks_loss": 0.0221, + "llm_learning_rate": 3.7180137438580933e-06, + "loss": 0.8106, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003718013743858093, + "step": 3493 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4386, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.7172891830022694e-06, + "loss": 0.8127, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0003717289183002269, + "step": 3494 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4385, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.7165644880958503e-06, + "loss": 0.8013, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000371656448809585, + "step": 3495 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4384, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.7158396592186396e-06, + "loss": 0.7612, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00037158396592186396, + "step": 3496 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4383, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.715114696450457e-06, + "loss": 0.7502, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037151146964504566, + "step": 3497 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4382, + "image_masks_loss": 0.0282, + "llm_learning_rate": 3.714389599871138e-06, + "loss": 0.7861, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00037143895998711376, + "step": 3498 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4381, + "image_masks_loss": 0.0111, + "llm_learning_rate": 3.713664369560531e-06, + "loss": 0.7706, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 0.0003713664369560531, + "step": 3499 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.438, + "image_masks_loss": 0.0254, + "llm_learning_rate": 3.7129390055985003e-06, + "loss": 0.8624, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00037129390055985, + "step": 3500 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4379, + "image_masks_loss": 0.0412, + "llm_learning_rate": 3.7122135080649247e-06, + "loss": 0.8156, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00037122135080649245, + "step": 3501 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4378, + "image_masks_loss": 0.0156, + "llm_learning_rate": 3.7114878770396974e-06, + "loss": 0.8424, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0003711487877039697, + "step": 3502 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4377, + "image_masks_loss": 0.0167, + "llm_learning_rate": 3.7107621126027265e-06, + "loss": 0.7062, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00037107621126027265, + "step": 3503 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4376, + "image_masks_loss": 0.0172, + "llm_learning_rate": 3.710036214833935e-06, + "loss": 0.7605, + "output_text_mask_loss": 0.0801, + "predictor_learning_rate": 0.0003710036214833935, + "step": 3504 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4375, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.7093101838132605e-06, + "loss": 0.7771, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.00037093101838132605, + "step": 3505 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4374, + "image_masks_loss": 0.0284, + "llm_learning_rate": 3.708584019620655e-06, + "loss": 0.8078, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00037085840196206547, + "step": 3506 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4373, + "image_masks_loss": 0.0107, + "llm_learning_rate": 3.7078577223360847e-06, + "loss": 0.7358, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.00037078577223360843, + "step": 3507 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4372, + "image_masks_loss": 0.0357, + "llm_learning_rate": 3.707131292039532e-06, + "loss": 0.7829, + "output_text_mask_loss": 0.1495, + "predictor_learning_rate": 0.00037071312920395315, + "step": 3508 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4371, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.706404728810993e-06, + "loss": 0.7564, + "output_text_mask_loss": 0.1045, + "predictor_learning_rate": 0.00037064047288109923, + "step": 3509 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.437, + "image_masks_loss": 0.0238, + "llm_learning_rate": 3.705678032730477e-06, + "loss": 0.8268, + "output_text_mask_loss": 0.0595, + "predictor_learning_rate": 0.00037056780327304767, + "step": 3510 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4369, + "image_masks_loss": 0.028, + "llm_learning_rate": 3.7049512038780113e-06, + "loss": 0.8372, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003704951203878011, + "step": 3511 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4368, + "image_masks_loss": 0.0074, + "llm_learning_rate": 3.7042242423336354e-06, + "loss": 0.8696, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003704224242333635, + "step": 3512 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4367, + "image_masks_loss": 0.0111, + "llm_learning_rate": 3.703497148177403e-06, + "loss": 0.7884, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003703497148177403, + "step": 3513 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4366, + "image_masks_loss": 0.0091, + "llm_learning_rate": 3.7027699214893843e-06, + "loss": 0.8171, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003702769921489384, + "step": 3514 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4365, + "image_masks_loss": 0.0166, + "llm_learning_rate": 3.7020425623496627e-06, + "loss": 0.7217, + "output_text_mask_loss": 0.2075, + "predictor_learning_rate": 0.00037020425623496625, + "step": 3515 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4364, + "image_masks_loss": 0.0282, + "llm_learning_rate": 3.7013150708383367e-06, + "loss": 0.9118, + "output_text_mask_loss": 0.0694, + "predictor_learning_rate": 0.00037013150708383363, + "step": 3516 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4363, + "image_masks_loss": 0.0086, + "llm_learning_rate": 3.7005874470355187e-06, + "loss": 0.8786, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00037005874470355185, + "step": 3517 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4362, + "image_masks_loss": 0.0236, + "llm_learning_rate": 3.6998596910213375e-06, + "loss": 0.7746, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00036998596910213375, + "step": 3518 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4361, + "image_masks_loss": 0.0183, + "llm_learning_rate": 3.6991318028759344e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00036991318028759343, + "step": 3519 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.436, + "image_masks_loss": 0.0138, + "llm_learning_rate": 3.6984037826794657e-06, + "loss": 0.7691, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00036984037826794656, + "step": 3520 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4359, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.6976756305121035e-06, + "loss": 0.8158, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.0003697675630512103, + "step": 3521 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4358, + "image_masks_loss": 0.0036, + "llm_learning_rate": 3.696947346454034e-06, + "loss": 0.8446, + "output_text_mask_loss": 0.1495, + "predictor_learning_rate": 0.0003696947346454034, + "step": 3522 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4357, + "image_masks_loss": 0.028, + "llm_learning_rate": 3.6962189305854558e-06, + "loss": 0.723, + "output_text_mask_loss": 0.0984, + "predictor_learning_rate": 0.00036962189305854556, + "step": 3523 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4355, + "image_masks_loss": 0.0265, + "llm_learning_rate": 3.695490382986584e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0003695490382986584, + "step": 3524 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4354, + "image_masks_loss": 0.0086, + "llm_learning_rate": 3.6947617037376495e-06, + "loss": 0.7448, + "output_text_mask_loss": 0.1076, + "predictor_learning_rate": 0.0003694761703737649, + "step": 3525 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4353, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.694032892918894e-06, + "loss": 0.8113, + "output_text_mask_loss": 0.0969, + "predictor_learning_rate": 0.00036940328929188935, + "step": 3526 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4352, + "image_masks_loss": 0.0233, + "llm_learning_rate": 3.6933039506105773e-06, + "loss": 0.7139, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003693303950610577, + "step": 3527 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4351, + "image_masks_loss": 0.0161, + "llm_learning_rate": 3.692574876892971e-06, + "loss": 0.8413, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.00036925748768929705, + "step": 3528 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.435, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.6918456718463635e-06, + "loss": 0.7698, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00036918456718463633, + "step": 3529 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4349, + "image_masks_loss": 0.0309, + "llm_learning_rate": 3.691116335551055e-06, + "loss": 0.7088, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00036911163355510544, + "step": 3530 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4348, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.690386868087364e-06, + "loss": 0.6891, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0003690386868087364, + "step": 3531 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4347, + "image_masks_loss": 0.0305, + "llm_learning_rate": 3.6896572695356186e-06, + "loss": 0.6969, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0003689657269535618, + "step": 3532 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4346, + "image_masks_loss": 0.0096, + "llm_learning_rate": 3.688927539976165e-06, + "loss": 0.722, + "output_text_mask_loss": 0.0725, + "predictor_learning_rate": 0.00036889275399761646, + "step": 3533 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4345, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.6881976794893625e-06, + "loss": 0.838, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00036881976794893626, + "step": 3534 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4344, + "image_masks_loss": 0.0217, + "llm_learning_rate": 3.687467688155585e-06, + "loss": 0.7262, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00036874676881555845, + "step": 3535 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4343, + "image_masks_loss": 0.0296, + "llm_learning_rate": 3.68673756605522e-06, + "loss": 0.7934, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00036867375660552197, + "step": 3536 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4342, + "image_masks_loss": 0.0298, + "llm_learning_rate": 3.6860073132686718e-06, + "loss": 0.7547, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003686007313268672, + "step": 3537 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4341, + "image_masks_loss": 0.0486, + "llm_learning_rate": 3.685276929876356e-06, + "loss": 0.7239, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0003685276929876356, + "step": 3538 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.434, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.6845464159587053e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003684546415958705, + "step": 3539 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4339, + "image_masks_loss": 0.0224, + "llm_learning_rate": 3.683815771596164e-06, + "loss": 0.8519, + "output_text_mask_loss": 0.0267, + "predictor_learning_rate": 0.0003683815771596164, + "step": 3540 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4338, + "image_masks_loss": 0.0326, + "llm_learning_rate": 3.683084996869194e-06, + "loss": 0.788, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00036830849968691936, + "step": 3541 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4337, + "image_masks_loss": 0.0197, + "llm_learning_rate": 3.6823540918582677e-06, + "loss": 0.7969, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0003682354091858268, + "step": 3542 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4336, + "image_masks_loss": 0.011, + "llm_learning_rate": 3.681623056643876e-06, + "loss": 0.7283, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003681623056643876, + "step": 3543 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4335, + "image_masks_loss": 0.0334, + "llm_learning_rate": 3.680891891306521e-06, + "loss": 0.7915, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003680891891306521, + "step": 3544 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4334, + "image_masks_loss": 0.022, + "llm_learning_rate": 3.6801605959267208e-06, + "loss": 0.6564, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00036801605959267203, + "step": 3545 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4333, + "image_masks_loss": 0.0198, + "llm_learning_rate": 3.6794291705850067e-06, + "loss": 0.816, + "output_text_mask_loss": 0.0441, + "predictor_learning_rate": 0.00036794291705850065, + "step": 3546 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4332, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.6786976153619257e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00036786976153619256, + "step": 3547 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4331, + "image_masks_loss": 0.032, + "llm_learning_rate": 3.6779659303380378e-06, + "loss": 0.8178, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00036779659303380375, + "step": 3548 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.433, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.677234115593918e-06, + "loss": 0.7781, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00036772341155939174, + "step": 3549 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4329, + "image_masks_loss": 0.0182, + "llm_learning_rate": 3.676502171210155e-06, + "loss": 0.7195, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.00036765021712101547, + "step": 3550 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4328, + "image_masks_loss": 0.0135, + "llm_learning_rate": 3.6757700972673523e-06, + "loss": 0.813, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 0.0003675770097267352, + "step": 3551 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4327, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.675037893846127e-06, + "loss": 0.8345, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003675037893846127, + "step": 3552 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4326, + "image_masks_loss": 0.0212, + "llm_learning_rate": 3.674305561027112e-06, + "loss": 0.8039, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0003674305561027112, + "step": 3553 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4325, + "image_masks_loss": 0.0169, + "llm_learning_rate": 3.6735730988909536e-06, + "loss": 0.7646, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.0003673573098890953, + "step": 3554 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4324, + "image_masks_loss": 0.004, + "llm_learning_rate": 3.672840507518311e-06, + "loss": 0.8061, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00036728405075183107, + "step": 3555 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4323, + "image_masks_loss": 0.015, + "llm_learning_rate": 3.6721077869898593e-06, + "loss": 0.6927, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003672107786989859, + "step": 3556 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4322, + "image_masks_loss": 0.0052, + "llm_learning_rate": 3.6713749373862874e-06, + "loss": 0.7816, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0003671374937386287, + "step": 3557 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4321, + "image_masks_loss": 0.0232, + "llm_learning_rate": 3.670641958788298e-06, + "loss": 0.789, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00036706419587882977, + "step": 3558 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.432, + "image_masks_loss": 0.015, + "llm_learning_rate": 3.669908851276609e-06, + "loss": 0.7426, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.00036699088512766087, + "step": 3559 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4319, + "image_masks_loss": 0.0064, + "llm_learning_rate": 3.669175614931951e-06, + "loss": 0.8531, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003669175614931951, + "step": 3560 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4318, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.6684422498350705e-06, + "loss": 0.7627, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.000366844224983507, + "step": 3561 + }, + { + "epoch": 0.36, + "gumbel_tau": 0.4317, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.6677087560667263e-06, + "loss": 0.8378, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0003667708756066726, + "step": 3562 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4316, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.666975133707693e-06, + "loss": 0.7167, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00036669751337076926, + "step": 3563 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4315, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.6662413828387593e-06, + "loss": 0.7151, + "output_text_mask_loss": 0.0477, + "predictor_learning_rate": 0.0003666241382838759, + "step": 3564 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4314, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.665507503540726e-06, + "loss": 0.7982, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0003665507503540726, + "step": 3565 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4313, + "image_masks_loss": 0.0099, + "llm_learning_rate": 3.664773495894411e-06, + "loss": 0.7494, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003664773495894411, + "step": 3566 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4312, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.664039359980644e-06, + "loss": 0.726, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003664039359980644, + "step": 3567 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4311, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.6633050958802698e-06, + "loss": 0.6774, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00036633050958802694, + "step": 3568 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4309, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.6625707036741476e-06, + "loss": 0.7485, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.00036625707036741474, + "step": 3569 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4308, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.6618361834431493e-06, + "loss": 0.7625, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.0003661836183443149, + "step": 3570 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4307, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.661101535268163e-06, + "loss": 0.7212, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0003661101535268163, + "step": 3571 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4306, + "image_masks_loss": 0.0052, + "llm_learning_rate": 3.660366759230089e-06, + "loss": 0.6156, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003660366759230089, + "step": 3572 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4305, + "image_masks_loss": 0.0147, + "llm_learning_rate": 3.6596318554098438e-06, + "loss": 0.7531, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00036596318554098434, + "step": 3573 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4304, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.658896823888355e-06, + "loss": 0.8038, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0003658896823888355, + "step": 3574 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4303, + "image_masks_loss": 0.0334, + "llm_learning_rate": 3.658161664746566e-06, + "loss": 0.8304, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0003658161664746566, + "step": 3575 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4302, + "image_masks_loss": 0.028, + "llm_learning_rate": 3.6574263780654355e-06, + "loss": 0.7983, + "output_text_mask_loss": 0.0523, + "predictor_learning_rate": 0.00036574263780654354, + "step": 3576 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4301, + "image_masks_loss": 0.0169, + "llm_learning_rate": 3.6566909639259353e-06, + "loss": 0.7824, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003656690963925935, + "step": 3577 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.43, + "image_masks_loss": 0.0298, + "llm_learning_rate": 3.6559554224090476e-06, + "loss": 0.7863, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00036559554224090476, + "step": 3578 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4299, + "image_masks_loss": 0.0125, + "llm_learning_rate": 3.6552197535957756e-06, + "loss": 0.8078, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.00036552197535957755, + "step": 3579 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4298, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.6544839575671305e-06, + "loss": 0.7677, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00036544839575671303, + "step": 3580 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4297, + "image_masks_loss": 0.0125, + "llm_learning_rate": 3.653748034404141e-06, + "loss": 0.7014, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0003653748034404141, + "step": 3581 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4296, + "image_masks_loss": 0.0051, + "llm_learning_rate": 3.6530119841878468e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00036530119841878466, + "step": 3582 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4295, + "image_masks_loss": 0.0145, + "llm_learning_rate": 3.652275806999306e-06, + "loss": 0.7663, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00036522758069993056, + "step": 3583 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4294, + "image_masks_loss": 0.01, + "llm_learning_rate": 3.651539502919586e-06, + "loss": 0.8425, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0003651539502919586, + "step": 3584 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4293, + "image_masks_loss": 0.0232, + "llm_learning_rate": 3.650803072029771e-06, + "loss": 0.7181, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.0003650803072029771, + "step": 3585 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4292, + "image_masks_loss": 0.0077, + "llm_learning_rate": 3.6500665144109587e-06, + "loss": 0.542, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00036500665144109586, + "step": 3586 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4291, + "image_masks_loss": 0.0109, + "llm_learning_rate": 3.6493298301442602e-06, + "loss": 0.7258, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00036493298301442604, + "step": 3587 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.429, + "image_masks_loss": 0.0178, + "llm_learning_rate": 3.648593019310801e-06, + "loss": 0.7035, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.00036485930193108006, + "step": 3588 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4289, + "image_masks_loss": 0.011, + "llm_learning_rate": 3.64785608199172e-06, + "loss": 0.7848, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00036478560819917197, + "step": 3589 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4288, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.6471190182681697e-06, + "loss": 0.7605, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.000364711901826817, + "step": 3590 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4287, + "image_masks_loss": 0.0311, + "llm_learning_rate": 3.646381828221318e-06, + "loss": 0.7462, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003646381828221318, + "step": 3591 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4286, + "image_masks_loss": 0.0319, + "llm_learning_rate": 3.645644511932347e-06, + "loss": 0.7431, + "output_text_mask_loss": 0.1129, + "predictor_learning_rate": 0.0003645644511932347, + "step": 3592 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4285, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.6449070694824502e-06, + "loss": 0.7446, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.000364490706948245, + "step": 3593 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4284, + "image_masks_loss": 0.0183, + "llm_learning_rate": 3.6441695009528368e-06, + "loss": 0.7969, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00036441695009528365, + "step": 3594 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4283, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.6434318064247287e-06, + "loss": 0.7665, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00036434318064247287, + "step": 3595 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4282, + "image_masks_loss": 0.0381, + "llm_learning_rate": 3.642693985979364e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003642693985979364, + "step": 3596 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4281, + "image_masks_loss": 0.007, + "llm_learning_rate": 3.6419560396979923e-06, + "loss": 0.8152, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00036419560396979924, + "step": 3597 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.428, + "image_masks_loss": 0.0231, + "llm_learning_rate": 3.641217967661877e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00036412179676618766, + "step": 3598 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4279, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.6404797699522974e-06, + "loss": 0.7493, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0003640479769952297, + "step": 3599 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4278, + "image_masks_loss": 0.0207, + "llm_learning_rate": 3.639741446650546e-06, + "loss": 0.778, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00036397414466505455, + "step": 3600 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4277, + "image_masks_loss": 0.0167, + "llm_learning_rate": 3.639002997837927e-06, + "loss": 0.8163, + "output_text_mask_loss": 0.0458, + "predictor_learning_rate": 0.0003639002997837927, + "step": 3601 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4276, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.638264423595761e-06, + "loss": 0.7428, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003638264423595761, + "step": 3602 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4275, + "image_masks_loss": 0.0137, + "llm_learning_rate": 3.6375257240053817e-06, + "loss": 0.8582, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00036375257240053816, + "step": 3603 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4274, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.636786899148136e-06, + "loss": 0.6419, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003636786899148136, + "step": 3604 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4273, + "image_masks_loss": 0.0111, + "llm_learning_rate": 3.6360479491053846e-06, + "loss": 0.607, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00036360479491053845, + "step": 3605 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4272, + "image_masks_loss": 0.0083, + "llm_learning_rate": 3.635308873958503e-06, + "loss": 0.7534, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00036353088739585025, + "step": 3606 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4271, + "image_masks_loss": 0.0273, + "llm_learning_rate": 3.634569673788878e-06, + "loss": 0.7915, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.0003634569673788878, + "step": 3607 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.427, + "image_masks_loss": 0.0402, + "llm_learning_rate": 3.6338303486779146e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00036338303486779143, + "step": 3608 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4269, + "image_masks_loss": 0.0172, + "llm_learning_rate": 3.633090898707028e-06, + "loss": 0.7275, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00036330908987070276, + "step": 3609 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4268, + "image_masks_loss": 0.0137, + "llm_learning_rate": 3.632351323957647e-06, + "loss": 0.6722, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00036323513239576465, + "step": 3610 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4267, + "image_masks_loss": 0.0334, + "llm_learning_rate": 3.631611624511215e-06, + "loss": 0.7263, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00036316116245112154, + "step": 3611 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4266, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.630871800449192e-06, + "loss": 0.7842, + "output_text_mask_loss": 0.0931, + "predictor_learning_rate": 0.0003630871800449192, + "step": 3612 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4265, + "image_masks_loss": 0.0026, + "llm_learning_rate": 3.6301318518530475e-06, + "loss": 0.8012, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.00036301318518530473, + "step": 3613 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4264, + "image_masks_loss": 0.0111, + "llm_learning_rate": 3.6293917788042654e-06, + "loss": 0.7594, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00036293917788042654, + "step": 3614 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4263, + "image_masks_loss": 0.0182, + "llm_learning_rate": 3.628651581384346e-06, + "loss": 0.8917, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.00036286515813843456, + "step": 3615 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4262, + "image_masks_loss": 0.0105, + "llm_learning_rate": 3.6279112596747997e-06, + "loss": 0.7585, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00036279112596747994, + "step": 3616 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4261, + "image_masks_loss": 0.0087, + "llm_learning_rate": 3.6271708137571527e-06, + "loss": 0.8208, + "output_text_mask_loss": 0.0366, + "predictor_learning_rate": 0.0003627170813757152, + "step": 3617 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.426, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.6264302437129458e-06, + "loss": 0.701, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.00036264302437129454, + "step": 3618 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4259, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.625689549623731e-06, + "loss": 0.6756, + "output_text_mask_loss": 0.0127, + "predictor_learning_rate": 0.0003625689549623731, + "step": 3619 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4258, + "image_masks_loss": 0.0128, + "llm_learning_rate": 3.6249487315710756e-06, + "loss": 0.7005, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00036249487315710753, + "step": 3620 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4257, + "image_masks_loss": 0.0229, + "llm_learning_rate": 3.6242077896365602e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0393, + "predictor_learning_rate": 0.000362420778963656, + "step": 3621 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4256, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.623466723901779e-06, + "loss": 0.6424, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00036234667239017787, + "step": 3622 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4255, + "image_masks_loss": 0.0153, + "llm_learning_rate": 3.62272553444834e-06, + "loss": 0.757, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00036227255344483403, + "step": 3623 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4254, + "image_masks_loss": 0.0243, + "llm_learning_rate": 3.6219842213578643e-06, + "loss": 0.7458, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0003621984221357864, + "step": 3624 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4253, + "image_masks_loss": 0.0102, + "llm_learning_rate": 3.621242784711987e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.00036212427847119865, + "step": 3625 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4252, + "image_masks_loss": 0.0174, + "llm_learning_rate": 3.6205012245923565e-06, + "loss": 0.7322, + "output_text_mask_loss": 0.0349, + "predictor_learning_rate": 0.0003620501224592356, + "step": 3626 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4251, + "image_masks_loss": 0.0326, + "llm_learning_rate": 3.619759541080635e-06, + "loss": 0.7625, + "output_text_mask_loss": 0.0599, + "predictor_learning_rate": 0.0003619759541080635, + "step": 3627 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.425, + "image_masks_loss": 0.0259, + "llm_learning_rate": 3.6190177342585e-06, + "loss": 0.7384, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00036190177342585, + "step": 3628 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4249, + "image_masks_loss": 0.033, + "llm_learning_rate": 3.6182758042076383e-06, + "loss": 0.7286, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003618275804207638, + "step": 3629 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4248, + "image_masks_loss": 0.0137, + "llm_learning_rate": 3.6175337510097542e-06, + "loss": 0.7492, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.0003617533751009754, + "step": 3630 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4247, + "image_masks_loss": 0.0713, + "llm_learning_rate": 3.6167915747465646e-06, + "loss": 0.781, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0003616791574746565, + "step": 3631 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4246, + "image_masks_loss": 0.0248, + "llm_learning_rate": 3.6160492754998e-06, + "loss": 0.8246, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00036160492754998, + "step": 3632 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4245, + "image_masks_loss": 0.0205, + "llm_learning_rate": 3.6153068533512024e-06, + "loss": 0.8042, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.00036153068533512024, + "step": 3633 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4244, + "image_masks_loss": 0.0507, + "llm_learning_rate": 3.61456430838253e-06, + "loss": 0.7556, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.00036145643083825295, + "step": 3634 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4243, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.613821640675554e-06, + "loss": 0.6943, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003613821640675554, + "step": 3635 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4242, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.6130788503120574e-06, + "loss": 0.8245, + "output_text_mask_loss": 0.0225, + "predictor_learning_rate": 0.0003613078850312057, + "step": 3636 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4241, + "image_masks_loss": 0.0056, + "llm_learning_rate": 3.6123359373738375e-06, + "loss": 0.7114, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003612335937373837, + "step": 3637 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.424, + "image_masks_loss": 0.0176, + "llm_learning_rate": 3.611592901942708e-06, + "loss": 0.7782, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003611592901942708, + "step": 3638 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4239, + "image_masks_loss": 0.0218, + "llm_learning_rate": 3.610849744100492e-06, + "loss": 0.8309, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003610849744100492, + "step": 3639 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4238, + "image_masks_loss": 0.0164, + "llm_learning_rate": 3.6101064639290272e-06, + "loss": 0.8635, + "output_text_mask_loss": 0.0736, + "predictor_learning_rate": 0.00036101064639290273, + "step": 3640 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4237, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.609363061510166e-06, + "loss": 0.7996, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 0.00036093630615101655, + "step": 3641 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4236, + "image_masks_loss": 0.019, + "llm_learning_rate": 3.608619536925775e-06, + "loss": 0.7808, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00036086195369257747, + "step": 3642 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4235, + "image_masks_loss": 0.0184, + "llm_learning_rate": 3.6078758902577295e-06, + "loss": 0.7522, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 0.0003607875890257729, + "step": 3643 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4234, + "image_masks_loss": 0.0038, + "llm_learning_rate": 3.6071321215879236e-06, + "loss": 0.7953, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00036071321215879235, + "step": 3644 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4233, + "image_masks_loss": 0.0101, + "llm_learning_rate": 3.606388230998263e-06, + "loss": 0.7544, + "output_text_mask_loss": 0.0341, + "predictor_learning_rate": 0.00036063882309982624, + "step": 3645 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4232, + "image_masks_loss": 0.0084, + "llm_learning_rate": 3.605644218570665e-06, + "loss": 0.6976, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003605644218570665, + "step": 3646 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4231, + "image_masks_loss": 0.0305, + "llm_learning_rate": 3.6049000843870634e-06, + "loss": 0.7625, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.0003604900084387063, + "step": 3647 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.423, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.6041558285294036e-06, + "loss": 0.8257, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00036041558285294033, + "step": 3648 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4229, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.6034114510796446e-06, + "loss": 0.7181, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00036034114510796446, + "step": 3649 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4228, + "image_masks_loss": 0.0077, + "llm_learning_rate": 3.602666952119759e-06, + "loss": 0.8297, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00036026669521197586, + "step": 3650 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4227, + "image_masks_loss": 0.0066, + "llm_learning_rate": 3.6019223317317327e-06, + "loss": 0.7077, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0003601922331731732, + "step": 3651 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4226, + "image_masks_loss": 0.0139, + "llm_learning_rate": 3.601177589997564e-06, + "loss": 0.8359, + "output_text_mask_loss": 0.0645, + "predictor_learning_rate": 0.0003601177589997564, + "step": 3652 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4225, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.6004327269992666e-06, + "loss": 0.6916, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00036004327269992665, + "step": 3653 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4224, + "image_masks_loss": 0.0207, + "llm_learning_rate": 3.5996877428188664e-06, + "loss": 0.7465, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003599687742818866, + "step": 3654 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4223, + "image_masks_loss": 0.0072, + "llm_learning_rate": 3.598942637538403e-06, + "loss": 0.7754, + "output_text_mask_loss": 0.0357, + "predictor_learning_rate": 0.00035989426375384025, + "step": 3655 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4222, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.5981974112399277e-06, + "loss": 0.7909, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 0.00035981974112399273, + "step": 3656 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4221, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.597452064005507e-06, + "loss": 0.7766, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0003597452064005507, + "step": 3657 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.422, + "image_masks_loss": 0.0128, + "llm_learning_rate": 3.5967065959172216e-06, + "loss": 0.6028, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00035967065959172215, + "step": 3658 + }, + { + "epoch": 0.37, + "gumbel_tau": 0.4219, + "image_masks_loss": 0.0213, + "llm_learning_rate": 3.5959610070571625e-06, + "loss": 0.7929, + "output_text_mask_loss": 0.0744, + "predictor_learning_rate": 0.00035959610070571623, + "step": 3659 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4218, + "image_masks_loss": 0.0148, + "llm_learning_rate": 3.595215297507436e-06, + "loss": 0.7804, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.0003595215297507436, + "step": 3660 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4217, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.5944694673501625e-06, + "loss": 0.6381, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0003594469467350162, + "step": 3661 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4216, + "image_masks_loss": 0.0269, + "llm_learning_rate": 3.5937235166674723e-06, + "loss": 0.7431, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0003593723516667472, + "step": 3662 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4215, + "image_masks_loss": 0.0146, + "llm_learning_rate": 3.5929774455415124e-06, + "loss": 0.7804, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003592977445541512, + "step": 3663 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4214, + "image_masks_loss": 0.0207, + "llm_learning_rate": 3.592231254054442e-06, + "loss": 0.7842, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.00035922312540544415, + "step": 3664 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4213, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.591484942288433e-06, + "loss": 0.6504, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0003591484942288433, + "step": 3665 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4212, + "image_masks_loss": 0.0227, + "llm_learning_rate": 3.5907385103256705e-06, + "loss": 0.7534, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000359073851032567, + "step": 3666 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4211, + "image_masks_loss": 0.0176, + "llm_learning_rate": 3.589991958248354e-06, + "loss": 0.8607, + "output_text_mask_loss": 0.1167, + "predictor_learning_rate": 0.0003589991958248354, + "step": 3667 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.421, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.589245286138695e-06, + "loss": 0.6879, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 0.00035892452861386947, + "step": 3668 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4209, + "image_masks_loss": 0.0037, + "llm_learning_rate": 3.5884984940789196e-06, + "loss": 0.7627, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.00035884984940789196, + "step": 3669 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4208, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.5877515821512654e-06, + "loss": 0.7371, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.0003587751582151265, + "step": 3670 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4207, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.5870045504379837e-06, + "loss": 0.7896, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.0003587004550437983, + "step": 3671 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4206, + "image_masks_loss": 0.0376, + "llm_learning_rate": 3.5862573990213393e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003586257399021339, + "step": 3672 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4205, + "image_masks_loss": 0.0139, + "llm_learning_rate": 3.5855101279836113e-06, + "loss": 0.7652, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0003585510127983611, + "step": 3673 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4204, + "image_masks_loss": 0.0236, + "llm_learning_rate": 3.5847627374070898e-06, + "loss": 0.8212, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 0.00035847627374070896, + "step": 3674 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4203, + "image_masks_loss": 0.0173, + "llm_learning_rate": 3.5840152273740796e-06, + "loss": 0.7548, + "output_text_mask_loss": 0.0011, + "predictor_learning_rate": 0.00035840152273740794, + "step": 3675 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4202, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.5832675979668976e-06, + "loss": 0.8701, + "output_text_mask_loss": 0.0469, + "predictor_learning_rate": 0.0003583267597966897, + "step": 3676 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4201, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.5825198492678753e-06, + "loss": 0.7861, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003582519849267875, + "step": 3677 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.42, + "image_masks_loss": 0.017, + "llm_learning_rate": 3.581771981359356e-06, + "loss": 0.7601, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00035817719813593555, + "step": 3678 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4199, + "image_masks_loss": 0.0027, + "llm_learning_rate": 3.5810239943236963e-06, + "loss": 0.7576, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0003581023994323696, + "step": 3679 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4198, + "image_masks_loss": 0.0212, + "llm_learning_rate": 3.5802758882432666e-06, + "loss": 0.6348, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.0003580275888243266, + "step": 3680 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4197, + "image_masks_loss": 0.0217, + "llm_learning_rate": 3.5795276632004498e-06, + "loss": 0.7964, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.000357952766320045, + "step": 3681 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4196, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.578779319277642e-06, + "loss": 0.7613, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00035787793192776417, + "step": 3682 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4195, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.578030856557253e-06, + "loss": 0.7526, + "output_text_mask_loss": 0.0463, + "predictor_learning_rate": 0.00035780308565572526, + "step": 3683 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4194, + "image_masks_loss": 0.018, + "llm_learning_rate": 3.5772822751217053e-06, + "loss": 0.7921, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.0003577282275121705, + "step": 3684 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4193, + "image_masks_loss": 0.0172, + "llm_learning_rate": 3.576533575053433e-06, + "loss": 0.8067, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0003576533575053433, + "step": 3685 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4192, + "image_masks_loss": 0.0091, + "llm_learning_rate": 3.5757847564348857e-06, + "loss": 0.8091, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00035757847564348857, + "step": 3686 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4191, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.5750358193485245e-06, + "loss": 0.8265, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003575035819348524, + "step": 3687 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.419, + "image_masks_loss": 0.0163, + "llm_learning_rate": 3.5742867638768254e-06, + "loss": 0.7659, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0003574286763876825, + "step": 3688 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4189, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.5735375901022745e-06, + "loss": 0.673, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.0003573537590102274, + "step": 3689 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4188, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.5727882981073725e-06, + "loss": 0.7604, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003572788298107372, + "step": 3690 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4187, + "image_masks_loss": 0.0127, + "llm_learning_rate": 3.5720388879746342e-06, + "loss": 0.84, + "output_text_mask_loss": 0.0018, + "predictor_learning_rate": 0.0003572038887974634, + "step": 3691 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4186, + "image_masks_loss": 0.0206, + "llm_learning_rate": 3.571289359786585e-06, + "loss": 0.8518, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.00035712893597865847, + "step": 3692 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4185, + "image_masks_loss": 0.0184, + "llm_learning_rate": 3.570539713625766e-06, + "loss": 0.7536, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00035705397136257655, + "step": 3693 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4184, + "image_masks_loss": 0.0139, + "llm_learning_rate": 3.569789949574729e-06, + "loss": 0.7347, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.0003569789949574729, + "step": 3694 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4183, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.569040067716039e-06, + "loss": 0.717, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0003569040067716039, + "step": 3695 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4182, + "image_masks_loss": 0.0259, + "llm_learning_rate": 3.5682900681322767e-06, + "loss": 0.8391, + "output_text_mask_loss": 0.0458, + "predictor_learning_rate": 0.0003568290068132277, + "step": 3696 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4181, + "image_masks_loss": 0.0137, + "llm_learning_rate": 3.5675399509060326e-06, + "loss": 0.7936, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.00035675399509060325, + "step": 3697 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.418, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.5667897161199105e-06, + "loss": 0.6995, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00035667897161199103, + "step": 3698 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4179, + "image_masks_loss": 0.0228, + "llm_learning_rate": 3.5660393638565295e-06, + "loss": 0.8359, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003566039363856529, + "step": 3699 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4178, + "image_masks_loss": 0.0058, + "llm_learning_rate": 3.565288894198519e-06, + "loss": 0.8504, + "output_text_mask_loss": 0.0553, + "predictor_learning_rate": 0.00035652888941985184, + "step": 3700 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4177, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.5645383072285224e-06, + "loss": 0.7984, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003564538307228522, + "step": 3701 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4176, + "image_masks_loss": 0.0176, + "llm_learning_rate": 3.5637876030291956e-06, + "loss": 0.731, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.00035637876030291955, + "step": 3702 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4175, + "image_masks_loss": 0.0187, + "llm_learning_rate": 3.5630367816832095e-06, + "loss": 0.6351, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00035630367816832096, + "step": 3703 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4174, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.562285843273245e-06, + "loss": 0.7142, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.0003562285843273245, + "step": 3704 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4173, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.561534787881996e-06, + "loss": 0.838, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0003561534787881996, + "step": 3705 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4172, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.5607836155921733e-06, + "loss": 0.7405, + "output_text_mask_loss": 0.0225, + "predictor_learning_rate": 0.0003560783615592173, + "step": 3706 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4171, + "image_masks_loss": 0.0269, + "llm_learning_rate": 3.5600323264864963e-06, + "loss": 0.8595, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003560032326486496, + "step": 3707 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.417, + "image_masks_loss": 0.0111, + "llm_learning_rate": 3.559280920647697e-06, + "loss": 0.7468, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0003559280920647697, + "step": 3708 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4169, + "image_masks_loss": 0.0179, + "llm_learning_rate": 3.558529398158524e-06, + "loss": 0.8, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00035585293981585235, + "step": 3709 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4168, + "image_masks_loss": 0.0228, + "llm_learning_rate": 3.557777759101736e-06, + "loss": 0.6795, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00035577777591017356, + "step": 3710 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4168, + "image_masks_loss": 0.0353, + "llm_learning_rate": 3.5570260035601043e-06, + "loss": 0.844, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003557026003560104, + "step": 3711 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4167, + "image_masks_loss": 0.0166, + "llm_learning_rate": 3.556274131616415e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 0.00035562741316164147, + "step": 3712 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4166, + "image_masks_loss": 0.0084, + "llm_learning_rate": 3.555522143353466e-06, + "loss": 0.7176, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0003555522143353466, + "step": 3713 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4165, + "image_masks_loss": 0.0086, + "llm_learning_rate": 3.5547700388540673e-06, + "loss": 0.7156, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00035547700388540674, + "step": 3714 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4164, + "image_masks_loss": 0.01, + "llm_learning_rate": 3.5540178182010422e-06, + "loss": 0.7567, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003554017818201042, + "step": 3715 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4163, + "image_masks_loss": 0.0256, + "llm_learning_rate": 3.553265481477228e-06, + "loss": 0.7871, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0003553265481477228, + "step": 3716 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4162, + "image_masks_loss": 0.0427, + "llm_learning_rate": 3.5525130287654723e-06, + "loss": 0.7093, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003552513028765472, + "step": 3717 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4161, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.551760460148638e-06, + "loss": 0.8311, + "output_text_mask_loss": 0.038, + "predictor_learning_rate": 0.00035517604601486376, + "step": 3718 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.416, + "image_masks_loss": 0.0071, + "llm_learning_rate": 3.5510077757095984e-06, + "loss": 0.8459, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00035510077757095983, + "step": 3719 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4159, + "image_masks_loss": 0.0158, + "llm_learning_rate": 3.550254975531242e-06, + "loss": 0.8998, + "output_text_mask_loss": 0.0839, + "predictor_learning_rate": 0.0003550254975531242, + "step": 3720 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4158, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.5495020596964674e-06, + "loss": 0.7234, + "output_text_mask_loss": 0.135, + "predictor_learning_rate": 0.0003549502059696467, + "step": 3721 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4157, + "image_masks_loss": 0.015, + "llm_learning_rate": 3.5487490282881894e-06, + "loss": 0.7884, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.0003548749028288189, + "step": 3722 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4156, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.5479958813893323e-06, + "loss": 0.8541, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003547995881389332, + "step": 3723 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4155, + "image_masks_loss": 0.0071, + "llm_learning_rate": 3.547242619082834e-06, + "loss": 1.0547, + "output_text_mask_loss": 0.6042, + "predictor_learning_rate": 0.00035472426190828336, + "step": 3724 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4154, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.5464892414516457e-06, + "loss": 1.059, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.0003546489241451645, + "step": 3725 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4153, + "image_masks_loss": 0.0147, + "llm_learning_rate": 3.5457357485787326e-06, + "loss": 0.9864, + "output_text_mask_loss": 0.0801, + "predictor_learning_rate": 0.0003545735748578732, + "step": 3726 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4152, + "image_masks_loss": 0.0139, + "llm_learning_rate": 3.5449821405470676e-06, + "loss": 0.9396, + "output_text_mask_loss": 0.1541, + "predictor_learning_rate": 0.0003544982140547067, + "step": 3727 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4151, + "image_masks_loss": 0.0186, + "llm_learning_rate": 3.5442284174396422e-06, + "loss": 0.7654, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.0003544228417439642, + "step": 3728 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.415, + "image_masks_loss": 0.0098, + "llm_learning_rate": 3.543474579339459e-06, + "loss": 0.7605, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.00035434745793394584, + "step": 3729 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4149, + "image_masks_loss": 0.0313, + "llm_learning_rate": 3.5427206263295293e-06, + "loss": 0.7533, + "output_text_mask_loss": 0.0097, + "predictor_learning_rate": 0.0003542720626329529, + "step": 3730 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4148, + "image_masks_loss": 0.0198, + "llm_learning_rate": 3.541966558492882e-06, + "loss": 0.7098, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00035419665584928816, + "step": 3731 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4147, + "image_masks_loss": 0.0063, + "llm_learning_rate": 3.5412123759125566e-06, + "loss": 0.8001, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.00035412123759125564, + "step": 3732 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4146, + "image_masks_loss": 0.0328, + "llm_learning_rate": 3.5404580786716052e-06, + "loss": 0.7651, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00035404580786716053, + "step": 3733 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4145, + "image_masks_loss": 0.0347, + "llm_learning_rate": 3.539703666853092e-06, + "loss": 0.7801, + "output_text_mask_loss": 0.0736, + "predictor_learning_rate": 0.0003539703666853092, + "step": 3734 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4144, + "image_masks_loss": 0.0222, + "llm_learning_rate": 3.5389491405400967e-06, + "loss": 0.7356, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003538949140540096, + "step": 3735 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4143, + "image_masks_loss": 0.0206, + "llm_learning_rate": 3.538194499815707e-06, + "loss": 0.8912, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.0003538194499815707, + "step": 3736 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4142, + "image_masks_loss": 0.0188, + "llm_learning_rate": 3.5374397447630255e-06, + "loss": 0.7942, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.0003537439744763025, + "step": 3737 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4141, + "image_masks_loss": 0.0124, + "llm_learning_rate": 3.5366848754651696e-06, + "loss": 0.7638, + "output_text_mask_loss": 0.0341, + "predictor_learning_rate": 0.0003536684875465169, + "step": 3738 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.414, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.535929892005266e-06, + "loss": 0.6979, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 0.00035359298920052657, + "step": 3739 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4139, + "image_masks_loss": 0.0052, + "llm_learning_rate": 3.535174794466455e-06, + "loss": 0.7433, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0003535174794466455, + "step": 3740 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4138, + "image_masks_loss": 0.0212, + "llm_learning_rate": 3.5344195829318896e-06, + "loss": 0.7593, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.0003534419582931889, + "step": 3741 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4137, + "image_masks_loss": 0.0119, + "llm_learning_rate": 3.533664257484736e-06, + "loss": 0.7038, + "output_text_mask_loss": 0.1083, + "predictor_learning_rate": 0.0003533664257484736, + "step": 3742 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4136, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.5329088182081724e-06, + "loss": 0.75, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003532908818208172, + "step": 3743 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4135, + "image_masks_loss": 0.0303, + "llm_learning_rate": 3.5321532651853884e-06, + "loss": 0.7663, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003532153265185388, + "step": 3744 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4134, + "image_masks_loss": 0.0222, + "llm_learning_rate": 3.5313975984995887e-06, + "loss": 0.7567, + "output_text_mask_loss": 0.0584, + "predictor_learning_rate": 0.0003531397598499588, + "step": 3745 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4133, + "image_masks_loss": 0.0157, + "llm_learning_rate": 3.5306418182339873e-06, + "loss": 0.695, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003530641818233987, + "step": 3746 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4132, + "image_masks_loss": 0.0107, + "llm_learning_rate": 3.529885924471813e-06, + "loss": 0.7026, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0003529885924471813, + "step": 3747 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4131, + "image_masks_loss": 0.0131, + "llm_learning_rate": 3.5291299172963068e-06, + "loss": 0.7602, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.00035291299172963064, + "step": 3748 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.413, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.5283737967907224e-06, + "loss": 0.7204, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003528373796790722, + "step": 3749 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4129, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.527617563038324e-06, + "loss": 0.7455, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0003527617563038324, + "step": 3750 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4128, + "image_masks_loss": 0.0068, + "llm_learning_rate": 3.5268612161223916e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0288, + "predictor_learning_rate": 0.0003526861216122391, + "step": 3751 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4127, + "image_masks_loss": 0.0164, + "llm_learning_rate": 3.526104756126214e-06, + "loss": 0.7367, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003526104756126214, + "step": 3752 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4126, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.525348183133096e-06, + "loss": 0.8882, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003525348183133096, + "step": 3753 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4125, + "image_masks_loss": 0.0126, + "llm_learning_rate": 3.524591497226352e-06, + "loss": 0.8447, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.0003524591497226352, + "step": 3754 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4124, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.5238346984893097e-06, + "loss": 0.6788, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.0003523834698489309, + "step": 3755 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4124, + "image_masks_loss": 0.0105, + "llm_learning_rate": 3.5230777870053102e-06, + "loss": 0.7731, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000352307778700531, + "step": 3756 + }, + { + "epoch": 0.38, + "gumbel_tau": 0.4123, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.5223207628577054e-06, + "loss": 0.8039, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.0003522320762857705, + "step": 3757 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4122, + "image_masks_loss": 0.0305, + "llm_learning_rate": 3.5215636261298614e-06, + "loss": 0.8329, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00035215636261298614, + "step": 3758 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4121, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.520806376905156e-06, + "loss": 0.6653, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003520806376905156, + "step": 3759 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.412, + "image_masks_loss": 0.018, + "llm_learning_rate": 3.5200490152669774e-06, + "loss": 0.771, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00035200490152669775, + "step": 3760 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4119, + "image_masks_loss": 0.023, + "llm_learning_rate": 3.5192915412987304e-06, + "loss": 0.8074, + "output_text_mask_loss": 0.0359, + "predictor_learning_rate": 0.00035192915412987305, + "step": 3761 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4118, + "image_masks_loss": 0.0109, + "llm_learning_rate": 3.518533955083828e-06, + "loss": 0.7399, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003518533955083828, + "step": 3762 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4117, + "image_masks_loss": 0.034, + "llm_learning_rate": 3.5177762567056977e-06, + "loss": 0.7268, + "output_text_mask_loss": 0.0257, + "predictor_learning_rate": 0.0003517776256705698, + "step": 3763 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4116, + "image_masks_loss": 0.0243, + "llm_learning_rate": 3.5170184462477797e-06, + "loss": 0.7393, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00035170184462477794, + "step": 3764 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4115, + "image_masks_loss": 0.0072, + "llm_learning_rate": 3.5162605237935243e-06, + "loss": 0.7314, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0003516260523793524, + "step": 3765 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4114, + "image_masks_loss": 0.0296, + "llm_learning_rate": 3.515502489426397e-06, + "loss": 0.824, + "output_text_mask_loss": 0.0645, + "predictor_learning_rate": 0.0003515502489426397, + "step": 3766 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4113, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.514744343229872e-06, + "loss": 0.7804, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003514744343229872, + "step": 3767 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4112, + "image_masks_loss": 0.019, + "llm_learning_rate": 3.5139860852874414e-06, + "loss": 0.8015, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.0003513986085287441, + "step": 3768 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4111, + "image_masks_loss": 0.0233, + "llm_learning_rate": 3.513227715682604e-06, + "loss": 0.8389, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003513227715682604, + "step": 3769 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.411, + "image_masks_loss": 0.0209, + "llm_learning_rate": 3.512469234498873e-06, + "loss": 0.8022, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.0003512469234498873, + "step": 3770 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4109, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.5117106418197756e-06, + "loss": 0.7431, + "output_text_mask_loss": 0.0717, + "predictor_learning_rate": 0.00035117106418197755, + "step": 3771 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4108, + "image_masks_loss": 0.0362, + "llm_learning_rate": 3.5109519377288494e-06, + "loss": 0.7349, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 0.0003510951937728849, + "step": 3772 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4107, + "image_masks_loss": 0.0319, + "llm_learning_rate": 3.5101931223096423e-06, + "loss": 0.776, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0003510193122309642, + "step": 3773 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4106, + "image_masks_loss": 0.0161, + "llm_learning_rate": 3.50943419564572e-06, + "loss": 0.7359, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.000350943419564572, + "step": 3774 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4105, + "image_masks_loss": 0.023, + "llm_learning_rate": 3.5086751578206546e-06, + "loss": 0.7876, + "output_text_mask_loss": 0.0479, + "predictor_learning_rate": 0.00035086751578206543, + "step": 3775 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4104, + "image_masks_loss": 0.0157, + "llm_learning_rate": 3.507916008918034e-06, + "loss": 0.7658, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0003507916008918034, + "step": 3776 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4103, + "image_masks_loss": 0.0168, + "llm_learning_rate": 3.5071567490214585e-06, + "loss": 0.7827, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.00035071567490214587, + "step": 3777 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4102, + "image_masks_loss": 0.0399, + "llm_learning_rate": 3.506397378214538e-06, + "loss": 0.8083, + "output_text_mask_loss": 0.0858, + "predictor_learning_rate": 0.00035063973782145376, + "step": 3778 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4101, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.5056378965808962e-06, + "loss": 0.7606, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0003505637896580896, + "step": 3779 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.41, + "image_masks_loss": 0.0163, + "llm_learning_rate": 3.504878304204169e-06, + "loss": 0.873, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 0.00035048783042041686, + "step": 3780 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4099, + "image_masks_loss": 0.0374, + "llm_learning_rate": 3.504118601168006e-06, + "loss": 0.8646, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0003504118601168006, + "step": 3781 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4098, + "image_masks_loss": 0.0084, + "llm_learning_rate": 3.5033587875560656e-06, + "loss": 0.6354, + "output_text_mask_loss": 0.0702, + "predictor_learning_rate": 0.00035033587875560654, + "step": 3782 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4097, + "image_masks_loss": 0.0145, + "llm_learning_rate": 3.50259886345202e-06, + "loss": 0.7669, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.00035025988634520194, + "step": 3783 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4096, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.5018388289395543e-06, + "loss": 0.7822, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.00035018388289395543, + "step": 3784 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4095, + "image_masks_loss": 0.0143, + "llm_learning_rate": 3.501078684102366e-06, + "loss": 0.7535, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00035010786841023657, + "step": 3785 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4094, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.500318429024162e-06, + "loss": 0.7927, + "output_text_mask_loss": 0.0896, + "predictor_learning_rate": 0.0003500318429024162, + "step": 3786 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4093, + "image_masks_loss": 0.0067, + "llm_learning_rate": 3.499558063788665e-06, + "loss": 0.7563, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.0003499558063788665, + "step": 3787 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4092, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.498797588479608e-06, + "loss": 0.7332, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00034987975884796075, + "step": 3788 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4092, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.498037003180736e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0003498037003180736, + "step": 3789 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4091, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.4972763079758053e-06, + "loss": 0.769, + "output_text_mask_loss": 0.1038, + "predictor_learning_rate": 0.00034972763079758055, + "step": 3790 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.409, + "image_masks_loss": 0.034, + "llm_learning_rate": 3.496515502948587e-06, + "loss": 0.7794, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003496515502948587, + "step": 3791 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4089, + "image_masks_loss": 0.0221, + "llm_learning_rate": 3.4957545881828614e-06, + "loss": 0.7594, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.0003495754588182861, + "step": 3792 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4088, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.494993563762423e-06, + "loss": 0.8093, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00034949935637624223, + "step": 3793 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4087, + "image_masks_loss": 0.0209, + "llm_learning_rate": 3.4942324297710767e-06, + "loss": 0.8021, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00034942324297710764, + "step": 3794 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4086, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.4934711862926405e-06, + "loss": 0.7286, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00034934711862926403, + "step": 3795 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4085, + "image_masks_loss": 0.034, + "llm_learning_rate": 3.4927098334109444e-06, + "loss": 0.7568, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0003492709833410944, + "step": 3796 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4084, + "image_masks_loss": 0.0051, + "llm_learning_rate": 3.491948371209831e-06, + "loss": 0.8261, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.0003491948371209831, + "step": 3797 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4083, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.491186799773153e-06, + "loss": 0.7479, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00034911867997731527, + "step": 3798 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4082, + "image_masks_loss": 0.01, + "llm_learning_rate": 3.4904251191847767e-06, + "loss": 0.6794, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00034904251191847764, + "step": 3799 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4081, + "image_masks_loss": 0.0129, + "llm_learning_rate": 3.489663329528581e-06, + "loss": 0.8087, + "output_text_mask_loss": 0.0607, + "predictor_learning_rate": 0.0003489663329528581, + "step": 3800 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.408, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.4889014308884546e-06, + "loss": 0.9232, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00034889014308884543, + "step": 3801 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4079, + "image_masks_loss": 0.0313, + "llm_learning_rate": 3.488139423348299e-06, + "loss": 0.6948, + "output_text_mask_loss": 0.0011, + "predictor_learning_rate": 0.00034881394233482986, + "step": 3802 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4078, + "image_masks_loss": 0.018, + "llm_learning_rate": 3.48737730699203e-06, + "loss": 0.8553, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000348737730699203, + "step": 3803 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4077, + "image_masks_loss": 0.0378, + "llm_learning_rate": 3.4866150819035726e-06, + "loss": 0.7279, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00034866150819035726, + "step": 3804 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4076, + "image_masks_loss": 0.0193, + "llm_learning_rate": 3.4858527481668647e-06, + "loss": 0.7175, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00034858527481668644, + "step": 3805 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4075, + "image_masks_loss": 0.008, + "llm_learning_rate": 3.485090305865856e-06, + "loss": 0.7734, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00034850903058658556, + "step": 3806 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4074, + "image_masks_loss": 0.0225, + "llm_learning_rate": 3.484327755084509e-06, + "loss": 0.7801, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003484327755084509, + "step": 3807 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4073, + "image_masks_loss": 0.0191, + "llm_learning_rate": 3.483565095906797e-06, + "loss": 0.7182, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00034835650959067974, + "step": 3808 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4072, + "image_masks_loss": 0.0214, + "llm_learning_rate": 3.4828023284167062e-06, + "loss": 0.7768, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0003482802328416706, + "step": 3809 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4071, + "image_masks_loss": 0.0324, + "llm_learning_rate": 3.4820394526982337e-06, + "loss": 0.8328, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00034820394526982335, + "step": 3810 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.407, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.4812764688353894e-06, + "loss": 0.6878, + "output_text_mask_loss": 0.05, + "predictor_learning_rate": 0.0003481276468835389, + "step": 3811 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4069, + "image_masks_loss": 0.0191, + "llm_learning_rate": 3.4805133769121933e-06, + "loss": 0.767, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0003480513376912193, + "step": 3812 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4068, + "image_masks_loss": 0.0269, + "llm_learning_rate": 3.479750177012682e-06, + "loss": 0.7432, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0003479750177012682, + "step": 3813 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4067, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.478986869220898e-06, + "loss": 0.7778, + "output_text_mask_loss": 0.0213, + "predictor_learning_rate": 0.00034789868692208976, + "step": 3814 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4067, + "image_masks_loss": 0.0484, + "llm_learning_rate": 3.478223453620899e-06, + "loss": 0.7653, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.0003478223453620899, + "step": 3815 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4066, + "image_masks_loss": 0.0301, + "llm_learning_rate": 3.477459930296755e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00034774599302967547, + "step": 3816 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4065, + "image_masks_loss": 0.0172, + "llm_learning_rate": 3.476696299332546e-06, + "loss": 0.8032, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003476696299332546, + "step": 3817 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4064, + "image_masks_loss": 0.0076, + "llm_learning_rate": 3.475932560812365e-06, + "loss": 0.6792, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003475932560812365, + "step": 3818 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4063, + "image_masks_loss": 0.005, + "llm_learning_rate": 3.4751687148203167e-06, + "loss": 0.7987, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00034751687148203164, + "step": 3819 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4062, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.4744047614405172e-06, + "loss": 0.7494, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003474404761440517, + "step": 3820 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4061, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.473640700757096e-06, + "loss": 0.693, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003473640700757096, + "step": 3821 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.406, + "image_masks_loss": 0.0128, + "llm_learning_rate": 3.4728765328541906e-06, + "loss": 0.7766, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0003472876532854191, + "step": 3822 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4059, + "image_masks_loss": 0.0081, + "llm_learning_rate": 3.472112257815955e-06, + "loss": 0.802, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00034721122578159545, + "step": 3823 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4058, + "image_masks_loss": 0.0324, + "llm_learning_rate": 3.471347875726553e-06, + "loss": 0.7675, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0003471347875726553, + "step": 3824 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4057, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.4705833866701584e-06, + "loss": 0.7678, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003470583386670158, + "step": 3825 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4056, + "image_masks_loss": 0.0179, + "llm_learning_rate": 3.4698187907309593e-06, + "loss": 0.8418, + "output_text_mask_loss": 0.1076, + "predictor_learning_rate": 0.0003469818790730959, + "step": 3826 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4055, + "image_masks_loss": 0.0254, + "llm_learning_rate": 3.4690540879931562e-06, + "loss": 0.8149, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.0003469054087993156, + "step": 3827 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4054, + "image_masks_loss": 0.0052, + "llm_learning_rate": 3.4682892785409565e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.0003468289278540956, + "step": 3828 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4053, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.4675243624585857e-06, + "loss": 0.8174, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003467524362458586, + "step": 3829 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4052, + "image_masks_loss": 0.0477, + "llm_learning_rate": 3.4667593398302774e-06, + "loss": 0.7303, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.0003466759339830277, + "step": 3830 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4051, + "image_masks_loss": 0.0349, + "llm_learning_rate": 3.465994210740276e-06, + "loss": 0.5987, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003465994210740276, + "step": 3831 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.405, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.4652289752728412e-06, + "loss": 0.7515, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.0003465228975272841, + "step": 3832 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4049, + "image_masks_loss": 0.0139, + "llm_learning_rate": 3.4644636335122416e-06, + "loss": 0.7608, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.00034644636335122415, + "step": 3833 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4048, + "image_masks_loss": 0.0406, + "llm_learning_rate": 3.463698185542759e-06, + "loss": 0.8, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00034636981855427586, + "step": 3834 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4047, + "image_masks_loss": 0.0191, + "llm_learning_rate": 3.4629326314486846e-06, + "loss": 0.743, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00034629326314486843, + "step": 3835 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4046, + "image_masks_loss": 0.0296, + "llm_learning_rate": 3.462166971314325e-06, + "loss": 0.838, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00034621669713143244, + "step": 3836 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4045, + "image_masks_loss": 0.0301, + "llm_learning_rate": 3.4614012052239954e-06, + "loss": 0.7211, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0003461401205223995, + "step": 3837 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4044, + "image_masks_loss": 0.0098, + "llm_learning_rate": 3.4606353332620226e-06, + "loss": 0.844, + "output_text_mask_loss": 0.0675, + "predictor_learning_rate": 0.00034606353332620224, + "step": 3838 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4044, + "image_masks_loss": 0.0186, + "llm_learning_rate": 3.4598693555127476e-06, + "loss": 0.8138, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00034598693555127475, + "step": 3839 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4043, + "image_masks_loss": 0.0542, + "llm_learning_rate": 3.459103272060521e-06, + "loss": 0.7851, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003459103272060521, + "step": 3840 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4042, + "image_masks_loss": 0.0404, + "llm_learning_rate": 3.4583370829897064e-06, + "loss": 0.8317, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.0003458337082989706, + "step": 3841 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4041, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.457570788384677e-06, + "loss": 0.7292, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0003457570788384677, + "step": 3842 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.404, + "image_masks_loss": 0.0387, + "llm_learning_rate": 3.4568043883298197e-06, + "loss": 0.812, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00034568043883298195, + "step": 3843 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4039, + "image_masks_loss": 0.0313, + "llm_learning_rate": 3.456037882909532e-06, + "loss": 0.7252, + "output_text_mask_loss": 0.0523, + "predictor_learning_rate": 0.0003456037882909532, + "step": 3844 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4038, + "image_masks_loss": 0.061, + "llm_learning_rate": 3.4552712722082225e-06, + "loss": 0.7177, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00034552712722082224, + "step": 3845 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4037, + "image_masks_loss": 0.015, + "llm_learning_rate": 3.454504556310314e-06, + "loss": 0.7483, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0003454504556310314, + "step": 3846 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4036, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.4537377353002366e-06, + "loss": 0.8395, + "output_text_mask_loss": 0.0721, + "predictor_learning_rate": 0.0003453737735300236, + "step": 3847 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4035, + "image_masks_loss": 0.029, + "llm_learning_rate": 3.452970809262436e-06, + "loss": 0.8051, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00034529708092624357, + "step": 3848 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4034, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.452203778281367e-06, + "loss": 0.6978, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0003452203778281367, + "step": 3849 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4033, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.4514366424414975e-06, + "loss": 0.785, + "output_text_mask_loss": 0.0227, + "predictor_learning_rate": 0.0003451436642441497, + "step": 3850 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4032, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.4506694018273044e-06, + "loss": 0.8672, + "output_text_mask_loss": 0.0992, + "predictor_learning_rate": 0.00034506694018273044, + "step": 3851 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4031, + "image_masks_loss": 0.0147, + "llm_learning_rate": 3.4499020565232805e-06, + "loss": 0.7173, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000344990205652328, + "step": 3852 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.403, + "image_masks_loss": 0.0286, + "llm_learning_rate": 3.4491346066139258e-06, + "loss": 0.7462, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00034491346066139257, + "step": 3853 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4029, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.448367052183754e-06, + "loss": 0.7691, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00034483670521837534, + "step": 3854 + }, + { + "epoch": 0.39, + "gumbel_tau": 0.4028, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.447599393317291e-06, + "loss": 0.7286, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003447599393317291, + "step": 3855 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4027, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.446831630099071e-06, + "loss": 0.8368, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.0003446831630099071, + "step": 3856 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4026, + "image_masks_loss": 0.025, + "llm_learning_rate": 3.4460637626136427e-06, + "loss": 0.6642, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.00034460637626136426, + "step": 3857 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4025, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.4452957909455662e-06, + "loss": 0.8161, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00034452957909455664, + "step": 3858 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4025, + "image_masks_loss": 0.0309, + "llm_learning_rate": 3.4445277151794122e-06, + "loss": 0.8605, + "output_text_mask_loss": 0.0847, + "predictor_learning_rate": 0.0003444527715179412, + "step": 3859 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4024, + "image_masks_loss": 0.0046, + "llm_learning_rate": 3.4437595353997613e-06, + "loss": 0.7246, + "output_text_mask_loss": 0.0317, + "predictor_learning_rate": 0.0003443759535399761, + "step": 3860 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4023, + "image_masks_loss": 0.0099, + "llm_learning_rate": 3.4429912516912086e-06, + "loss": 0.7442, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00034429912516912084, + "step": 3861 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4022, + "image_masks_loss": 0.0233, + "llm_learning_rate": 3.442222864138359e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.00034422228641383583, + "step": 3862 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4021, + "image_masks_loss": 0.0217, + "llm_learning_rate": 3.441454372825829e-06, + "loss": 0.7842, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.00034414543728258283, + "step": 3863 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.402, + "image_masks_loss": 0.0244, + "llm_learning_rate": 3.4406857778382457e-06, + "loss": 0.8318, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00034406857778382453, + "step": 3864 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4019, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.4399170792602497e-06, + "loss": 0.8531, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00034399170792602496, + "step": 3865 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4018, + "image_masks_loss": 0.0265, + "llm_learning_rate": 3.4391482771764915e-06, + "loss": 0.8366, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00034391482771764916, + "step": 3866 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4017, + "image_masks_loss": 0.0223, + "llm_learning_rate": 3.438379371671633e-06, + "loss": 0.7364, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00034383793716716326, + "step": 3867 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4016, + "image_masks_loss": 0.0176, + "llm_learning_rate": 3.4376103628303477e-06, + "loss": 0.7806, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 0.0003437610362830347, + "step": 3868 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4015, + "image_masks_loss": 0.0157, + "llm_learning_rate": 3.436841250737321e-06, + "loss": 0.7208, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0003436841250737321, + "step": 3869 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4014, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.4360720354772488e-06, + "loss": 0.7903, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00034360720354772485, + "step": 3870 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4013, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.4353027171348396e-06, + "loss": 0.8456, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00034353027171348395, + "step": 3871 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4012, + "image_masks_loss": 0.0256, + "llm_learning_rate": 3.4345332957948118e-06, + "loss": 0.718, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003434533295794812, + "step": 3872 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4011, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.4337637715418957e-06, + "loss": 0.6282, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.00034337637715418955, + "step": 3873 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.401, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.4329941444608333e-06, + "loss": 0.7268, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00034329941444608334, + "step": 3874 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4009, + "image_masks_loss": 0.0229, + "llm_learning_rate": 3.4322244146363777e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.00034322244146363774, + "step": 3875 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4008, + "image_masks_loss": 0.0027, + "llm_learning_rate": 3.431454582153293e-06, + "loss": 0.7581, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003431454582153293, + "step": 3876 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4007, + "image_masks_loss": 0.0064, + "llm_learning_rate": 3.430684647096354e-06, + "loss": 0.7871, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.0003430684647096354, + "step": 3877 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4007, + "image_masks_loss": 0.0336, + "llm_learning_rate": 3.42991460955035e-06, + "loss": 0.9153, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00034299146095503495, + "step": 3878 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4006, + "image_masks_loss": 0.0066, + "llm_learning_rate": 3.429144469600077e-06, + "loss": 0.6789, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003429144469600077, + "step": 3879 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4005, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.4283742273303457e-06, + "loss": 0.7944, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.00034283742273303453, + "step": 3880 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4004, + "image_masks_loss": 0.0181, + "llm_learning_rate": 3.4276038828259765e-06, + "loss": 0.8414, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00034276038828259765, + "step": 3881 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4003, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.426833436171802e-06, + "loss": 0.7449, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00034268334361718016, + "step": 3882 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4002, + "image_masks_loss": 0.0143, + "llm_learning_rate": 3.426062887452665e-06, + "loss": 0.7068, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00034260628874526643, + "step": 3883 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4001, + "image_masks_loss": 0.0202, + "llm_learning_rate": 3.4252922367534196e-06, + "loss": 0.7492, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.000342529223675342, + "step": 3884 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.4, + "image_masks_loss": 0.0322, + "llm_learning_rate": 3.4245214841589326e-06, + "loss": 0.7758, + "output_text_mask_loss": 0.0645, + "predictor_learning_rate": 0.0003424521484158932, + "step": 3885 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3999, + "image_masks_loss": 0.01, + "llm_learning_rate": 3.4237506297540805e-06, + "loss": 0.6522, + "output_text_mask_loss": 0.0159, + "predictor_learning_rate": 0.00034237506297540805, + "step": 3886 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3998, + "image_masks_loss": 0.0042, + "llm_learning_rate": 3.4229796736237508e-06, + "loss": 0.7517, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.0003422979673623751, + "step": 3887 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3997, + "image_masks_loss": 0.0051, + "llm_learning_rate": 3.4222086158528445e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0003422208615852844, + "step": 3888 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3996, + "image_masks_loss": 0.0132, + "llm_learning_rate": 3.4214374565262708e-06, + "loss": 0.6968, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003421437456526271, + "step": 3889 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3995, + "image_masks_loss": 0.0328, + "llm_learning_rate": 3.4206661957289515e-06, + "loss": 0.722, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00034206661957289513, + "step": 3890 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3994, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.4198948335458203e-06, + "loss": 0.7467, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000341989483354582, + "step": 3891 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3993, + "image_masks_loss": 0.0107, + "llm_learning_rate": 3.419123370061822e-06, + "loss": 0.7329, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00034191233700618217, + "step": 3892 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3992, + "image_masks_loss": 0.0144, + "llm_learning_rate": 3.4183518053619102e-06, + "loss": 0.7801, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000341835180536191, + "step": 3893 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3991, + "image_masks_loss": 0.0132, + "llm_learning_rate": 3.417580139531052e-06, + "loss": 0.7129, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.00034175801395310517, + "step": 3894 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.399, + "image_masks_loss": 0.0085, + "llm_learning_rate": 3.416808372654225e-06, + "loss": 0.8314, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0003416808372654225, + "step": 3895 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.399, + "image_masks_loss": 0.0236, + "llm_learning_rate": 3.416036504816417e-06, + "loss": 0.6543, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003416036504816417, + "step": 3896 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3989, + "image_masks_loss": 0.0124, + "llm_learning_rate": 3.4152645361026306e-06, + "loss": 0.9119, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.000341526453610263, + "step": 3897 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3988, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.4144924665978736e-06, + "loss": 0.7172, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00034144924665978734, + "step": 3898 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3987, + "image_masks_loss": 0.0151, + "llm_learning_rate": 3.41372029638717e-06, + "loss": 0.8102, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.00034137202963871693, + "step": 3899 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3986, + "image_masks_loss": 0.019, + "llm_learning_rate": 3.4129480255555507e-06, + "loss": 0.8218, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00034129480255555507, + "step": 3900 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3985, + "image_masks_loss": 0.0357, + "llm_learning_rate": 3.4121756541880626e-06, + "loss": 0.7238, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003412175654188062, + "step": 3901 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3984, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.4114031823697595e-06, + "loss": 0.762, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003411403182369759, + "step": 3902 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3983, + "image_masks_loss": 0.0381, + "llm_learning_rate": 3.410630610185707e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0003410630610185707, + "step": 3903 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3982, + "image_masks_loss": 0.0202, + "llm_learning_rate": 3.4098579377209846e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00034098579377209843, + "step": 3904 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3981, + "image_masks_loss": 0.0261, + "llm_learning_rate": 3.409085165060679e-06, + "loss": 0.8302, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.00034090851650606785, + "step": 3905 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.398, + "image_masks_loss": 0.0028, + "llm_learning_rate": 3.408312292289889e-06, + "loss": 0.7836, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0003408312292289889, + "step": 3906 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3979, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.407539319493728e-06, + "loss": 0.7483, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00034075393194937277, + "step": 3907 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3978, + "image_masks_loss": 0.0033, + "llm_learning_rate": 3.4067662467573147e-06, + "loss": 0.8146, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0003406766246757315, + "step": 3908 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3977, + "image_masks_loss": 0.0185, + "llm_learning_rate": 3.405993074165783e-06, + "loss": 0.8586, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003405993074165783, + "step": 3909 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3976, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.4052198018042753e-06, + "loss": 0.7152, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 0.00034052198018042753, + "step": 3910 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3975, + "image_masks_loss": 0.0172, + "llm_learning_rate": 3.4044464297579477e-06, + "loss": 0.8004, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00034044464297579475, + "step": 3911 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3975, + "image_masks_loss": 0.0186, + "llm_learning_rate": 3.403672958111964e-06, + "loss": 0.7941, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00034036729581119634, + "step": 3912 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3974, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.402899386951501e-06, + "loss": 0.7606, + "output_text_mask_loss": 0.0576, + "predictor_learning_rate": 0.00034028993869515006, + "step": 3913 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3973, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.402125716361747e-06, + "loss": 0.7734, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003402125716361747, + "step": 3914 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3972, + "image_masks_loss": 0.0142, + "llm_learning_rate": 3.401351946427899e-06, + "loss": 0.7872, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00034013519464278993, + "step": 3915 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3971, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.4005780772351678e-06, + "loss": 0.7494, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00034005780772351676, + "step": 3916 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.397, + "image_masks_loss": 0.0148, + "llm_learning_rate": 3.399804108868772e-06, + "loss": 0.7669, + "output_text_mask_loss": 0.029, + "predictor_learning_rate": 0.0003399804108868772, + "step": 3917 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3969, + "image_masks_loss": 0.0203, + "llm_learning_rate": 3.399030041413944e-06, + "loss": 0.7826, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00033990300414139444, + "step": 3918 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3968, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.3982558749559253e-06, + "loss": 0.8138, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003398255874955925, + "step": 3919 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3967, + "image_masks_loss": 0.0408, + "llm_learning_rate": 3.397481609579969e-06, + "loss": 0.7946, + "output_text_mask_loss": 0.1373, + "predictor_learning_rate": 0.0003397481609579969, + "step": 3920 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3966, + "image_masks_loss": 0.0193, + "llm_learning_rate": 3.396707245371339e-06, + "loss": 0.7262, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003396707245371339, + "step": 3921 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3965, + "image_masks_loss": 0.0099, + "llm_learning_rate": 3.395932782415309e-06, + "loss": 0.7608, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.00033959327824153085, + "step": 3922 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3964, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.3951582207971658e-06, + "loss": 0.8268, + "output_text_mask_loss": 0.1129, + "predictor_learning_rate": 0.00033951582207971656, + "step": 3923 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3963, + "image_masks_loss": 0.0298, + "llm_learning_rate": 3.394383560602206e-06, + "loss": 0.7666, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00033943835606022054, + "step": 3924 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3962, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.3936088019157354e-06, + "loss": 0.7395, + "output_text_mask_loss": 0.1404, + "predictor_learning_rate": 0.0003393608801915735, + "step": 3925 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3961, + "image_masks_loss": 0.012, + "llm_learning_rate": 3.3928339448230728e-06, + "loss": 0.6879, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003392833944823073, + "step": 3926 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.396, + "image_masks_loss": 0.0039, + "llm_learning_rate": 3.3920589894095484e-06, + "loss": 0.7598, + "output_text_mask_loss": 0.0576, + "predictor_learning_rate": 0.00033920589894095485, + "step": 3927 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.396, + "image_masks_loss": 0.0183, + "llm_learning_rate": 3.391283935760501e-06, + "loss": 0.7344, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00033912839357605007, + "step": 3928 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3959, + "image_masks_loss": 0.0201, + "llm_learning_rate": 3.3905087839612818e-06, + "loss": 0.7825, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00033905087839612816, + "step": 3929 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3958, + "image_masks_loss": 0.0205, + "llm_learning_rate": 3.389733534097251e-06, + "loss": 0.7483, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00033897335340972505, + "step": 3930 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3957, + "image_masks_loss": 0.0119, + "llm_learning_rate": 3.388958186253782e-06, + "loss": 0.802, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003388958186253782, + "step": 3931 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3956, + "image_masks_loss": 0.0241, + "llm_learning_rate": 3.388182740516257e-06, + "loss": 0.6775, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003388182740516257, + "step": 3932 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3955, + "image_masks_loss": 0.0105, + "llm_learning_rate": 3.3874071969700706e-06, + "loss": 0.8027, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00033874071969700704, + "step": 3933 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3954, + "image_masks_loss": 0.0271, + "llm_learning_rate": 3.386631555700627e-06, + "loss": 0.7364, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.00033866315557006264, + "step": 3934 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3953, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.3858558167933407e-06, + "loss": 0.7711, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00033858558167933407, + "step": 3935 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3952, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.3850799803336397e-06, + "loss": 0.8036, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.00033850799803336395, + "step": 3936 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3951, + "image_masks_loss": 0.0435, + "llm_learning_rate": 3.38430404640696e-06, + "loss": 0.813, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.000338430404640696, + "step": 3937 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.395, + "image_masks_loss": 0.0278, + "llm_learning_rate": 3.3835280150987488e-06, + "loss": 0.7831, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00033835280150987484, + "step": 3938 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3949, + "image_masks_loss": 0.0425, + "llm_learning_rate": 3.3827518864944646e-06, + "loss": 0.7381, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00033827518864944646, + "step": 3939 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3948, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.381975660679576e-06, + "loss": 0.7294, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003381975660679576, + "step": 3940 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3947, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.381199337739564e-06, + "loss": 0.6871, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003381199337739564, + "step": 3941 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3946, + "image_masks_loss": 0.0186, + "llm_learning_rate": 3.3804229177599168e-06, + "loss": 0.7441, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00033804229177599165, + "step": 3942 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3946, + "image_masks_loss": 0.0185, + "llm_learning_rate": 3.3796464008261375e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00033796464008261376, + "step": 3943 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3945, + "image_masks_loss": 0.0332, + "llm_learning_rate": 3.3788697870237374e-06, + "loss": 0.722, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0003378869787023737, + "step": 3944 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3944, + "image_masks_loss": 0.034, + "llm_learning_rate": 3.3780930764382385e-06, + "loss": 0.7747, + "output_text_mask_loss": 0.0194, + "predictor_learning_rate": 0.00033780930764382383, + "step": 3945 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3943, + "image_masks_loss": 0.0126, + "llm_learning_rate": 3.377316269155174e-06, + "loss": 0.749, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0003377316269155174, + "step": 3946 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3942, + "image_masks_loss": 0.0124, + "llm_learning_rate": 3.376539365260089e-06, + "loss": 0.7873, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00033765393652600886, + "step": 3947 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3941, + "image_masks_loss": 0.0061, + "llm_learning_rate": 3.3757623648385353e-06, + "loss": 0.8444, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.0003375762364838535, + "step": 3948 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.394, + "image_masks_loss": 0.0049, + "llm_learning_rate": 3.3749852679760796e-06, + "loss": 0.7814, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00033749852679760796, + "step": 3949 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3939, + "image_masks_loss": 0.0252, + "llm_learning_rate": 3.374208074758298e-06, + "loss": 0.8011, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 0.00033742080747582975, + "step": 3950 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3938, + "image_masks_loss": 0.0288, + "llm_learning_rate": 3.3734307852707748e-06, + "loss": 0.8915, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00033734307852707746, + "step": 3951 + }, + { + "epoch": 0.4, + "gumbel_tau": 0.3937, + "image_masks_loss": 0.017, + "llm_learning_rate": 3.3726533995991085e-06, + "loss": 0.7105, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003372653399599108, + "step": 3952 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3936, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.371875917828906e-06, + "loss": 0.7791, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003371875917828906, + "step": 3953 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3935, + "image_masks_loss": 0.0467, + "llm_learning_rate": 3.3710983400457865e-06, + "loss": 0.8051, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.0003371098340045786, + "step": 3954 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3934, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.370320666335376e-06, + "loss": 0.7149, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003370320666335376, + "step": 3955 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3933, + "image_masks_loss": 0.0164, + "llm_learning_rate": 3.3695428967833156e-06, + "loss": 0.8157, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00033695428967833155, + "step": 3956 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3933, + "image_masks_loss": 0.0127, + "llm_learning_rate": 3.368765031475255e-06, + "loss": 0.8274, + "output_text_mask_loss": 0.0085, + "predictor_learning_rate": 0.0003368765031475255, + "step": 3957 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3932, + "image_masks_loss": 0.0058, + "llm_learning_rate": 3.3679870704968535e-06, + "loss": 0.8209, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.00033679870704968537, + "step": 3958 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3931, + "image_masks_loss": 0.0123, + "llm_learning_rate": 3.367209013933783e-06, + "loss": 0.7306, + "output_text_mask_loss": 0.2319, + "predictor_learning_rate": 0.0003367209013933783, + "step": 3959 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.393, + "image_masks_loss": 0.0078, + "llm_learning_rate": 3.366430861871724e-06, + "loss": 0.8033, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0003366430861871724, + "step": 3960 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3929, + "image_masks_loss": 0.0507, + "llm_learning_rate": 3.3656526143963687e-06, + "loss": 0.7341, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00033656526143963685, + "step": 3961 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3928, + "image_masks_loss": 0.0123, + "llm_learning_rate": 3.364874271593419e-06, + "loss": 0.8566, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00033648742715934187, + "step": 3962 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3927, + "image_masks_loss": 0.0081, + "llm_learning_rate": 3.3640958335485887e-06, + "loss": 0.6772, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00033640958335485885, + "step": 3963 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3926, + "image_masks_loss": 0.0319, + "llm_learning_rate": 3.3633173003475996e-06, + "loss": 0.6831, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.0003363317300347599, + "step": 3964 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3925, + "image_masks_loss": 0.0298, + "llm_learning_rate": 3.3625386720761865e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00033625386720761866, + "step": 3965 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3924, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.361759948820095e-06, + "loss": 0.8712, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00033617599488200943, + "step": 3966 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3923, + "image_masks_loss": 0.0181, + "llm_learning_rate": 3.3609811306650763e-06, + "loss": 0.7461, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0003360981130665076, + "step": 3967 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3922, + "image_masks_loss": 0.0359, + "llm_learning_rate": 3.360202217696899e-06, + "loss": 0.7724, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003360202217696899, + "step": 3968 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3921, + "image_masks_loss": 0.014, + "llm_learning_rate": 3.359423210001338e-06, + "loss": 0.7091, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003359423210001338, + "step": 3969 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.392, + "image_masks_loss": 0.0199, + "llm_learning_rate": 3.358644107664178e-06, + "loss": 0.7931, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0003358644107664178, + "step": 3970 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.392, + "image_masks_loss": 0.0299, + "llm_learning_rate": 3.357864910771216e-06, + "loss": 0.7271, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00033578649107712154, + "step": 3971 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3919, + "image_masks_loss": 0.019, + "llm_learning_rate": 3.3570856194082598e-06, + "loss": 0.7573, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00033570856194082595, + "step": 3972 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3918, + "image_masks_loss": 0.0119, + "llm_learning_rate": 3.3563062336611257e-06, + "loss": 0.6972, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 0.00033563062336611254, + "step": 3973 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3917, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.355526753615642e-06, + "loss": 0.7853, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.0003355526753615642, + "step": 3974 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3916, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.354747179357647e-06, + "loss": 0.8101, + "output_text_mask_loss": 0.0771, + "predictor_learning_rate": 0.00033547471793576465, + "step": 3975 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3915, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.3539675109729885e-06, + "loss": 0.732, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00033539675109729884, + "step": 3976 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3914, + "image_masks_loss": 0.0385, + "llm_learning_rate": 3.353187748547525e-06, + "loss": 0.7016, + "output_text_mask_loss": 0.0717, + "predictor_learning_rate": 0.0003353187748547525, + "step": 3977 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3913, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.3524078921671265e-06, + "loss": 0.7882, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00033524078921671265, + "step": 3978 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3912, + "image_masks_loss": 0.0188, + "llm_learning_rate": 3.351627941917673e-06, + "loss": 0.8265, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00033516279419176727, + "step": 3979 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3911, + "image_masks_loss": 0.0164, + "llm_learning_rate": 3.3508478978850526e-06, + "loss": 0.7446, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.00033508478978850524, + "step": 3980 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.391, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.350067760155167e-06, + "loss": 0.7382, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00033500677601551665, + "step": 3981 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3909, + "image_masks_loss": 0.0212, + "llm_learning_rate": 3.3492875288139257e-06, + "loss": 0.702, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00033492875288139255, + "step": 3982 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3908, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.348507203947251e-06, + "loss": 0.7996, + "output_text_mask_loss": 0.0732, + "predictor_learning_rate": 0.0003348507203947251, + "step": 3983 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3908, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.347726785641072e-06, + "loss": 0.7312, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003347726785641072, + "step": 3984 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3907, + "image_masks_loss": 0.0061, + "llm_learning_rate": 3.346946273981332e-06, + "loss": 0.7196, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003346946273981332, + "step": 3985 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3906, + "image_masks_loss": 0.0168, + "llm_learning_rate": 3.346165669053982e-06, + "loss": 0.8441, + "output_text_mask_loss": 0.0402, + "predictor_learning_rate": 0.00033461656690539815, + "step": 3986 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3905, + "image_masks_loss": 0.0178, + "llm_learning_rate": 3.3453849709449827e-06, + "loss": 0.802, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0003345384970944982, + "step": 3987 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3904, + "image_masks_loss": 0.0357, + "llm_learning_rate": 3.344604179740308e-06, + "loss": 0.756, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0003344604179740308, + "step": 3988 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3903, + "image_masks_loss": 0.0209, + "llm_learning_rate": 3.3438232955259392e-06, + "loss": 0.7403, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0003343823295525939, + "step": 3989 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3902, + "image_masks_loss": 0.0158, + "llm_learning_rate": 3.3430423183878704e-06, + "loss": 0.7762, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.000334304231838787, + "step": 3990 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3901, + "image_masks_loss": 0.0173, + "llm_learning_rate": 3.3422612484121025e-06, + "loss": 0.7875, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00033422612484121025, + "step": 3991 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.39, + "image_masks_loss": 0.0184, + "llm_learning_rate": 3.3414800856846506e-06, + "loss": 0.7584, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00033414800856846504, + "step": 3992 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3899, + "image_masks_loss": 0.0194, + "llm_learning_rate": 3.3406988302915376e-06, + "loss": 0.8329, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00033406988302915375, + "step": 3993 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3898, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.339917482318797e-06, + "loss": 0.7787, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 0.0003339917482318797, + "step": 3994 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3897, + "image_masks_loss": 0.0151, + "llm_learning_rate": 3.3391360418524714e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0003339136041852471, + "step": 3995 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3897, + "image_masks_loss": 0.0241, + "llm_learning_rate": 3.338354508978617e-06, + "loss": 0.7123, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0003338354508978617, + "step": 3996 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3896, + "image_masks_loss": 0.0244, + "llm_learning_rate": 3.3375728837832954e-06, + "loss": 0.7752, + "output_text_mask_loss": 0.0744, + "predictor_learning_rate": 0.0003337572883783295, + "step": 3997 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3895, + "image_masks_loss": 0.0277, + "llm_learning_rate": 3.3367911663525833e-06, + "loss": 0.7113, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 0.0003336791166352583, + "step": 3998 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3894, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.336009356772563e-06, + "loss": 0.844, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003336009356772563, + "step": 3999 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3893, + "image_masks_loss": 0.0161, + "llm_learning_rate": 3.3352274551293305e-06, + "loss": 0.7744, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00033352274551293307, + "step": 4000 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3892, + "image_masks_loss": 0.0188, + "llm_learning_rate": 3.3344454615089913e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.0315, + "predictor_learning_rate": 0.0003334445461508991, + "step": 4001 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3891, + "image_masks_loss": 0.038, + "llm_learning_rate": 3.3336633759976585e-06, + "loss": 0.7135, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.00033336633759976583, + "step": 4002 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.389, + "image_masks_loss": 0.0448, + "llm_learning_rate": 3.332881198681458e-06, + "loss": 0.7091, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003332881198681458, + "step": 4003 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3889, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.3320989296465246e-06, + "loss": 0.7904, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 0.0003332098929646524, + "step": 4004 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3888, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.3313165689790034e-06, + "loss": 0.7108, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 0.0003331316568979003, + "step": 4005 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3887, + "image_masks_loss": 0.0273, + "llm_learning_rate": 3.33053411676505e-06, + "loss": 0.7751, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.000333053411676505, + "step": 4006 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3886, + "image_masks_loss": 0.0183, + "llm_learning_rate": 3.3297515730908297e-06, + "loss": 0.7153, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.00033297515730908297, + "step": 4007 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3885, + "image_masks_loss": 0.0051, + "llm_learning_rate": 3.3289689380425185e-06, + "loss": 0.8115, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0003328968938042518, + "step": 4008 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3885, + "image_masks_loss": 0.0142, + "llm_learning_rate": 3.3281862117063008e-06, + "loss": 0.705, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00033281862117063004, + "step": 4009 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3884, + "image_masks_loss": 0.042, + "llm_learning_rate": 3.3274033941683727e-06, + "loss": 0.699, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00033274033941683724, + "step": 4010 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3883, + "image_masks_loss": 0.0213, + "llm_learning_rate": 3.3266204855149403e-06, + "loss": 0.7126, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000332662048551494, + "step": 4011 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3882, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.3258374858322183e-06, + "loss": 0.8065, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.00033258374858322184, + "step": 4012 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3881, + "image_masks_loss": 0.0238, + "llm_learning_rate": 3.3250543952064334e-06, + "loss": 0.7853, + "output_text_mask_loss": 0.032, + "predictor_learning_rate": 0.0003325054395206433, + "step": 4013 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.388, + "image_masks_loss": 0.0406, + "llm_learning_rate": 3.3242712137238207e-06, + "loss": 0.6075, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003324271213723821, + "step": 4014 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3879, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.323487941470626e-06, + "loss": 0.7898, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003323487941470626, + "step": 4015 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3878, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.3227045785331046e-06, + "loss": 0.7321, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00033227045785331046, + "step": 4016 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3877, + "image_masks_loss": 0.0196, + "llm_learning_rate": 3.3219211249975226e-06, + "loss": 0.7888, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0003321921124997522, + "step": 4017 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3876, + "image_masks_loss": 0.0252, + "llm_learning_rate": 3.3211375809501556e-06, + "loss": 0.8208, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.00033211375809501557, + "step": 4018 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3875, + "image_masks_loss": 0.0328, + "llm_learning_rate": 3.3203539464772895e-06, + "loss": 0.7738, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00033203539464772894, + "step": 4019 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3875, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.319570221665219e-06, + "loss": 0.6955, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00033195702216652184, + "step": 4020 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3874, + "image_masks_loss": 0.0083, + "llm_learning_rate": 3.3187864066002507e-06, + "loss": 0.7739, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00033187864066002505, + "step": 4021 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3873, + "image_masks_loss": 0.0254, + "llm_learning_rate": 3.3180025013686994e-06, + "loss": 0.8789, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003318002501368699, + "step": 4022 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3872, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.3172185060568903e-06, + "loss": 0.782, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.000331721850605689, + "step": 4023 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3871, + "image_masks_loss": 0.0156, + "llm_learning_rate": 3.316434420751159e-06, + "loss": 0.8132, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003316434420751159, + "step": 4024 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.387, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.3156502455378512e-06, + "loss": 0.8241, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00033156502455378514, + "step": 4025 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3869, + "image_masks_loss": 0.022, + "llm_learning_rate": 3.314865980503321e-06, + "loss": 0.8327, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0003314865980503321, + "step": 4026 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3868, + "image_masks_loss": 0.023, + "llm_learning_rate": 3.314081625733935e-06, + "loss": 0.7222, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00033140816257339344, + "step": 4027 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3867, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.3132971813160656e-06, + "loss": 0.7795, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003313297181316066, + "step": 4028 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3866, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.3125126473361004e-06, + "loss": 0.7676, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.00033125126473361, + "step": 4029 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3865, + "image_masks_loss": 0.0241, + "llm_learning_rate": 3.3117280238804316e-06, + "loss": 0.701, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00033117280238804313, + "step": 4030 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3864, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.310943311035466e-06, + "loss": 0.6908, + "output_text_mask_loss": 0.1389, + "predictor_learning_rate": 0.0003310943311035466, + "step": 4031 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3864, + "image_masks_loss": 0.0317, + "llm_learning_rate": 3.3101585088876155e-06, + "loss": 0.6691, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.0003310158508887615, + "step": 4032 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3863, + "image_masks_loss": 0.0158, + "llm_learning_rate": 3.3093736175233058e-06, + "loss": 0.7072, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00033093736175233057, + "step": 4033 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3862, + "image_masks_loss": 0.0359, + "llm_learning_rate": 3.3085886370289712e-06, + "loss": 0.851, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003308588637028971, + "step": 4034 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3861, + "image_masks_loss": 0.0142, + "llm_learning_rate": 3.307803567491054e-06, + "loss": 0.639, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003307803567491054, + "step": 4035 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.386, + "image_masks_loss": 0.0052, + "llm_learning_rate": 3.3070184089960084e-06, + "loss": 0.7966, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003307018408996008, + "step": 4036 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3859, + "image_masks_loss": 0.0187, + "llm_learning_rate": 3.306233161630299e-06, + "loss": 0.7828, + "output_text_mask_loss": 0.004, + "predictor_learning_rate": 0.0003306233161630299, + "step": 4037 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3858, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.305447825480397e-06, + "loss": 0.7276, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003305447825480397, + "step": 4038 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3857, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.3046624006327873e-06, + "loss": 0.8302, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0003304662400632787, + "step": 4039 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3856, + "image_masks_loss": 0.0252, + "llm_learning_rate": 3.303876887173962e-06, + "loss": 0.6973, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003303876887173962, + "step": 4040 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3855, + "image_masks_loss": 0.0184, + "llm_learning_rate": 3.303091285190422e-06, + "loss": 0.7775, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00033030912851904224, + "step": 4041 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3854, + "image_masks_loss": 0.0123, + "llm_learning_rate": 3.3023055947686814e-06, + "loss": 0.7657, + "output_text_mask_loss": 0.0334, + "predictor_learning_rate": 0.00033023055947686814, + "step": 4042 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3854, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.301519815995262e-06, + "loss": 0.6783, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003301519815995262, + "step": 4043 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3853, + "image_masks_loss": 0.0437, + "llm_learning_rate": 3.3007339489566953e-06, + "loss": 0.7504, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0003300733948956695, + "step": 4044 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3852, + "image_masks_loss": 0.0208, + "llm_learning_rate": 3.299947993739522e-06, + "loss": 0.723, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032999479937395213, + "step": 4045 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3851, + "image_masks_loss": 0.0425, + "llm_learning_rate": 3.299161950430294e-06, + "loss": 0.8118, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0003299161950430294, + "step": 4046 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.385, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.298375819115572e-06, + "loss": 0.6151, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032983758191155717, + "step": 4047 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3849, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.297589599881926e-06, + "loss": 0.798, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0003297589599881926, + "step": 4048 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3848, + "image_masks_loss": 0.0158, + "llm_learning_rate": 3.296803292815937e-06, + "loss": 0.734, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.0003296803292815937, + "step": 4049 + }, + { + "epoch": 0.41, + "gumbel_tau": 0.3847, + "image_masks_loss": 0.0178, + "llm_learning_rate": 3.2960168980041957e-06, + "loss": 0.6988, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00032960168980041953, + "step": 4050 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3846, + "image_masks_loss": 0.0256, + "llm_learning_rate": 3.2952304155332993e-06, + "loss": 0.7232, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0003295230415533299, + "step": 4051 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3845, + "image_masks_loss": 0.015, + "llm_learning_rate": 3.2944438454898585e-06, + "loss": 0.7335, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00032944438454898583, + "step": 4052 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3844, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.293657187960492e-06, + "loss": 0.773, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032936571879604913, + "step": 4053 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3844, + "image_masks_loss": 0.0206, + "llm_learning_rate": 3.292870443031828e-06, + "loss": 0.8794, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0003292870443031828, + "step": 4054 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3843, + "image_masks_loss": 0.0238, + "llm_learning_rate": 3.2920836107905048e-06, + "loss": 0.7421, + "output_text_mask_loss": 0.069, + "predictor_learning_rate": 0.00032920836107905044, + "step": 4055 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3842, + "image_masks_loss": 0.0271, + "llm_learning_rate": 3.29129669132317e-06, + "loss": 0.8874, + "output_text_mask_loss": 0.0793, + "predictor_learning_rate": 0.00032912966913231694, + "step": 4056 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3841, + "image_masks_loss": 0.0132, + "llm_learning_rate": 3.2905096847164807e-06, + "loss": 0.8603, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.00032905096847164804, + "step": 4057 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.384, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.289722591057105e-06, + "loss": 0.7081, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0003289722591057105, + "step": 4058 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3839, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.288935410431718e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.0003288935410431718, + "step": 4059 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3838, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.2881481429270067e-06, + "loss": 0.7536, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.0003288148142927007, + "step": 4060 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3837, + "image_masks_loss": 0.0212, + "llm_learning_rate": 3.287360788629666e-06, + "loss": 0.7421, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003287360788629666, + "step": 4061 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3836, + "image_masks_loss": 0.0179, + "llm_learning_rate": 3.2865733476264016e-06, + "loss": 0.66, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032865733476264013, + "step": 4062 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3835, + "image_masks_loss": 0.0332, + "llm_learning_rate": 3.285785820003928e-06, + "loss": 0.834, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00032857858200039275, + "step": 4063 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3835, + "image_masks_loss": 0.0211, + "llm_learning_rate": 3.2849982058489697e-06, + "loss": 0.7382, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032849982058489696, + "step": 4064 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3834, + "image_masks_loss": 0.0473, + "llm_learning_rate": 3.28421050524826e-06, + "loss": 0.7885, + "output_text_mask_loss": 0.0018, + "predictor_learning_rate": 0.00032842105052482593, + "step": 4065 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3833, + "image_masks_loss": 0.0168, + "llm_learning_rate": 3.2834227182885437e-06, + "loss": 0.7999, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00032834227182885437, + "step": 4066 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3832, + "image_masks_loss": 0.004, + "llm_learning_rate": 3.282634845056572e-06, + "loss": 0.7786, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003282634845056572, + "step": 4067 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3831, + "image_masks_loss": 0.0119, + "llm_learning_rate": 3.281846885639108e-06, + "loss": 0.8075, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00032818468856391083, + "step": 4068 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.383, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.2810588401229235e-06, + "loss": 0.7797, + "output_text_mask_loss": 0.0977, + "predictor_learning_rate": 0.0003281058840122924, + "step": 4069 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3829, + "image_masks_loss": 0.0179, + "llm_learning_rate": 3.2802707085948e-06, + "loss": 0.7445, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00032802707085947995, + "step": 4070 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3828, + "image_masks_loss": 0.0212, + "llm_learning_rate": 3.2794824911415276e-06, + "loss": 0.8472, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003279482491141527, + "step": 4071 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3827, + "image_masks_loss": 0.028, + "llm_learning_rate": 3.278694187849907e-06, + "loss": 0.7035, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00032786941878499074, + "step": 4072 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3826, + "image_masks_loss": 0.0263, + "llm_learning_rate": 3.2779057988067485e-06, + "loss": 0.6627, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032779057988067484, + "step": 4073 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3825, + "image_masks_loss": 0.0364, + "llm_learning_rate": 3.277117324098871e-06, + "loss": 0.7931, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00032771173240988703, + "step": 4074 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3825, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.2763287638131014e-06, + "loss": 0.7871, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032763287638131014, + "step": 4075 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3824, + "image_masks_loss": 0.0182, + "llm_learning_rate": 3.2755401180362804e-06, + "loss": 0.8087, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 0.00032755401180362804, + "step": 4076 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3823, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.274751386855254e-06, + "loss": 0.8873, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0003274751386855254, + "step": 4077 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3822, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.273962570356879e-06, + "loss": 0.7209, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032739625703568785, + "step": 4078 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3821, + "image_masks_loss": 0.0087, + "llm_learning_rate": 3.273173668628022e-06, + "loss": 0.7616, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032731736686280216, + "step": 4079 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.382, + "image_masks_loss": 0.0242, + "llm_learning_rate": 3.272384681755558e-06, + "loss": 0.8182, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.0003272384681755558, + "step": 4080 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3819, + "image_masks_loss": 0.0364, + "llm_learning_rate": 3.2715956098263723e-06, + "loss": 0.7136, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00032715956098263723, + "step": 4081 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3818, + "image_masks_loss": 0.0179, + "llm_learning_rate": 3.27080645292736e-06, + "loss": 0.8113, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00032708064529273596, + "step": 4082 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3817, + "image_masks_loss": 0.0083, + "llm_learning_rate": 3.2700172111454244e-06, + "loss": 0.7618, + "output_text_mask_loss": 0.0774, + "predictor_learning_rate": 0.0003270017211145424, + "step": 4083 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3816, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.2692278845674772e-06, + "loss": 0.7581, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003269227884567477, + "step": 4084 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3816, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.2684384732804425e-06, + "loss": 0.7661, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0003268438473280442, + "step": 4085 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3815, + "image_masks_loss": 0.0101, + "llm_learning_rate": 3.2676489773712515e-06, + "loss": 0.8268, + "output_text_mask_loss": 0.0033, + "predictor_learning_rate": 0.00032676489773712515, + "step": 4086 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3814, + "image_masks_loss": 0.0083, + "llm_learning_rate": 3.266859396926846e-06, + "loss": 0.6408, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.00032668593969268457, + "step": 4087 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3813, + "image_masks_loss": 0.0047, + "llm_learning_rate": 3.266069732034175e-06, + "loss": 0.6802, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.0003266069732034175, + "step": 4088 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3812, + "image_masks_loss": 0.0223, + "llm_learning_rate": 3.265279982780199e-06, + "loss": 0.7712, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003265279982780199, + "step": 4089 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3811, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.2644901492518864e-06, + "loss": 0.7638, + "output_text_mask_loss": 0.161, + "predictor_learning_rate": 0.0003264490149251886, + "step": 4090 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.381, + "image_masks_loss": 0.0146, + "llm_learning_rate": 3.2637002315362153e-06, + "loss": 0.7272, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003263700231536215, + "step": 4091 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3809, + "image_masks_loss": 0.0039, + "llm_learning_rate": 3.2629102297201744e-06, + "loss": 0.7501, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003262910229720174, + "step": 4092 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3808, + "image_masks_loss": 0.011, + "llm_learning_rate": 3.2621201438907597e-06, + "loss": 0.7695, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.00032621201438907596, + "step": 4093 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3807, + "image_masks_loss": 0.0216, + "llm_learning_rate": 3.2613299741349765e-06, + "loss": 0.714, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00032613299741349763, + "step": 4094 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3807, + "image_masks_loss": 0.0189, + "llm_learning_rate": 3.260539720539841e-06, + "loss": 0.7717, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00032605397205398405, + "step": 4095 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3806, + "image_masks_loss": 0.0185, + "llm_learning_rate": 3.259749383192378e-06, + "loss": 0.724, + "output_text_mask_loss": 0.0402, + "predictor_learning_rate": 0.0003259749383192378, + "step": 4096 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3805, + "image_masks_loss": 0.0307, + "llm_learning_rate": 3.25895896217962e-06, + "loss": 0.8119, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000325895896217962, + "step": 4097 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3804, + "image_masks_loss": 0.0087, + "llm_learning_rate": 3.2581684575886106e-06, + "loss": 0.6557, + "output_text_mask_loss": 0.087, + "predictor_learning_rate": 0.00032581684575886104, + "step": 4098 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3803, + "image_masks_loss": 0.0307, + "llm_learning_rate": 3.2573778695064025e-06, + "loss": 0.771, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.0003257377869506402, + "step": 4099 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3802, + "image_masks_loss": 0.0252, + "llm_learning_rate": 3.256587198020056e-06, + "loss": 0.6952, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003256587198020056, + "step": 4100 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3801, + "image_masks_loss": 0.0171, + "llm_learning_rate": 3.2557964432166415e-06, + "loss": 0.7696, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00032557964432166416, + "step": 4101 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.38, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.25500560518324e-06, + "loss": 0.725, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032550056051832397, + "step": 4102 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3799, + "image_masks_loss": 0.0163, + "llm_learning_rate": 3.2542146840069393e-06, + "loss": 0.774, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00032542146840069394, + "step": 4103 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3798, + "image_masks_loss": 0.0282, + "llm_learning_rate": 3.2534236797748374e-06, + "loss": 0.7858, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003253423679774837, + "step": 4104 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3798, + "image_masks_loss": 0.0187, + "llm_learning_rate": 3.2526325925740428e-06, + "loss": 0.6989, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.00032526325925740423, + "step": 4105 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3797, + "image_masks_loss": 0.0124, + "llm_learning_rate": 3.2518414224916696e-06, + "loss": 0.7133, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00032518414224916694, + "step": 4106 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3796, + "image_masks_loss": 0.0174, + "llm_learning_rate": 3.251050169614844e-06, + "loss": 0.8092, + "output_text_mask_loss": 0.0011, + "predictor_learning_rate": 0.00032510501696148437, + "step": 4107 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3795, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.2502588340307016e-06, + "loss": 0.8585, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00032502588340307015, + "step": 4108 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3794, + "image_masks_loss": 0.0208, + "llm_learning_rate": 3.249467415826385e-06, + "loss": 0.7436, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0003249467415826385, + "step": 4109 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3793, + "image_masks_loss": 0.011, + "llm_learning_rate": 3.248675915089047e-06, + "loss": 0.7927, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0003248675915089047, + "step": 4110 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3792, + "image_masks_loss": 0.0353, + "llm_learning_rate": 3.2478843319058495e-06, + "loss": 0.7539, + "output_text_mask_loss": 0.0423, + "predictor_learning_rate": 0.0003247884331905849, + "step": 4111 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3791, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.247092666363964e-06, + "loss": 0.684, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0003247092666363964, + "step": 4112 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.379, + "image_masks_loss": 0.0069, + "llm_learning_rate": 3.24630091855057e-06, + "loss": 0.718, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.000324630091855057, + "step": 4113 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.379, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.2455090885528563e-06, + "loss": 0.7381, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0003245509088552856, + "step": 4114 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3789, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.2447171764580217e-06, + "loss": 0.7205, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0003244717176458021, + "step": 4115 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3788, + "image_masks_loss": 0.0102, + "llm_learning_rate": 3.2439251823532725e-06, + "loss": 0.7248, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003243925182353272, + "step": 4116 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3787, + "image_masks_loss": 0.0296, + "llm_learning_rate": 3.2431331063258243e-06, + "loss": 0.7467, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0003243133106325824, + "step": 4117 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3786, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.242340948462904e-06, + "loss": 0.8015, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0003242340948462904, + "step": 4118 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3785, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.2415487088517457e-06, + "loss": 0.7583, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.0003241548708851745, + "step": 4119 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3784, + "image_masks_loss": 0.0256, + "llm_learning_rate": 3.24075638757959e-06, + "loss": 0.8019, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00032407563875795896, + "step": 4120 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3783, + "image_masks_loss": 0.0171, + "llm_learning_rate": 3.2399639847336924e-06, + "loss": 0.7036, + "output_text_mask_loss": 0.0196, + "predictor_learning_rate": 0.0003239963984733692, + "step": 4121 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3782, + "image_masks_loss": 0.0208, + "llm_learning_rate": 3.2391715004013126e-06, + "loss": 0.7604, + "output_text_mask_loss": 0.0144, + "predictor_learning_rate": 0.0003239171500401312, + "step": 4122 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3781, + "image_masks_loss": 0.0098, + "llm_learning_rate": 3.2383789346697204e-06, + "loss": 0.8525, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 0.000323837893466972, + "step": 4123 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3781, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.237586287626195e-06, + "loss": 0.6503, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00032375862876261944, + "step": 4124 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.378, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.236793559358025e-06, + "loss": 0.8363, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003236793559358024, + "step": 4125 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3779, + "image_masks_loss": 0.013, + "llm_learning_rate": 3.236000749952507e-06, + "loss": 0.8325, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003236000749952507, + "step": 4126 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3778, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.235207859496947e-06, + "loss": 0.7111, + "output_text_mask_loss": 0.09, + "predictor_learning_rate": 0.00032352078594969464, + "step": 4127 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3777, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.23441488807866e-06, + "loss": 0.762, + "output_text_mask_loss": 0.0816, + "predictor_learning_rate": 0.0003234414888078659, + "step": 4128 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3776, + "image_masks_loss": 0.029, + "llm_learning_rate": 3.233621835784969e-06, + "loss": 0.8444, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.0003233621835784969, + "step": 4129 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3775, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.232828702703208e-06, + "loss": 0.7063, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003232828702703208, + "step": 4130 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3774, + "image_masks_loss": 0.0157, + "llm_learning_rate": 3.2320354889207177e-06, + "loss": 0.719, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00032320354889207176, + "step": 4131 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3773, + "image_masks_loss": 0.0128, + "llm_learning_rate": 3.2312421945248496e-06, + "loss": 0.7719, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.00032312421945248493, + "step": 4132 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3773, + "image_masks_loss": 0.0074, + "llm_learning_rate": 3.2304488196029614e-06, + "loss": 0.7683, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0003230448819602961, + "step": 4133 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3772, + "image_masks_loss": 0.0071, + "llm_learning_rate": 3.2296553642424226e-06, + "loss": 0.7349, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00032296553642424225, + "step": 4134 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3771, + "image_masks_loss": 0.0389, + "llm_learning_rate": 3.2288618285306095e-06, + "loss": 0.8412, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.0003228861828530609, + "step": 4135 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.377, + "image_masks_loss": 0.0216, + "llm_learning_rate": 3.2280682125549086e-06, + "loss": 0.7178, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.00032280682125549083, + "step": 4136 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3769, + "image_masks_loss": 0.0218, + "llm_learning_rate": 3.227274516402714e-06, + "loss": 0.7527, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.0003227274516402714, + "step": 4137 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3768, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.2264807401614305e-06, + "loss": 0.7184, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.000322648074016143, + "step": 4138 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3767, + "image_masks_loss": 0.0176, + "llm_learning_rate": 3.2256868839184694e-06, + "loss": 0.717, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.00032256868839184696, + "step": 4139 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3766, + "image_masks_loss": 0.0215, + "llm_learning_rate": 3.2248929477612516e-06, + "loss": 0.6872, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00032248929477612514, + "step": 4140 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3765, + "image_masks_loss": 0.0168, + "llm_learning_rate": 3.224098931777209e-06, + "loss": 0.6804, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032240989317772087, + "step": 4141 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3765, + "image_masks_loss": 0.0286, + "llm_learning_rate": 3.223304836053779e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00032233048360537787, + "step": 4142 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3764, + "image_masks_loss": 0.0132, + "llm_learning_rate": 3.222510660678408e-06, + "loss": 0.6928, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003222510660678408, + "step": 4143 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3763, + "image_masks_loss": 0.0099, + "llm_learning_rate": 3.221716405738555e-06, + "loss": 0.7124, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00032217164057385546, + "step": 4144 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3762, + "image_masks_loss": 0.0105, + "llm_learning_rate": 3.2209220713216833e-06, + "loss": 0.7578, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003220922071321683, + "step": 4145 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3761, + "image_masks_loss": 0.0105, + "llm_learning_rate": 3.220127657515267e-06, + "loss": 0.764, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00032201276575152663, + "step": 4146 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.376, + "image_masks_loss": 0.0055, + "llm_learning_rate": 3.2193331644067895e-06, + "loss": 0.7011, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00032193331644067894, + "step": 4147 + }, + { + "epoch": 0.42, + "gumbel_tau": 0.3759, + "image_masks_loss": 0.0353, + "llm_learning_rate": 3.2185385920837415e-06, + "loss": 0.8934, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.00032185385920837414, + "step": 4148 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3758, + "image_masks_loss": 0.0515, + "llm_learning_rate": 3.2177439406336237e-06, + "loss": 0.8111, + "output_text_mask_loss": 0.0359, + "predictor_learning_rate": 0.00032177439406336233, + "step": 4149 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3757, + "image_masks_loss": 0.0056, + "llm_learning_rate": 3.2169492101439436e-06, + "loss": 0.7129, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00032169492101439434, + "step": 4150 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3757, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.21615440070222e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.000321615440070222, + "step": 4151 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3756, + "image_masks_loss": 0.0484, + "llm_learning_rate": 3.2153595123959793e-06, + "loss": 0.7745, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0003215359512395979, + "step": 4152 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3755, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.2145645453127548e-06, + "loss": 0.846, + "output_text_mask_loss": 0.1015, + "predictor_learning_rate": 0.0003214564545312755, + "step": 4153 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3754, + "image_masks_loss": 0.0313, + "llm_learning_rate": 3.2137694995400915e-06, + "loss": 0.7665, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0003213769499540091, + "step": 4154 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3753, + "image_masks_loss": 0.0166, + "llm_learning_rate": 3.2129743751655406e-06, + "loss": 0.5189, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032129743751655403, + "step": 4155 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3752, + "image_masks_loss": 0.0233, + "llm_learning_rate": 3.212179172276663e-06, + "loss": 0.7979, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003212179172276663, + "step": 4156 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3751, + "image_masks_loss": 0.0163, + "llm_learning_rate": 3.2113838909610297e-06, + "loss": 0.7669, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.00032113838909610294, + "step": 4157 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.375, + "image_masks_loss": 0.0203, + "llm_learning_rate": 3.2105885313062175e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.0584, + "predictor_learning_rate": 0.00032105885313062175, + "step": 4158 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.375, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.2097930933998127e-06, + "loss": 0.6347, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032097930933998125, + "step": 4159 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3749, + "image_masks_loss": 0.0167, + "llm_learning_rate": 3.208997577329412e-06, + "loss": 0.7236, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00032089975773294115, + "step": 4160 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3748, + "image_masks_loss": 0.0298, + "llm_learning_rate": 3.2082019831826196e-06, + "loss": 0.7598, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0003208201983182619, + "step": 4161 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3747, + "image_masks_loss": 0.0087, + "llm_learning_rate": 3.207406311047046e-06, + "loss": 0.7453, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003207406311047046, + "step": 4162 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3746, + "image_masks_loss": 0.0243, + "llm_learning_rate": 3.2066105610103148e-06, + "loss": 0.7099, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00032066105610103145, + "step": 4163 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3745, + "image_masks_loss": 0.0275, + "llm_learning_rate": 3.2058147331600543e-06, + "loss": 0.7576, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0003205814733160054, + "step": 4164 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3744, + "image_masks_loss": 0.0353, + "llm_learning_rate": 3.2050188275839035e-06, + "loss": 0.7942, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003205018827583903, + "step": 4165 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3743, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.204222844369508e-06, + "loss": 0.7444, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003204222844369508, + "step": 4166 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3742, + "image_masks_loss": 0.0136, + "llm_learning_rate": 3.2034267836045253e-06, + "loss": 0.762, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0003203426783604525, + "step": 4167 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3742, + "image_masks_loss": 0.012, + "llm_learning_rate": 3.202630645376618e-06, + "loss": 0.8061, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.0003202630645376618, + "step": 4168 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3741, + "image_masks_loss": 0.0139, + "llm_learning_rate": 3.2018344297734587e-06, + "loss": 0.744, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.00032018344297734587, + "step": 4169 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.374, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.2010381368827297e-06, + "loss": 0.7068, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000320103813688273, + "step": 4170 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3739, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.2002417667921193e-06, + "loss": 0.8267, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0003200241766792119, + "step": 4171 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3738, + "image_masks_loss": 0.0223, + "llm_learning_rate": 3.199445319589325e-06, + "loss": 0.7852, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003199445319589325, + "step": 4172 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3737, + "image_masks_loss": 0.0174, + "llm_learning_rate": 3.1986487953620547e-06, + "loss": 0.6973, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0003198648795362055, + "step": 4173 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3736, + "image_masks_loss": 0.0395, + "llm_learning_rate": 3.1978521941980227e-06, + "loss": 0.7784, + "output_text_mask_loss": 0.0425, + "predictor_learning_rate": 0.00031978521941980226, + "step": 4174 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3735, + "image_masks_loss": 0.014, + "llm_learning_rate": 3.1970555161849525e-06, + "loss": 0.7253, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00031970555161849523, + "step": 4175 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3735, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.196258761410577e-06, + "loss": 0.8016, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031962587614105763, + "step": 4176 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3734, + "image_masks_loss": 0.033, + "llm_learning_rate": 3.1954619299626347e-06, + "loss": 0.7906, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0003195461929962635, + "step": 4177 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3733, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.194665021928877e-06, + "loss": 0.8298, + "output_text_mask_loss": 0.1312, + "predictor_learning_rate": 0.00031946650219288766, + "step": 4178 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3732, + "image_masks_loss": 0.0207, + "llm_learning_rate": 3.1938680373970586e-06, + "loss": 0.7939, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003193868037397059, + "step": 4179 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3731, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.193070976454947e-06, + "loss": 0.8148, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.00031930709764549473, + "step": 4180 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.373, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.1922738391903156e-06, + "loss": 0.7706, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 0.00031922738391903155, + "step": 4181 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3729, + "image_masks_loss": 0.0324, + "llm_learning_rate": 3.191476625690947e-06, + "loss": 0.7753, + "output_text_mask_loss": 0.0648, + "predictor_learning_rate": 0.0003191476625690947, + "step": 4182 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3728, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.190679336044632e-06, + "loss": 0.6874, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0003190679336044632, + "step": 4183 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3727, + "image_masks_loss": 0.0226, + "llm_learning_rate": 3.1898819703391704e-06, + "loss": 0.7049, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.000318988197033917, + "step": 4184 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3727, + "image_masks_loss": 0.0136, + "llm_learning_rate": 3.189084528662369e-06, + "loss": 0.82, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0003189084528662369, + "step": 4185 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3726, + "image_masks_loss": 0.0052, + "llm_learning_rate": 3.1882870111020446e-06, + "loss": 0.7919, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00031882870111020446, + "step": 4186 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3725, + "image_masks_loss": 0.0218, + "llm_learning_rate": 3.1874894177460217e-06, + "loss": 0.8143, + "output_text_mask_loss": 0.0736, + "predictor_learning_rate": 0.0003187489417746022, + "step": 4187 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3724, + "image_masks_loss": 0.0214, + "llm_learning_rate": 3.186691748682133e-06, + "loss": 0.5729, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003186691748682133, + "step": 4188 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3723, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.1858940039982197e-06, + "loss": 0.7653, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031858940039982195, + "step": 4189 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3722, + "image_masks_loss": 0.0214, + "llm_learning_rate": 3.1850961837821303e-06, + "loss": 0.7582, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.000318509618378213, + "step": 4190 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3721, + "image_masks_loss": 0.0181, + "llm_learning_rate": 3.1842982881217227e-06, + "loss": 0.7744, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003184298288121722, + "step": 4191 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.372, + "image_masks_loss": 0.019, + "llm_learning_rate": 3.183500317104864e-06, + "loss": 0.812, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00031835003171048635, + "step": 4192 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.372, + "image_masks_loss": 0.0042, + "llm_learning_rate": 3.1827022708194277e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0523, + "predictor_learning_rate": 0.00031827022708194277, + "step": 4193 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3719, + "image_masks_loss": 0.0109, + "llm_learning_rate": 3.1819041493532975e-06, + "loss": 0.7285, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.00031819041493532974, + "step": 4194 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3718, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.1811059527943622e-06, + "loss": 0.7916, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0003181105952794362, + "step": 4195 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3717, + "image_masks_loss": 0.0078, + "llm_learning_rate": 3.1803076812305235e-06, + "loss": 0.8759, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.0003180307681230523, + "step": 4196 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3716, + "image_masks_loss": 0.0163, + "llm_learning_rate": 3.1795093347496864e-06, + "loss": 0.7923, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.0003179509334749686, + "step": 4197 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3715, + "image_masks_loss": 0.0189, + "llm_learning_rate": 3.1787109134397687e-06, + "loss": 0.7538, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.0003178710913439768, + "step": 4198 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3714, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.1779124173886934e-06, + "loss": 0.7777, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.0003177912417388693, + "step": 4199 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3713, + "image_masks_loss": 0.0298, + "llm_learning_rate": 3.1771138466843925e-06, + "loss": 0.6173, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003177113846684392, + "step": 4200 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3713, + "image_masks_loss": 0.0393, + "llm_learning_rate": 3.1763152014148064e-06, + "loss": 0.7315, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00031763152014148064, + "step": 4201 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3712, + "image_masks_loss": 0.0236, + "llm_learning_rate": 3.175516481667884e-06, + "loss": 0.7975, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.00031755164816678837, + "step": 4202 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3711, + "image_masks_loss": 0.0259, + "llm_learning_rate": 3.174717687531582e-06, + "loss": 0.7327, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0003174717687531582, + "step": 4203 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.371, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.1739188190938656e-06, + "loss": 0.8015, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.00031739188190938654, + "step": 4204 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3709, + "image_masks_loss": 0.0194, + "llm_learning_rate": 3.1731198764427075e-06, + "loss": 0.7324, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.0003173119876442707, + "step": 4205 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3708, + "image_masks_loss": 0.0235, + "llm_learning_rate": 3.1723208596660894e-06, + "loss": 0.7982, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003172320859666089, + "step": 4206 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3707, + "image_masks_loss": 0.0225, + "llm_learning_rate": 3.1715217688520006e-06, + "loss": 0.6514, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003171521768852, + "step": 4207 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3706, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.1707226040884385e-06, + "loss": 0.8324, + "output_text_mask_loss": 0.0125, + "predictor_learning_rate": 0.00031707226040884385, + "step": 4208 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3706, + "image_masks_loss": 0.0193, + "llm_learning_rate": 3.16992336546341e-06, + "loss": 0.7028, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.00031699233654634097, + "step": 4209 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3705, + "image_masks_loss": 0.0211, + "llm_learning_rate": 3.1691240530649282e-06, + "loss": 0.8469, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.00031691240530649277, + "step": 4210 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3704, + "image_masks_loss": 0.0089, + "llm_learning_rate": 3.1683246669810146e-06, + "loss": 0.7661, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.00031683246669810146, + "step": 4211 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3703, + "image_masks_loss": 0.0054, + "llm_learning_rate": 3.167525207299701e-06, + "loss": 0.7208, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.00031675252072997005, + "step": 4212 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3702, + "image_masks_loss": 0.0191, + "llm_learning_rate": 3.166725674109024e-06, + "loss": 0.8633, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.0003166725674109024, + "step": 4213 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3701, + "image_masks_loss": 0.0163, + "llm_learning_rate": 3.1659260674970303e-06, + "loss": 0.7412, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00031659260674970303, + "step": 4214 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.37, + "image_masks_loss": 0.0137, + "llm_learning_rate": 3.1651263875517756e-06, + "loss": 0.7125, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00031651263875517757, + "step": 4215 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3699, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.164326634361322e-06, + "loss": 0.8138, + "output_text_mask_loss": 0.0729, + "predictor_learning_rate": 0.0003164326634361322, + "step": 4216 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3699, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.163526808013738e-06, + "loss": 0.8264, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 0.0003163526808013738, + "step": 4217 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3698, + "image_masks_loss": 0.0278, + "llm_learning_rate": 3.1627269085971057e-06, + "loss": 0.8216, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031627269085971057, + "step": 4218 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3697, + "image_masks_loss": 0.0107, + "llm_learning_rate": 3.1619269361995093e-06, + "loss": 0.6966, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003161926936199509, + "step": 4219 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3696, + "image_masks_loss": 0.0213, + "llm_learning_rate": 3.1611268909090443e-06, + "loss": 0.7643, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003161126890909044, + "step": 4220 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3695, + "image_masks_loss": 0.0317, + "llm_learning_rate": 3.1603267728138133e-06, + "loss": 0.7475, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0003160326772813813, + "step": 4221 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3694, + "image_masks_loss": 0.0091, + "llm_learning_rate": 3.159526582001927e-06, + "loss": 0.8026, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0003159526582001927, + "step": 4222 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3693, + "image_masks_loss": 0.0077, + "llm_learning_rate": 3.1587263185615057e-06, + "loss": 0.7746, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031587263185615055, + "step": 4223 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3692, + "image_masks_loss": 0.012, + "llm_learning_rate": 3.1579259825806736e-06, + "loss": 0.7574, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031579259825806734, + "step": 4224 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3692, + "image_masks_loss": 0.0288, + "llm_learning_rate": 3.1571255741475676e-06, + "loss": 0.7073, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003157125574147567, + "step": 4225 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3691, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.15632509335033e-06, + "loss": 0.7524, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00031563250933503296, + "step": 4226 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.369, + "image_masks_loss": 0.0135, + "llm_learning_rate": 3.15552454027711e-06, + "loss": 0.7303, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00031555245402771097, + "step": 4227 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3689, + "image_masks_loss": 0.0221, + "llm_learning_rate": 3.1547239150160687e-06, + "loss": 0.7655, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.0003154723915016068, + "step": 4228 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3688, + "image_masks_loss": 0.0068, + "llm_learning_rate": 3.153923217655371e-06, + "loss": 0.7507, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00031539232176553707, + "step": 4229 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3687, + "image_masks_loss": 0.0234, + "llm_learning_rate": 3.1531224482831924e-06, + "loss": 0.7209, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00031531224482831923, + "step": 4230 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3686, + "image_masks_loss": 0.0233, + "llm_learning_rate": 3.152321606987714e-06, + "loss": 0.7496, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0003152321606987714, + "step": 4231 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3685, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.151520693857128e-06, + "loss": 0.7813, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003151520693857128, + "step": 4232 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3685, + "image_masks_loss": 0.0218, + "llm_learning_rate": 3.1507197089796322e-06, + "loss": 0.8743, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0003150719708979632, + "step": 4233 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3684, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.1499186524434315e-06, + "loss": 0.8088, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00031499186524434317, + "step": 4234 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3683, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.1491175243367427e-06, + "loss": 0.787, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00031491175243367425, + "step": 4235 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3682, + "image_masks_loss": 0.0294, + "llm_learning_rate": 3.1483163247477856e-06, + "loss": 0.8617, + "output_text_mask_loss": 0.0568, + "predictor_learning_rate": 0.0003148316324747785, + "step": 4236 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3681, + "image_masks_loss": 0.0203, + "llm_learning_rate": 3.14751505376479e-06, + "loss": 0.7406, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.000314751505376479, + "step": 4237 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.368, + "image_masks_loss": 0.0183, + "llm_learning_rate": 3.1467137114759954e-06, + "loss": 0.7884, + "output_text_mask_loss": 0.0317, + "predictor_learning_rate": 0.0003146713711475995, + "step": 4238 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3679, + "image_masks_loss": 0.022, + "llm_learning_rate": 3.1459122979696467e-06, + "loss": 0.7805, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.00031459122979696463, + "step": 4239 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3679, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.145110813333996e-06, + "loss": 0.7748, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031451108133339955, + "step": 4240 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3678, + "image_masks_loss": 0.0084, + "llm_learning_rate": 3.1443092576573057e-06, + "loss": 0.7469, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00031443092576573055, + "step": 4241 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3677, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.1435076310278457e-06, + "loss": 0.7197, + "output_text_mask_loss": 0.0196, + "predictor_learning_rate": 0.00031435076310278455, + "step": 4242 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3676, + "image_masks_loss": 0.0232, + "llm_learning_rate": 3.142705933533891e-06, + "loss": 0.7194, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003142705933533891, + "step": 4243 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3675, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.1419041652637285e-06, + "loss": 0.7421, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003141904165263728, + "step": 4244 + }, + { + "epoch": 0.43, + "gumbel_tau": 0.3674, + "image_masks_loss": 0.0126, + "llm_learning_rate": 3.1411023263056485e-06, + "loss": 0.6631, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003141102326305648, + "step": 4245 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3673, + "image_masks_loss": 0.0128, + "llm_learning_rate": 3.140300416747953e-06, + "loss": 0.7676, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.00031403004167479526, + "step": 4246 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3672, + "image_masks_loss": 0.0197, + "llm_learning_rate": 3.139498436678948e-06, + "loss": 0.7517, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003139498436678948, + "step": 4247 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3672, + "image_masks_loss": 0.0032, + "llm_learning_rate": 3.1386963861869524e-06, + "loss": 0.6232, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00031386963861869524, + "step": 4248 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3671, + "image_masks_loss": 0.0098, + "llm_learning_rate": 3.1378942653602875e-06, + "loss": 0.7112, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0003137894265360287, + "step": 4249 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.367, + "image_masks_loss": 0.0216, + "llm_learning_rate": 3.1370920742872846e-06, + "loss": 0.6511, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031370920742872845, + "step": 4250 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3669, + "image_masks_loss": 0.0153, + "llm_learning_rate": 3.1362898130562842e-06, + "loss": 0.6554, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.0003136289813056284, + "step": 4251 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3668, + "image_masks_loss": 0.0172, + "llm_learning_rate": 3.1354874817556323e-06, + "loss": 0.797, + "output_text_mask_loss": 0.0698, + "predictor_learning_rate": 0.00031354874817556325, + "step": 4252 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3667, + "image_masks_loss": 0.0048, + "llm_learning_rate": 3.134685080473683e-06, + "loss": 0.7917, + "output_text_mask_loss": 0.0372, + "predictor_learning_rate": 0.0003134685080473683, + "step": 4253 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3666, + "image_masks_loss": 0.0223, + "llm_learning_rate": 3.1338826092987994e-06, + "loss": 0.7723, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.0003133882609298799, + "step": 4254 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3666, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.1330800683193507e-06, + "loss": 0.791, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00031330800683193507, + "step": 4255 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3665, + "image_masks_loss": 0.0151, + "llm_learning_rate": 3.1322774576237147e-06, + "loss": 0.7055, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00031322774576237147, + "step": 4256 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3664, + "image_masks_loss": 0.0076, + "llm_learning_rate": 3.131474777300277e-06, + "loss": 0.7499, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 0.0003131474777300277, + "step": 4257 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3663, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.13067202743743e-06, + "loss": 0.7831, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000313067202743743, + "step": 4258 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3662, + "image_masks_loss": 0.0083, + "llm_learning_rate": 3.129869208123575e-06, + "loss": 0.6869, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003129869208123575, + "step": 4259 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3661, + "image_masks_loss": 0.0058, + "llm_learning_rate": 3.1290663194471188e-06, + "loss": 0.7105, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00031290663194471184, + "step": 4260 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.366, + "image_masks_loss": 0.0199, + "llm_learning_rate": 3.128263361496479e-06, + "loss": 0.744, + "output_text_mask_loss": 0.0089, + "predictor_learning_rate": 0.0003128263361496479, + "step": 4261 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.366, + "image_masks_loss": 0.0248, + "llm_learning_rate": 3.1274603343600784e-06, + "loss": 0.7307, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031274603343600785, + "step": 4262 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3659, + "image_masks_loss": 0.0064, + "llm_learning_rate": 3.1266572381263483e-06, + "loss": 0.8116, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00031266572381263483, + "step": 4263 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3658, + "image_masks_loss": 0.0095, + "llm_learning_rate": 3.125854072883727e-06, + "loss": 0.8096, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003125854072883727, + "step": 4264 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3657, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.1250508387206617e-06, + "loss": 0.7688, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 0.00031250508387206617, + "step": 4265 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3656, + "image_masks_loss": 0.0211, + "llm_learning_rate": 3.124247535725605e-06, + "loss": 0.7392, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00031242475357256053, + "step": 4266 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3655, + "image_masks_loss": 0.0136, + "llm_learning_rate": 3.12344416398702e-06, + "loss": 0.7253, + "output_text_mask_loss": 0.0033, + "predictor_learning_rate": 0.00031234441639870197, + "step": 4267 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3654, + "image_masks_loss": 0.0067, + "llm_learning_rate": 3.122640723593375e-06, + "loss": 0.7151, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.00031226407235933747, + "step": 4268 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3653, + "image_masks_loss": 0.0198, + "llm_learning_rate": 3.121837214633146e-06, + "loss": 0.7725, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0003121837214633146, + "step": 4269 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3653, + "image_masks_loss": 0.0124, + "llm_learning_rate": 3.121033637194818e-06, + "loss": 0.7504, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00031210336371948177, + "step": 4270 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3652, + "image_masks_loss": 0.01, + "llm_learning_rate": 3.120229991366883e-06, + "loss": 0.7285, + "output_text_mask_loss": 0.0057, + "predictor_learning_rate": 0.0003120229991366882, + "step": 4271 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3651, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.1194262772378393e-06, + "loss": 0.7648, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.0003119426277237839, + "step": 4272 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.365, + "image_masks_loss": 0.0144, + "llm_learning_rate": 3.1186224948961945e-06, + "loss": 0.7165, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 0.00031186224948961944, + "step": 4273 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3649, + "image_masks_loss": 0.0095, + "llm_learning_rate": 3.1178186444304625e-06, + "loss": 0.8297, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00031178186444304624, + "step": 4274 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3648, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.1170147259291654e-06, + "loss": 0.7434, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 0.0003117014725929165, + "step": 4275 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3647, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.1162107394808322e-06, + "loss": 0.6913, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003116210739480832, + "step": 4276 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3647, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.115406685174e-06, + "loss": 0.6446, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003115406685174, + "step": 4277 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3646, + "image_masks_loss": 0.0242, + "llm_learning_rate": 3.1146025630972126e-06, + "loss": 0.653, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0003114602563097212, + "step": 4278 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3645, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.113798373339022e-06, + "loss": 0.7299, + "output_text_mask_loss": 0.0102, + "predictor_learning_rate": 0.00031137983733390217, + "step": 4279 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3644, + "image_masks_loss": 0.0217, + "llm_learning_rate": 3.1129941159879873e-06, + "loss": 0.7027, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031129941159879873, + "step": 4280 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3643, + "image_masks_loss": 0.0139, + "llm_learning_rate": 3.112189791132676e-06, + "loss": 0.7157, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00031121897911326756, + "step": 4281 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3642, + "image_masks_loss": 0.021, + "llm_learning_rate": 3.1113853988616605e-06, + "loss": 0.649, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.000311138539886166, + "step": 4282 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3641, + "image_masks_loss": 0.0177, + "llm_learning_rate": 3.1105809392635233e-06, + "loss": 0.7475, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00031105809392635233, + "step": 4283 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3641, + "image_masks_loss": 0.0189, + "llm_learning_rate": 3.1097764124268528e-06, + "loss": 0.8393, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00031097764124268524, + "step": 4284 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.364, + "image_masks_loss": 0.0146, + "llm_learning_rate": 3.1089718184402456e-06, + "loss": 0.8063, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.00031089718184402457, + "step": 4285 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3639, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.108167157392305e-06, + "loss": 0.685, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00031081671573923047, + "step": 4286 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3638, + "image_masks_loss": 0.0076, + "llm_learning_rate": 3.107362429371643e-06, + "loss": 0.8567, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0003107362429371643, + "step": 4287 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3637, + "image_masks_loss": 0.0187, + "llm_learning_rate": 3.1065576344668772e-06, + "loss": 0.8162, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003106557634466877, + "step": 4288 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3636, + "image_masks_loss": 0.0176, + "llm_learning_rate": 3.1057527727666336e-06, + "loss": 0.9168, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003105752772766633, + "step": 4289 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3635, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.1049478443595456e-06, + "loss": 0.7128, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.00031049478443595453, + "step": 4290 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3635, + "image_masks_loss": 0.0202, + "llm_learning_rate": 3.104142849334254e-06, + "loss": 0.7494, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0003104142849334254, + "step": 4291 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3634, + "image_masks_loss": 0.0077, + "llm_learning_rate": 3.1033377877794054e-06, + "loss": 0.6732, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.0003103337787779405, + "step": 4292 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3633, + "image_masks_loss": 0.0182, + "llm_learning_rate": 3.1025326597836558e-06, + "loss": 0.6172, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031025326597836555, + "step": 4293 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3632, + "image_masks_loss": 0.02, + "llm_learning_rate": 3.101727465435668e-06, + "loss": 0.7013, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00031017274654356676, + "step": 4294 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3631, + "image_masks_loss": 0.0169, + "llm_learning_rate": 3.1009222048241104e-06, + "loss": 0.7179, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000310092220482411, + "step": 4295 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.363, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.100116878037662e-06, + "loss": 0.7137, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00031001168780376615, + "step": 4296 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3629, + "image_masks_loss": 0.0311, + "llm_learning_rate": 3.099311485165006e-06, + "loss": 0.7796, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0003099311485165006, + "step": 4297 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3629, + "image_masks_loss": 0.0181, + "llm_learning_rate": 3.0985060262948345e-06, + "loss": 0.5632, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030985060262948345, + "step": 4298 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3628, + "image_masks_loss": 0.01, + "llm_learning_rate": 3.0977005015158456e-06, + "loss": 0.7172, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030977005015158456, + "step": 4299 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3627, + "image_masks_loss": 0.0228, + "llm_learning_rate": 3.0968949109167475e-06, + "loss": 0.741, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0003096894910916747, + "step": 4300 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3626, + "image_masks_loss": 0.0241, + "llm_learning_rate": 3.0960892545862515e-06, + "loss": 0.7339, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003096089254586251, + "step": 4301 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3625, + "image_masks_loss": 0.0201, + "llm_learning_rate": 3.095283532613079e-06, + "loss": 0.8081, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030952835326130784, + "step": 4302 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3624, + "image_masks_loss": 0.0085, + "llm_learning_rate": 3.0944777450859576e-06, + "loss": 0.799, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00030944777450859573, + "step": 4303 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3623, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.0936718920936236e-06, + "loss": 0.8304, + "output_text_mask_loss": 0.0265, + "predictor_learning_rate": 0.0003093671892093624, + "step": 4304 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3623, + "image_masks_loss": 0.0332, + "llm_learning_rate": 3.0928659737248173e-06, + "loss": 0.7974, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0003092865973724817, + "step": 4305 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3622, + "image_masks_loss": 0.0357, + "llm_learning_rate": 3.0920599900682907e-06, + "loss": 0.7057, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030920599900682903, + "step": 4306 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3621, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.0912539412127984e-06, + "loss": 0.7282, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.0003091253941212798, + "step": 4307 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.362, + "image_masks_loss": 0.0059, + "llm_learning_rate": 3.0904478272471054e-06, + "loss": 0.7441, + "output_text_mask_loss": 0.0257, + "predictor_learning_rate": 0.00030904478272471053, + "step": 4308 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3619, + "image_masks_loss": 0.0053, + "llm_learning_rate": 3.0896416482599827e-06, + "loss": 0.797, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003089641648259983, + "step": 4309 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3618, + "image_masks_loss": 0.0452, + "llm_learning_rate": 3.0888354043402087e-06, + "loss": 0.786, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00030888354043402086, + "step": 4310 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3617, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.0880290955765674e-06, + "loss": 0.5917, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00030880290955765673, + "step": 4311 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3617, + "image_masks_loss": 0.0322, + "llm_learning_rate": 3.0872227220578527e-06, + "loss": 0.8419, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00030872227220578525, + "step": 4312 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3616, + "image_masks_loss": 0.0416, + "llm_learning_rate": 3.086416283872864e-06, + "loss": 0.7479, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003086416283872864, + "step": 4313 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3615, + "image_masks_loss": 0.0175, + "llm_learning_rate": 3.085609781110408e-06, + "loss": 0.8034, + "output_text_mask_loss": 0.0479, + "predictor_learning_rate": 0.00030856097811104077, + "step": 4314 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3614, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.084803213859298e-06, + "loss": 0.6756, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0003084803213859298, + "step": 4315 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3613, + "image_masks_loss": 0.0083, + "llm_learning_rate": 3.083996582208356e-06, + "loss": 0.8208, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00030839965822083553, + "step": 4316 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3612, + "image_masks_loss": 0.0198, + "llm_learning_rate": 3.0831898862464094e-06, + "loss": 0.8201, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0003083189886246409, + "step": 4317 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3611, + "image_masks_loss": 0.0248, + "llm_learning_rate": 3.0823831260622937e-06, + "loss": 0.7264, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0003082383126062294, + "step": 4318 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3611, + "image_masks_loss": 0.0222, + "llm_learning_rate": 3.081576301744851e-06, + "loss": 0.7634, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 0.00030815763017448506, + "step": 4319 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.361, + "image_masks_loss": 0.0075, + "llm_learning_rate": 3.08076941338293e-06, + "loss": 0.7527, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.000308076941338293, + "step": 4320 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3609, + "image_masks_loss": 0.0128, + "llm_learning_rate": 3.079962461065388e-06, + "loss": 0.7707, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030799624610653874, + "step": 4321 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3608, + "image_masks_loss": 0.0112, + "llm_learning_rate": 3.0791554448810877e-06, + "loss": 0.7461, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.00030791554448810876, + "step": 4322 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3607, + "image_masks_loss": 0.0099, + "llm_learning_rate": 3.0783483649189e-06, + "loss": 0.746, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00030783483649188996, + "step": 4323 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3606, + "image_masks_loss": 0.037, + "llm_learning_rate": 3.0775412212677023e-06, + "loss": 0.797, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0003077541221267702, + "step": 4324 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3606, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.0767340140163783e-06, + "loss": 0.66, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0003076734014016378, + "step": 4325 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3605, + "image_masks_loss": 0.0244, + "llm_learning_rate": 3.0759267432538204e-06, + "loss": 0.6802, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000307592674325382, + "step": 4326 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3604, + "image_masks_loss": 0.0039, + "llm_learning_rate": 3.075119409068927e-06, + "loss": 0.7637, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00030751194090689266, + "step": 4327 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3603, + "image_masks_loss": 0.0054, + "llm_learning_rate": 3.074312011550603e-06, + "loss": 0.7561, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00030743120115506026, + "step": 4328 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3602, + "image_masks_loss": 0.0292, + "llm_learning_rate": 3.0735045507877615e-06, + "loss": 0.7636, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0003073504550787761, + "step": 4329 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3601, + "image_masks_loss": 0.0058, + "llm_learning_rate": 3.0726970268693208e-06, + "loss": 0.7633, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00030726970268693207, + "step": 4330 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.36, + "image_masks_loss": 0.0193, + "llm_learning_rate": 3.0718894398842076e-06, + "loss": 0.6796, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00030718894398842076, + "step": 4331 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.36, + "image_masks_loss": 0.0349, + "llm_learning_rate": 3.0710817899213563e-06, + "loss": 0.797, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0003071081789921356, + "step": 4332 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3599, + "image_masks_loss": 0.0143, + "llm_learning_rate": 3.070274077069706e-06, + "loss": 0.6456, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0003070274077069706, + "step": 4333 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3598, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.0694663014182035e-06, + "loss": 0.7148, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00030694663014182034, + "step": 4334 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3597, + "image_masks_loss": 0.0408, + "llm_learning_rate": 3.0686584630558035e-06, + "loss": 0.6977, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003068658463055803, + "step": 4335 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3596, + "image_masks_loss": 0.0303, + "llm_learning_rate": 3.067850562071468e-06, + "loss": 0.6969, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00030678505620714675, + "step": 4336 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3595, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.067042598554163e-06, + "loss": 0.7659, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 0.0003067042598554163, + "step": 4337 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3594, + "image_masks_loss": 0.0343, + "llm_learning_rate": 3.066234572592864e-06, + "loss": 0.7125, + "output_text_mask_loss": 0.0425, + "predictor_learning_rate": 0.00030662345725928643, + "step": 4338 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3594, + "image_masks_loss": 0.0372, + "llm_learning_rate": 3.065426484276553e-06, + "loss": 0.7865, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0003065426484276553, + "step": 4339 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3593, + "image_masks_loss": 0.0288, + "llm_learning_rate": 3.0646183336942182e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.0003064618333694218, + "step": 4340 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3592, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.063810120934854e-06, + "loss": 0.6674, + "output_text_mask_loss": 0.0338, + "predictor_learning_rate": 0.00030638101209348533, + "step": 4341 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.3591, + "image_masks_loss": 0.0066, + "llm_learning_rate": 3.063001846087464e-06, + "loss": 0.7147, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.0003063001846087464, + "step": 4342 + }, + { + "epoch": 0.44, + "gumbel_tau": 0.359, + "image_masks_loss": 0.0045, + "llm_learning_rate": 3.0621935092410564e-06, + "loss": 0.7448, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.00030621935092410563, + "step": 4343 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3589, + "image_masks_loss": 0.0305, + "llm_learning_rate": 3.061385110484647e-06, + "loss": 0.7565, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0003061385110484647, + "step": 4344 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3589, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.06057664990726e-06, + "loss": 0.7246, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00030605766499072594, + "step": 4345 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3588, + "image_masks_loss": 0.0051, + "llm_learning_rate": 3.0597681275979235e-06, + "loss": 0.7551, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 0.0003059768127597923, + "step": 4346 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3587, + "image_masks_loss": 0.0131, + "llm_learning_rate": 3.058959543645673e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003058959543645673, + "step": 4347 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3586, + "image_masks_loss": 0.0252, + "llm_learning_rate": 3.058150898139553e-06, + "loss": 0.8545, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003058150898139553, + "step": 4348 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3585, + "image_masks_loss": 0.0187, + "llm_learning_rate": 3.057342191168613e-06, + "loss": 0.705, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030573421911686126, + "step": 4349 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3584, + "image_masks_loss": 0.0051, + "llm_learning_rate": 3.0565334228219097e-06, + "loss": 0.8272, + "output_text_mask_loss": 0.0732, + "predictor_learning_rate": 0.00030565334228219095, + "step": 4350 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3583, + "image_masks_loss": 0.0191, + "llm_learning_rate": 3.055724593188505e-06, + "loss": 0.8331, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00030557245931885047, + "step": 4351 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3583, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.054915702357472e-06, + "loss": 0.7986, + "output_text_mask_loss": 0.0328, + "predictor_learning_rate": 0.00030549157023574715, + "step": 4352 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3582, + "image_masks_loss": 0.0065, + "llm_learning_rate": 3.0541067504178844e-06, + "loss": 0.7957, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003054106750417884, + "step": 4353 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3581, + "image_masks_loss": 0.0138, + "llm_learning_rate": 3.0532977374588274e-06, + "loss": 0.5948, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003053297737458827, + "step": 4354 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.358, + "image_masks_loss": 0.0124, + "llm_learning_rate": 3.052488663569392e-06, + "loss": 0.7532, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003052488663569392, + "step": 4355 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3579, + "image_masks_loss": 0.0257, + "llm_learning_rate": 3.0516795288386742e-06, + "loss": 0.732, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003051679528838674, + "step": 4356 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3578, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.0508703333557773e-06, + "loss": 0.7875, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0003050870333355777, + "step": 4357 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3578, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.050061077209812e-06, + "loss": 0.7991, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0003050061077209812, + "step": 4358 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3577, + "image_masks_loss": 0.0057, + "llm_learning_rate": 3.049251760489897e-06, + "loss": 0.7028, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0003049251760489897, + "step": 4359 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3576, + "image_masks_loss": 0.013, + "llm_learning_rate": 3.048442383285153e-06, + "loss": 0.7132, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00030484423832851526, + "step": 4360 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3575, + "image_masks_loss": 0.0078, + "llm_learning_rate": 3.0476329456847133e-06, + "loss": 0.7537, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0003047632945684713, + "step": 4361 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3574, + "image_masks_loss": 0.015, + "llm_learning_rate": 3.0468234477777137e-06, + "loss": 0.6633, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00030468234477777135, + "step": 4362 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3573, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.0460138896532983e-06, + "loss": 0.7848, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0003046013889653298, + "step": 4363 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3572, + "image_masks_loss": 0.0081, + "llm_learning_rate": 3.0452042714006157e-06, + "loss": 0.6794, + "output_text_mask_loss": 0.0146, + "predictor_learning_rate": 0.00030452042714006157, + "step": 4364 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3572, + "image_masks_loss": 0.0081, + "llm_learning_rate": 3.0443945931088266e-06, + "loss": 0.7001, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0003044394593108826, + "step": 4365 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3571, + "image_masks_loss": 0.0364, + "llm_learning_rate": 3.043584854867091e-06, + "loss": 0.7572, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.00030435848548670906, + "step": 4366 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.357, + "image_masks_loss": 0.0086, + "llm_learning_rate": 3.0427750567645803e-06, + "loss": 0.7621, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000304277505676458, + "step": 4367 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3569, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.0419651988904714e-06, + "loss": 0.8045, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00030419651988904714, + "step": 4368 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3568, + "image_masks_loss": 0.0168, + "llm_learning_rate": 3.041155281333948e-06, + "loss": 0.6613, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00030411552813339475, + "step": 4369 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3567, + "image_masks_loss": 0.0051, + "llm_learning_rate": 3.040345304184199e-06, + "loss": 0.7537, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 0.00030403453041841986, + "step": 4370 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3567, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.0395352675304218e-06, + "loss": 0.7944, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.00030395352675304215, + "step": 4371 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3566, + "image_masks_loss": 0.0215, + "llm_learning_rate": 3.03872517146182e-06, + "loss": 0.7861, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.00030387251714618195, + "step": 4372 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3565, + "image_masks_loss": 0.0214, + "llm_learning_rate": 3.0379150160676013e-06, + "loss": 0.7592, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0003037915016067601, + "step": 4373 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3564, + "image_masks_loss": 0.0188, + "llm_learning_rate": 3.0371048014369835e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00030371048014369835, + "step": 4374 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3563, + "image_masks_loss": 0.0087, + "llm_learning_rate": 3.0362945276591888e-06, + "loss": 0.7816, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00030362945276591886, + "step": 4375 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3562, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.035484194823445e-06, + "loss": 0.7812, + "output_text_mask_loss": 0.0992, + "predictor_learning_rate": 0.0003035484194823445, + "step": 4376 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3562, + "image_masks_loss": 0.0107, + "llm_learning_rate": 3.03467380301899e-06, + "loss": 0.8404, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00030346738030189894, + "step": 4377 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3561, + "image_masks_loss": 0.0271, + "llm_learning_rate": 3.0338633523350645e-06, + "loss": 0.8279, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0003033863352335064, + "step": 4378 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.356, + "image_masks_loss": 0.0206, + "llm_learning_rate": 3.0330528428609176e-06, + "loss": 0.7846, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0003033052842860917, + "step": 4379 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3559, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.032242274685804e-06, + "loss": 0.7528, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00030322422746858037, + "step": 4380 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3558, + "image_masks_loss": 0.017, + "llm_learning_rate": 3.0314316478989857e-06, + "loss": 0.7597, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00030314316478989855, + "step": 4381 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3557, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.030620962589731e-06, + "loss": 0.7565, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00030306209625897306, + "step": 4382 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3557, + "image_masks_loss": 0.0179, + "llm_learning_rate": 3.029810218847314e-06, + "loss": 0.5753, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030298102188473135, + "step": 4383 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3556, + "image_masks_loss": 0.0326, + "llm_learning_rate": 3.028999416761015e-06, + "loss": 0.7371, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00030289994167610154, + "step": 4384 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3555, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.028188556420123e-06, + "loss": 0.7107, + "output_text_mask_loss": 0.045, + "predictor_learning_rate": 0.00030281885564201227, + "step": 4385 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3554, + "image_masks_loss": 0.0237, + "llm_learning_rate": 3.027377637913929e-06, + "loss": 0.7339, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0003027377637913929, + "step": 4386 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3553, + "image_masks_loss": 0.0157, + "llm_learning_rate": 3.0265666613317363e-06, + "loss": 0.727, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00030265666613317363, + "step": 4387 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3552, + "image_masks_loss": 0.0173, + "llm_learning_rate": 3.02575562676285e-06, + "loss": 0.647, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00030257556267628493, + "step": 4388 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3551, + "image_masks_loss": 0.0233, + "llm_learning_rate": 3.0249445342965822e-06, + "loss": 0.8359, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.0003024944534296582, + "step": 4389 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3551, + "image_masks_loss": 0.01, + "llm_learning_rate": 3.0241333840222524e-06, + "loss": 0.8164, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.00030241333840222526, + "step": 4390 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.355, + "image_masks_loss": 0.0202, + "llm_learning_rate": 3.0233221760291882e-06, + "loss": 0.7945, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 0.0003023322176029188, + "step": 4391 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3549, + "image_masks_loss": 0.0144, + "llm_learning_rate": 3.0225109104067196e-06, + "loss": 0.6961, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0003022510910406719, + "step": 4392 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3548, + "image_masks_loss": 0.0376, + "llm_learning_rate": 3.021699587244186e-06, + "loss": 0.7222, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.0003021699587244186, + "step": 4393 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3547, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.0208882066309323e-06, + "loss": 0.7952, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.00030208882066309325, + "step": 4394 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3546, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.0200767686563088e-06, + "loss": 0.729, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00030200767686563084, + "step": 4395 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3546, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.0192652734096723e-06, + "loss": 0.723, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0003019265273409672, + "step": 4396 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3545, + "image_masks_loss": 0.0345, + "llm_learning_rate": 3.0184537209803875e-06, + "loss": 0.8351, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00030184537209803876, + "step": 4397 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3544, + "image_masks_loss": 0.0208, + "llm_learning_rate": 3.017642111457825e-06, + "loss": 0.7189, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00030176421114578244, + "step": 4398 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3543, + "image_masks_loss": 0.0275, + "llm_learning_rate": 3.016830444931359e-06, + "loss": 0.7675, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00030168304449313587, + "step": 4399 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3542, + "image_masks_loss": 0.0198, + "llm_learning_rate": 3.016018721490374e-06, + "loss": 0.6837, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00030160187214903737, + "step": 4400 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3541, + "image_masks_loss": 0.0319, + "llm_learning_rate": 3.015206941224258e-06, + "loss": 0.7135, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00030152069412242577, + "step": 4401 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3541, + "image_masks_loss": 0.0069, + "llm_learning_rate": 3.014395104222406e-06, + "loss": 0.7125, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0003014395104222406, + "step": 4402 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.354, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.01358321057422e-06, + "loss": 0.7535, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 0.00030135832105742196, + "step": 4403 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3539, + "image_masks_loss": 0.0073, + "llm_learning_rate": 3.0127712603691057e-06, + "loss": 0.8222, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00030127712603691053, + "step": 4404 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3538, + "image_masks_loss": 0.0187, + "llm_learning_rate": 3.011959253696479e-06, + "loss": 0.802, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00030119592536964786, + "step": 4405 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3537, + "image_masks_loss": 0.0238, + "llm_learning_rate": 3.0111471906457577e-06, + "loss": 0.7901, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0003011147190645758, + "step": 4406 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3536, + "image_masks_loss": 0.0172, + "llm_learning_rate": 3.01033507130637e-06, + "loss": 0.7282, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.000301033507130637, + "step": 4407 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3536, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.009522895767748e-06, + "loss": 0.7432, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 0.0003009522895767748, + "step": 4408 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3535, + "image_masks_loss": 0.0309, + "llm_learning_rate": 3.0087106641193285e-06, + "loss": 0.7569, + "output_text_mask_loss": 0.0652, + "predictor_learning_rate": 0.00030087106641193284, + "step": 4409 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3534, + "image_masks_loss": 0.0045, + "llm_learning_rate": 3.007898376450559e-06, + "loss": 0.7933, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.00030078983764505584, + "step": 4410 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3533, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.0070860328508888e-06, + "loss": 0.7567, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0003007086032850889, + "step": 4411 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3532, + "image_masks_loss": 0.0125, + "llm_learning_rate": 3.006273633409774e-06, + "loss": 0.7288, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0003006273633409774, + "step": 4412 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3531, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.0054611782166798e-06, + "loss": 0.7909, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 0.00030054611782166795, + "step": 4413 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3531, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.004648667361075e-06, + "loss": 0.7493, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.00030046486673610743, + "step": 4414 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.353, + "image_masks_loss": 0.0176, + "llm_learning_rate": 3.0038361009324342e-06, + "loss": 0.7455, + "output_text_mask_loss": 0.032, + "predictor_learning_rate": 0.0003003836100932434, + "step": 4415 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3529, + "image_masks_loss": 0.0066, + "llm_learning_rate": 3.00302347902024e-06, + "loss": 0.7749, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.000300302347902024, + "step": 4416 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3528, + "image_masks_loss": 0.0054, + "llm_learning_rate": 3.00221080171398e-06, + "loss": 0.6709, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.000300221080171398, + "step": 4417 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3527, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.001398069103148e-06, + "loss": 0.7569, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00030013980691031476, + "step": 4418 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3526, + "image_masks_loss": 0.0206, + "llm_learning_rate": 3.0005852812772436e-06, + "loss": 0.6923, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00030005852812772435, + "step": 4419 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3526, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.9997724383257738e-06, + "loss": 0.7143, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00029997724383257733, + "step": 4420 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3525, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.998959540338249e-06, + "loss": 0.7505, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00029989595403382487, + "step": 4421 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3524, + "image_masks_loss": 0.0224, + "llm_learning_rate": 2.9981465874041883e-06, + "loss": 0.7333, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002998146587404188, + "step": 4422 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3523, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.9973335796131155e-06, + "loss": 0.7311, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002997333579613115, + "step": 4423 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3522, + "image_masks_loss": 0.0154, + "llm_learning_rate": 2.996520517054562e-06, + "loss": 0.8387, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.0002996520517054562, + "step": 4424 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3521, + "image_masks_loss": 0.025, + "llm_learning_rate": 2.995707399818062e-06, + "loss": 0.8336, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00029957073998180617, + "step": 4425 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3521, + "image_masks_loss": 0.0028, + "llm_learning_rate": 2.99489422799316e-06, + "loss": 0.6894, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.00029948942279931596, + "step": 4426 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.352, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.9940810016694032e-06, + "loss": 0.7164, + "output_text_mask_loss": 0.053, + "predictor_learning_rate": 0.0002994081001669403, + "step": 4427 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3519, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.993267720936346e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.00029932677209363456, + "step": 4428 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3518, + "image_masks_loss": 0.0101, + "llm_learning_rate": 2.9924543858835483e-06, + "loss": 0.8059, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.0002992454385883548, + "step": 4429 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3517, + "image_masks_loss": 0.025, + "llm_learning_rate": 2.991640996600577e-06, + "loss": 0.7776, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002991640996600577, + "step": 4430 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3516, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.9908275531770038e-06, + "loss": 0.7472, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.00029908275531770036, + "step": 4431 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3516, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.990014055702407e-06, + "loss": 0.7207, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002990014055702407, + "step": 4432 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3515, + "image_masks_loss": 0.0156, + "llm_learning_rate": 2.9892005042663718e-06, + "loss": 0.752, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0002989200504266372, + "step": 4433 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3514, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.988386898958487e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0002988386898958487, + "step": 4434 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3513, + "image_masks_loss": 0.0179, + "llm_learning_rate": 2.987573239868348e-06, + "loss": 0.7648, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002987573239868348, + "step": 4435 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3512, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.9867595270855594e-06, + "loss": 0.8302, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0002986759527085559, + "step": 4436 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3511, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.9859457606997268e-06, + "loss": 0.8432, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.00029859457606997266, + "step": 4437 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3511, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.985131940800464e-06, + "loss": 0.7176, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0002985131940800464, + "step": 4438 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.351, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.9843180674773924e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00029843180674773924, + "step": 4439 + }, + { + "epoch": 0.45, + "gumbel_tau": 0.3509, + "image_masks_loss": 0.025, + "llm_learning_rate": 2.983504140820136e-06, + "loss": 0.8867, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0002983504140820136, + "step": 4440 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3508, + "image_masks_loss": 0.0162, + "llm_learning_rate": 2.982690160918326e-06, + "loss": 0.7797, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.0002982690160918326, + "step": 4441 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3507, + "image_masks_loss": 0.0217, + "llm_learning_rate": 2.9818761278616016e-06, + "loss": 0.8017, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00029818761278616014, + "step": 4442 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3507, + "image_masks_loss": 0.002, + "llm_learning_rate": 2.981062041739604e-06, + "loss": 0.7943, + "output_text_mask_loss": 0.0702, + "predictor_learning_rate": 0.00029810620417396037, + "step": 4443 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3506, + "image_masks_loss": 0.0119, + "llm_learning_rate": 2.980247902641983e-06, + "loss": 0.7241, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00029802479026419824, + "step": 4444 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3505, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.9794337106583927e-06, + "loss": 0.7718, + "output_text_mask_loss": 0.0862, + "predictor_learning_rate": 0.00029794337106583925, + "step": 4445 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3504, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.9786194658784947e-06, + "loss": 0.72, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002978619465878494, + "step": 4446 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3503, + "image_masks_loss": 0.0191, + "llm_learning_rate": 2.977805168391955e-06, + "loss": 0.7532, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0002977805168391955, + "step": 4447 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3502, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.9769908182884456e-06, + "loss": 0.6727, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00029769908182884455, + "step": 4448 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3502, + "image_masks_loss": 0.0101, + "llm_learning_rate": 2.9761764156576458e-06, + "loss": 0.6767, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00029761764156576456, + "step": 4449 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3501, + "image_masks_loss": 0.0196, + "llm_learning_rate": 2.9753619605892376e-06, + "loss": 0.7651, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00029753619605892375, + "step": 4450 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.35, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.9745474531729113e-06, + "loss": 0.7859, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.0002974547453172911, + "step": 4451 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3499, + "image_masks_loss": 0.0246, + "llm_learning_rate": 2.9737328934983626e-06, + "loss": 0.753, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.0002973732893498363, + "step": 4452 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3498, + "image_masks_loss": 0.0138, + "llm_learning_rate": 2.9729182816552927e-06, + "loss": 0.7571, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00029729182816552925, + "step": 4453 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3497, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.9721036177334083e-06, + "loss": 0.8085, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00029721036177334084, + "step": 4454 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3497, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.9712889018224217e-06, + "loss": 0.8006, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 0.0002971288901822422, + "step": 4455 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3496, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.9704741340120513e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00029704741340120513, + "step": 4456 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3495, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.9696593143920222e-06, + "loss": 0.7152, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002969659314392022, + "step": 4457 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3494, + "image_masks_loss": 0.0267, + "llm_learning_rate": 2.9688444430520623e-06, + "loss": 0.6663, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0002968844443052062, + "step": 4458 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3493, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.9680295200819087e-06, + "loss": 0.7733, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00029680295200819085, + "step": 4459 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3492, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.967214545571302e-06, + "loss": 0.7275, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00029672145455713017, + "step": 4460 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3492, + "image_masks_loss": 0.0137, + "llm_learning_rate": 2.9663995196099884e-06, + "loss": 0.6321, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002966399519609988, + "step": 4461 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3491, + "image_masks_loss": 0.009, + "llm_learning_rate": 2.9655844422877207e-06, + "loss": 0.7773, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00029655844422877203, + "step": 4462 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.349, + "image_masks_loss": 0.0439, + "llm_learning_rate": 2.9647693136942574e-06, + "loss": 0.7186, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002964769313694257, + "step": 4463 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3489, + "image_masks_loss": 0.0119, + "llm_learning_rate": 2.963954133919361e-06, + "loss": 0.7896, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.0002963954133919361, + "step": 4464 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3488, + "image_masks_loss": 0.021, + "llm_learning_rate": 2.9631389030528034e-06, + "loss": 0.7904, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002963138903052803, + "step": 4465 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3488, + "image_masks_loss": 0.0215, + "llm_learning_rate": 2.962323621184358e-06, + "loss": 0.8161, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.0002962323621184358, + "step": 4466 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3487, + "image_masks_loss": 0.0162, + "llm_learning_rate": 2.9615082884038055e-06, + "loss": 0.787, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.0002961508288403805, + "step": 4467 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3486, + "image_masks_loss": 0.0156, + "llm_learning_rate": 2.9606929048009327e-06, + "loss": 0.7226, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 0.00029606929048009325, + "step": 4468 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3485, + "image_masks_loss": 0.0116, + "llm_learning_rate": 2.95987747046553e-06, + "loss": 0.7443, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.00029598774704655303, + "step": 4469 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3484, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.959061985487397e-06, + "loss": 0.7372, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002959061985487397, + "step": 4470 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3483, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.9582464499563347e-06, + "loss": 0.6823, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00029582464499563343, + "step": 4471 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3483, + "image_masks_loss": 0.0213, + "llm_learning_rate": 2.957430863962153e-06, + "loss": 0.7267, + "output_text_mask_loss": 0.0553, + "predictor_learning_rate": 0.0002957430863962153, + "step": 4472 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3482, + "image_masks_loss": 0.0169, + "llm_learning_rate": 2.956615227594666e-06, + "loss": 0.7876, + "output_text_mask_loss": 0.0372, + "predictor_learning_rate": 0.0002956615227594666, + "step": 4473 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3481, + "image_masks_loss": 0.0056, + "llm_learning_rate": 2.9557995409436923e-06, + "loss": 0.7383, + "output_text_mask_loss": 0.0629, + "predictor_learning_rate": 0.00029557995409436923, + "step": 4474 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.348, + "image_masks_loss": 0.0226, + "llm_learning_rate": 2.9549838040990586e-06, + "loss": 0.8401, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00029549838040990583, + "step": 4475 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3479, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.9541680171505956e-06, + "loss": 0.7329, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002954168017150595, + "step": 4476 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3479, + "image_masks_loss": 0.0198, + "llm_learning_rate": 2.953352180188137e-06, + "loss": 0.7476, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.0002953352180188137, + "step": 4477 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3478, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.952536293301528e-06, + "loss": 0.8194, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00029525362933015276, + "step": 4478 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3477, + "image_masks_loss": 0.0057, + "llm_learning_rate": 2.951720356580614e-06, + "loss": 0.717, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00029517203565806137, + "step": 4479 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3476, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.9509043701152477e-06, + "loss": 0.767, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.00029509043701152475, + "step": 4480 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3475, + "image_masks_loss": 0.022, + "llm_learning_rate": 2.9500883339952886e-06, + "loss": 0.7589, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00029500883339952884, + "step": 4481 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3474, + "image_masks_loss": 0.0273, + "llm_learning_rate": 2.9492722483105994e-06, + "loss": 0.7873, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002949272248310599, + "step": 4482 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3474, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.948456113151049e-06, + "loss": 0.7643, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 0.0002948456113151049, + "step": 4483 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3473, + "image_masks_loss": 0.0205, + "llm_learning_rate": 2.947639928606513e-06, + "loss": 0.703, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00029476399286065125, + "step": 4484 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3472, + "image_masks_loss": 0.0147, + "llm_learning_rate": 2.9468236947668715e-06, + "loss": 0.8551, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00029468236947668714, + "step": 4485 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3471, + "image_masks_loss": 0.0192, + "llm_learning_rate": 2.946007411722009e-06, + "loss": 0.6628, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.0002946007411722009, + "step": 4486 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.347, + "image_masks_loss": 0.0425, + "llm_learning_rate": 2.9451910795618168e-06, + "loss": 0.7779, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0002945191079561816, + "step": 4487 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3469, + "image_masks_loss": 0.024, + "llm_learning_rate": 2.944374698376191e-06, + "loss": 0.8544, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0002944374698376191, + "step": 4488 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3469, + "image_masks_loss": 0.0182, + "llm_learning_rate": 2.943558268255034e-06, + "loss": 0.7779, + "output_text_mask_loss": 0.1183, + "predictor_learning_rate": 0.0002943558268255034, + "step": 4489 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3468, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.942741789288252e-06, + "loss": 0.7772, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0002942741789288252, + "step": 4490 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3467, + "image_masks_loss": 0.0238, + "llm_learning_rate": 2.9419252615657585e-06, + "loss": 0.7569, + "output_text_mask_loss": 0.1144, + "predictor_learning_rate": 0.0002941925261565758, + "step": 4491 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3466, + "image_masks_loss": 0.0353, + "llm_learning_rate": 2.9411086851774705e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.000294110868517747, + "step": 4492 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3465, + "image_masks_loss": 0.0222, + "llm_learning_rate": 2.940292060213312e-06, + "loss": 0.7218, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002940292060213312, + "step": 4493 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3465, + "image_masks_loss": 0.0242, + "llm_learning_rate": 2.939475386763211e-06, + "loss": 0.7724, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.0002939475386763211, + "step": 4494 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3464, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.9386586649171015e-06, + "loss": 0.7546, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002938658664917102, + "step": 4495 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3463, + "image_masks_loss": 0.0202, + "llm_learning_rate": 2.937841894764922e-06, + "loss": 0.6407, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002937841894764922, + "step": 4496 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3462, + "image_masks_loss": 0.0234, + "llm_learning_rate": 2.937025076396619e-06, + "loss": 0.7921, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0002937025076396619, + "step": 4497 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3461, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.9362082099021405e-06, + "loss": 0.7629, + "output_text_mask_loss": 0.0389, + "predictor_learning_rate": 0.00029362082099021404, + "step": 4498 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3461, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.9353912953714423e-06, + "loss": 0.7252, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00029353912953714425, + "step": 4499 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.346, + "image_masks_loss": 0.0163, + "llm_learning_rate": 2.934574332894485e-06, + "loss": 0.7298, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00029345743328944843, + "step": 4500 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3459, + "image_masks_loss": 0.017, + "llm_learning_rate": 2.9337573225612345e-06, + "loss": 0.7638, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00029337573225612343, + "step": 4501 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3458, + "image_masks_loss": 0.0226, + "llm_learning_rate": 2.9329402644616608e-06, + "loss": 0.8472, + "output_text_mask_loss": 0.0273, + "predictor_learning_rate": 0.00029329402644616605, + "step": 4502 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3457, + "image_masks_loss": 0.019, + "llm_learning_rate": 2.9321231586857407e-06, + "loss": 0.7014, + "output_text_mask_loss": 0.0362, + "predictor_learning_rate": 0.0002932123158685741, + "step": 4503 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3456, + "image_masks_loss": 0.0179, + "llm_learning_rate": 2.931306005323457e-06, + "loss": 0.7303, + "output_text_mask_loss": 0.0721, + "predictor_learning_rate": 0.0002931306005323457, + "step": 4504 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3456, + "image_masks_loss": 0.0147, + "llm_learning_rate": 2.930488804464795e-06, + "loss": 0.6983, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00029304888044647947, + "step": 4505 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3455, + "image_masks_loss": 0.0133, + "llm_learning_rate": 2.929671556199746e-06, + "loss": 0.7518, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002929671556199746, + "step": 4506 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3454, + "image_masks_loss": 0.0156, + "llm_learning_rate": 2.928854260618309e-06, + "loss": 0.712, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0002928854260618309, + "step": 4507 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3453, + "image_masks_loss": 0.0011, + "llm_learning_rate": 2.928036917810485e-06, + "loss": 0.8191, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00029280369178104847, + "step": 4508 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3452, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.927219527866283e-06, + "loss": 0.8052, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00029272195278662826, + "step": 4509 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3452, + "image_masks_loss": 0.0198, + "llm_learning_rate": 2.926402090875714e-06, + "loss": 0.8122, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002926402090875714, + "step": 4510 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3451, + "image_masks_loss": 0.0235, + "llm_learning_rate": 2.9255846069287973e-06, + "loss": 0.7648, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00029255846069287973, + "step": 4511 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.345, + "image_masks_loss": 0.0058, + "llm_learning_rate": 2.924767076115557e-06, + "loss": 0.682, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0002924767076115556, + "step": 4512 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3449, + "image_masks_loss": 0.0189, + "llm_learning_rate": 2.9239494985260187e-06, + "loss": 0.7793, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00029239494985260187, + "step": 4513 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3448, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.923131874250218e-06, + "loss": 0.6598, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002923131874250218, + "step": 4514 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3447, + "image_masks_loss": 0.0052, + "llm_learning_rate": 2.9223142033781925e-06, + "loss": 0.84, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00029223142033781924, + "step": 4515 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3447, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.9214964859999857e-06, + "loss": 0.7833, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00029214964859999854, + "step": 4516 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3446, + "image_masks_loss": 0.0241, + "llm_learning_rate": 2.9206787222056477e-06, + "loss": 0.747, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00029206787222056473, + "step": 4517 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3445, + "image_masks_loss": 0.0098, + "llm_learning_rate": 2.9198609120852317e-06, + "loss": 0.7409, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 0.00029198609120852315, + "step": 4518 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3444, + "image_masks_loss": 0.0066, + "llm_learning_rate": 2.9190430557287964e-06, + "loss": 0.7823, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 0.0002919043055728796, + "step": 4519 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3443, + "image_masks_loss": 0.0242, + "llm_learning_rate": 2.918225153226407e-06, + "loss": 0.7093, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002918225153226407, + "step": 4520 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3443, + "image_masks_loss": 0.0137, + "llm_learning_rate": 2.9174072046681325e-06, + "loss": 0.7908, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00029174072046681324, + "step": 4521 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3442, + "image_masks_loss": 0.0229, + "llm_learning_rate": 2.916589210144046e-06, + "loss": 0.6544, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002916589210144046, + "step": 4522 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3441, + "image_masks_loss": 0.0311, + "llm_learning_rate": 2.9157711697442286e-06, + "loss": 0.6596, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00029157711697442283, + "step": 4523 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.344, + "image_masks_loss": 0.0278, + "llm_learning_rate": 2.9149530835587637e-06, + "loss": 0.669, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.00029149530835587633, + "step": 4524 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3439, + "image_masks_loss": 0.0381, + "llm_learning_rate": 2.9141349516777407e-06, + "loss": 0.841, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00029141349516777405, + "step": 4525 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3439, + "image_masks_loss": 0.0271, + "llm_learning_rate": 2.9133167741912543e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00029133167741912544, + "step": 4526 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3438, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.912498551189404e-06, + "loss": 0.765, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00029124985511894044, + "step": 4527 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3437, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.9116802827622948e-06, + "loss": 0.7618, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00029116802827622946, + "step": 4528 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3436, + "image_masks_loss": 0.0263, + "llm_learning_rate": 2.9108619690000355e-06, + "loss": 0.7403, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00029108619690000354, + "step": 4529 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3435, + "image_masks_loss": 0.0133, + "llm_learning_rate": 2.9100436099927415e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0546, + "predictor_learning_rate": 0.00029100436099927413, + "step": 4530 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3434, + "image_masks_loss": 0.0169, + "llm_learning_rate": 2.909225205830532e-06, + "loss": 0.7112, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00029092252058305316, + "step": 4531 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3434, + "image_masks_loss": 0.0233, + "llm_learning_rate": 2.908406756603531e-06, + "loss": 0.8222, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.000290840675660353, + "step": 4532 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3433, + "image_masks_loss": 0.0181, + "llm_learning_rate": 2.907588262401868e-06, + "loss": 0.663, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002907588262401868, + "step": 4533 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3432, + "image_masks_loss": 0.0233, + "llm_learning_rate": 2.9067697233156776e-06, + "loss": 0.7499, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00029067697233156777, + "step": 4534 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3431, + "image_masks_loss": 0.0269, + "llm_learning_rate": 2.9059511394350993e-06, + "loss": 0.7107, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002905951139435099, + "step": 4535 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.343, + "image_masks_loss": 0.0278, + "llm_learning_rate": 2.9051325108502774e-06, + "loss": 0.7268, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0002905132510850277, + "step": 4536 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.343, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.904313837651361e-06, + "loss": 0.7561, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.0002904313837651361, + "step": 4537 + }, + { + "epoch": 0.46, + "gumbel_tau": 0.3429, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.9034951199285044e-06, + "loss": 0.673, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.0002903495119928504, + "step": 4538 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3428, + "image_masks_loss": 0.003, + "llm_learning_rate": 2.9026763577718663e-06, + "loss": 0.7386, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002902676357771866, + "step": 4539 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3427, + "image_masks_loss": 0.0257, + "llm_learning_rate": 2.901857551271611e-06, + "loss": 0.7712, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00029018575512716104, + "step": 4540 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3426, + "image_masks_loss": 0.0195, + "llm_learning_rate": 2.9010387005179073e-06, + "loss": 0.7091, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00029010387005179076, + "step": 4541 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3426, + "image_masks_loss": 0.0282, + "llm_learning_rate": 2.900219805600928e-06, + "loss": 0.673, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00029002198056009277, + "step": 4542 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3425, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.8994008666108523e-06, + "loss": 0.6853, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00028994008666108524, + "step": 4543 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3424, + "image_masks_loss": 0.0271, + "llm_learning_rate": 2.8985818836378638e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00028985818836378636, + "step": 4544 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3423, + "image_masks_loss": 0.0202, + "llm_learning_rate": 2.89776285677215e-06, + "loss": 0.7914, + "output_text_mask_loss": 0.069, + "predictor_learning_rate": 0.00028977628567721496, + "step": 4545 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3422, + "image_masks_loss": 0.017, + "llm_learning_rate": 2.896943786103905e-06, + "loss": 0.7105, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002896943786103905, + "step": 4546 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3422, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.896124671723326e-06, + "loss": 0.7672, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002896124671723326, + "step": 4547 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3421, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.8953055137206154e-06, + "loss": 0.7409, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00028953055137206153, + "step": 4548 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.342, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.8944863121859812e-06, + "loss": 0.6821, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002894486312185981, + "step": 4549 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3419, + "image_masks_loss": 0.0105, + "llm_learning_rate": 2.893667067209637e-06, + "loss": 0.7046, + "output_text_mask_loss": 0.0402, + "predictor_learning_rate": 0.00028936670672096365, + "step": 4550 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3418, + "image_masks_loss": 0.0053, + "llm_learning_rate": 2.8928477788817964e-06, + "loss": 0.772, + "output_text_mask_loss": 0.0717, + "predictor_learning_rate": 0.00028928477788817966, + "step": 4551 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3418, + "image_masks_loss": 0.0299, + "llm_learning_rate": 2.8920284472926846e-06, + "loss": 0.6649, + "output_text_mask_loss": 0.1602, + "predictor_learning_rate": 0.00028920284472926843, + "step": 4552 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3417, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.8912090725325265e-06, + "loss": 0.713, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.00028912090725325264, + "step": 4553 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3416, + "image_masks_loss": 0.0155, + "llm_learning_rate": 2.8903896546915544e-06, + "loss": 0.7334, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00028903896546915545, + "step": 4554 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3415, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.8895701938600036e-06, + "loss": 0.7752, + "output_text_mask_loss": 0.0473, + "predictor_learning_rate": 0.00028895701938600035, + "step": 4555 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3414, + "image_masks_loss": 0.0201, + "llm_learning_rate": 2.8887506901281153e-06, + "loss": 0.7589, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00028887506901281154, + "step": 4556 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3413, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.8879311435861353e-06, + "loss": 0.7432, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002887931143586135, + "step": 4557 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3413, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.8871115543243133e-06, + "loss": 0.7497, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00028871115543243134, + "step": 4558 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3412, + "image_masks_loss": 0.0237, + "llm_learning_rate": 2.8862919224329058e-06, + "loss": 0.7966, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00028862919224329056, + "step": 4559 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3411, + "image_masks_loss": 0.0374, + "llm_learning_rate": 2.8854722480021704e-06, + "loss": 0.6649, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 0.000288547224800217, + "step": 4560 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.341, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.884652531122372e-06, + "loss": 0.7466, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002884652531122372, + "step": 4561 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3409, + "image_masks_loss": 0.0056, + "llm_learning_rate": 2.8838327718837804e-06, + "loss": 0.7254, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000288383277188378, + "step": 4562 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3409, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.8830129703766696e-06, + "loss": 0.7372, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00028830129703766695, + "step": 4563 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3408, + "image_masks_loss": 0.0042, + "llm_learning_rate": 2.8821931266913173e-06, + "loss": 0.7082, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00028821931266913173, + "step": 4564 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3407, + "image_masks_loss": 0.0179, + "llm_learning_rate": 2.8813732409180062e-06, + "loss": 0.6671, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002881373240918006, + "step": 4565 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3406, + "image_masks_loss": 0.0137, + "llm_learning_rate": 2.880553313147024e-06, + "loss": 0.7157, + "output_text_mask_loss": 0.0702, + "predictor_learning_rate": 0.0002880553313147024, + "step": 4566 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3405, + "image_masks_loss": 0.0214, + "llm_learning_rate": 2.8797333434686643e-06, + "loss": 0.7097, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0002879733343468664, + "step": 4567 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3405, + "image_masks_loss": 0.0284, + "llm_learning_rate": 2.8789133319732225e-06, + "loss": 0.6953, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 0.00028789133319732224, + "step": 4568 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3404, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.8780932787510006e-06, + "loss": 0.7975, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00028780932787510004, + "step": 4569 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3403, + "image_masks_loss": 0.0176, + "llm_learning_rate": 2.877273183892305e-06, + "loss": 0.7327, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002877273183892305, + "step": 4570 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3402, + "image_masks_loss": 0.0248, + "llm_learning_rate": 2.8764530474874454e-06, + "loss": 0.7, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002876453047487445, + "step": 4571 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3401, + "image_masks_loss": 0.0166, + "llm_learning_rate": 2.8756328696267376e-06, + "loss": 0.7351, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.00028756328696267374, + "step": 4572 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3401, + "image_masks_loss": 0.0052, + "llm_learning_rate": 2.874812650400502e-06, + "loss": 0.5748, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002874812650400502, + "step": 4573 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.34, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.873992389899063e-06, + "loss": 0.8247, + "output_text_mask_loss": 0.0793, + "predictor_learning_rate": 0.0002873992389899063, + "step": 4574 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3399, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.873172088212748e-06, + "loss": 0.6425, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00028731720882127477, + "step": 4575 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3398, + "image_masks_loss": 0.0154, + "llm_learning_rate": 2.8723517454318915e-06, + "loss": 0.7477, + "output_text_mask_loss": 0.0479, + "predictor_learning_rate": 0.00028723517454318914, + "step": 4576 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3397, + "image_masks_loss": 0.025, + "llm_learning_rate": 2.8715313616468317e-06, + "loss": 0.8411, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.00028715313616468316, + "step": 4577 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3397, + "image_masks_loss": 0.0343, + "llm_learning_rate": 2.870710936947911e-06, + "loss": 0.7172, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00028707109369479104, + "step": 4578 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3396, + "image_masks_loss": 0.0132, + "llm_learning_rate": 2.869890471425476e-06, + "loss": 0.7158, + "output_text_mask_loss": 0.0755, + "predictor_learning_rate": 0.0002869890471425476, + "step": 4579 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3395, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.8690699651698784e-06, + "loss": 0.7887, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002869069965169878, + "step": 4580 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3394, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.868249418271474e-06, + "loss": 0.8377, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.00028682494182714734, + "step": 4581 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3393, + "image_masks_loss": 0.0115, + "llm_learning_rate": 2.8674288308206238e-06, + "loss": 0.7265, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002867428830820624, + "step": 4582 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3393, + "image_masks_loss": 0.0194, + "llm_learning_rate": 2.866608202907692e-06, + "loss": 0.7745, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 0.00028666082029076917, + "step": 4583 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3392, + "image_masks_loss": 0.0234, + "llm_learning_rate": 2.865787534623048e-06, + "loss": 0.7523, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.00028657875346230476, + "step": 4584 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3391, + "image_masks_loss": 0.0163, + "llm_learning_rate": 2.864966826057066e-06, + "loss": 0.6899, + "output_text_mask_loss": 0.0237, + "predictor_learning_rate": 0.0002864966826057066, + "step": 4585 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.339, + "image_masks_loss": 0.0087, + "llm_learning_rate": 2.864146077300125e-06, + "loss": 0.8246, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0002864146077300125, + "step": 4586 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3389, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.8633252884426065e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.00028633252884426063, + "step": 4587 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3389, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.862504459574898e-06, + "loss": 0.7808, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0002862504459574898, + "step": 4588 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3388, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.861683590787391e-06, + "loss": 0.7882, + "output_text_mask_loss": 0.0171, + "predictor_learning_rate": 0.0002861683590787391, + "step": 4589 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3387, + "image_masks_loss": 0.0178, + "llm_learning_rate": 2.860862682170481e-06, + "loss": 0.7328, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0002860862682170481, + "step": 4590 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3386, + "image_masks_loss": 0.0098, + "llm_learning_rate": 2.8600417338145685e-06, + "loss": 0.7636, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00028600417338145684, + "step": 4591 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3385, + "image_masks_loss": 0.0096, + "llm_learning_rate": 2.859220745810059e-06, + "loss": 0.7037, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0002859220745810059, + "step": 4592 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3385, + "image_masks_loss": 0.0182, + "llm_learning_rate": 2.8583997182473604e-06, + "loss": 0.8234, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00028583997182473605, + "step": 4593 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3384, + "image_masks_loss": 0.0039, + "llm_learning_rate": 2.8575786512168863e-06, + "loss": 0.7354, + "output_text_mask_loss": 0.1198, + "predictor_learning_rate": 0.0002857578651216886, + "step": 4594 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3383, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.856757544809055e-06, + "loss": 0.6342, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002856757544809055, + "step": 4595 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3382, + "image_masks_loss": 0.0216, + "llm_learning_rate": 2.8559363991142885e-06, + "loss": 0.7085, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00028559363991142886, + "step": 4596 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3381, + "image_masks_loss": 0.0208, + "llm_learning_rate": 2.8551152142230127e-06, + "loss": 0.7419, + "output_text_mask_loss": 0.1312, + "predictor_learning_rate": 0.00028551152142230124, + "step": 4597 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3381, + "image_masks_loss": 0.0146, + "llm_learning_rate": 2.8542939902256584e-06, + "loss": 0.7569, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0002854293990225658, + "step": 4598 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.338, + "image_masks_loss": 0.0156, + "llm_learning_rate": 2.8534727272126605e-06, + "loss": 0.6782, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000285347272721266, + "step": 4599 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3379, + "image_masks_loss": 0.0101, + "llm_learning_rate": 2.8526514252744587e-06, + "loss": 0.8082, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.00028526514252744583, + "step": 4600 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3378, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.8518300845014967e-06, + "loss": 0.7714, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00028518300845014965, + "step": 4601 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3377, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.8510087049842215e-06, + "loss": 0.7248, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00028510087049842214, + "step": 4602 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3377, + "image_masks_loss": 0.0296, + "llm_learning_rate": 2.850187286813087e-06, + "loss": 0.7497, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0002850187286813087, + "step": 4603 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3376, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.849365830078547e-06, + "loss": 0.8237, + "output_text_mask_loss": 0.0028, + "predictor_learning_rate": 0.0002849365830078547, + "step": 4604 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3375, + "image_masks_loss": 0.0179, + "llm_learning_rate": 2.8485443348710646e-06, + "loss": 0.7958, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0002848544334871064, + "step": 4605 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3374, + "image_masks_loss": 0.0165, + "llm_learning_rate": 2.8477228012811043e-06, + "loss": 0.7231, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002847722801281104, + "step": 4606 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3373, + "image_masks_loss": 0.0174, + "llm_learning_rate": 2.8469012293991333e-06, + "loss": 0.7221, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.00028469012293991334, + "step": 4607 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3373, + "image_masks_loss": 0.0301, + "llm_learning_rate": 2.846079619315627e-06, + "loss": 0.7788, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0002846079619315627, + "step": 4608 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3372, + "image_masks_loss": 0.0259, + "llm_learning_rate": 2.845257971121062e-06, + "loss": 0.7703, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002845257971121062, + "step": 4609 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3371, + "image_masks_loss": 0.0254, + "llm_learning_rate": 2.8444362849059193e-06, + "loss": 0.7104, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0002844436284905919, + "step": 4610 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.337, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.843614560760687e-06, + "loss": 0.814, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0002843614560760687, + "step": 4611 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3369, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.842792798775853e-06, + "loss": 0.86, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00028427927987758527, + "step": 4612 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3369, + "image_masks_loss": 0.0311, + "llm_learning_rate": 2.841970999041913e-06, + "loss": 0.8412, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0002841970999041913, + "step": 4613 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3368, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.841149161649365e-06, + "loss": 0.7989, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0002841149161649365, + "step": 4614 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3367, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.840327286688711e-06, + "loss": 0.7478, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002840327286688711, + "step": 4615 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3366, + "image_masks_loss": 0.0227, + "llm_learning_rate": 2.8395053742504582e-06, + "loss": 0.6661, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 0.0002839505374250458, + "step": 4616 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3365, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.838683424425118e-06, + "loss": 0.7118, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00028386834244251174, + "step": 4617 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3365, + "image_masks_loss": 0.0141, + "llm_learning_rate": 2.8378614373032036e-06, + "loss": 0.7677, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 0.00028378614373032035, + "step": 4618 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3364, + "image_masks_loss": 0.0303, + "llm_learning_rate": 2.837039412975236e-06, + "loss": 0.6697, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.0002837039412975236, + "step": 4619 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3363, + "image_masks_loss": 0.004, + "llm_learning_rate": 2.836217351531737e-06, + "loss": 0.7679, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 0.0002836217351531737, + "step": 4620 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3362, + "image_masks_loss": 0.0218, + "llm_learning_rate": 2.8353952530632344e-06, + "loss": 0.8292, + "output_text_mask_loss": 0.038, + "predictor_learning_rate": 0.0002835395253063234, + "step": 4621 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3362, + "image_masks_loss": 0.0187, + "llm_learning_rate": 2.8345731176602598e-06, + "loss": 0.7166, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.00028345731176602595, + "step": 4622 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3361, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.8337509454133475e-06, + "loss": 0.7431, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00028337509454133477, + "step": 4623 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.336, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.832928736413039e-06, + "loss": 0.7291, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0002832928736413039, + "step": 4624 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3359, + "image_masks_loss": 0.0094, + "llm_learning_rate": 2.832106490749876e-06, + "loss": 0.7019, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0002832106490749876, + "step": 4625 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3358, + "image_masks_loss": 0.0152, + "llm_learning_rate": 2.831284208514406e-06, + "loss": 0.7449, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00028312842085144056, + "step": 4626 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3358, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.8304618897971812e-06, + "loss": 0.751, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002830461889797181, + "step": 4627 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3357, + "image_masks_loss": 0.016, + "llm_learning_rate": 2.8296395346887577e-06, + "loss": 0.7358, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00028296395346887575, + "step": 4628 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3356, + "image_masks_loss": 0.0239, + "llm_learning_rate": 2.828817143279694e-06, + "loss": 0.8215, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.0002828817143279694, + "step": 4629 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3355, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.8279947156605534e-06, + "loss": 0.738, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00028279947156605533, + "step": 4630 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3354, + "image_masks_loss": 0.02, + "llm_learning_rate": 2.8271722519219048e-06, + "loss": 0.7654, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.00028271722519219045, + "step": 4631 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3354, + "image_masks_loss": 0.0149, + "llm_learning_rate": 2.8263497521543192e-06, + "loss": 0.7724, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.00028263497521543193, + "step": 4632 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3353, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.8255272164483717e-06, + "loss": 0.7812, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00028255272164483714, + "step": 4633 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3352, + "image_masks_loss": 0.0218, + "llm_learning_rate": 2.824704644894642e-06, + "loss": 0.7626, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002824704644894642, + "step": 4634 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.3351, + "image_masks_loss": 0.0207, + "llm_learning_rate": 2.823882037583714e-06, + "loss": 0.7894, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.0002823882037583714, + "step": 4635 + }, + { + "epoch": 0.47, + "gumbel_tau": 0.335, + "image_masks_loss": 0.0025, + "llm_learning_rate": 2.8230593946061736e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002823059394606173, + "step": 4636 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.335, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.822236716052614e-06, + "loss": 0.7326, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0002822236716052614, + "step": 4637 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3349, + "image_masks_loss": 0.0294, + "llm_learning_rate": 2.821414002013629e-06, + "loss": 0.6791, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002821414002013629, + "step": 4638 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3348, + "image_masks_loss": 0.0275, + "llm_learning_rate": 2.820591252579818e-06, + "loss": 0.8269, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00028205912525798175, + "step": 4639 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3347, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.819768467841785e-06, + "loss": 0.778, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00028197684678417845, + "step": 4640 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3346, + "image_masks_loss": 0.0237, + "llm_learning_rate": 2.8189456478901354e-06, + "loss": 0.7267, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00028189456478901355, + "step": 4641 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3346, + "image_masks_loss": 0.0124, + "llm_learning_rate": 2.818122792815481e-06, + "loss": 0.6915, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.0002818122792815481, + "step": 4642 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3345, + "image_masks_loss": 0.0277, + "llm_learning_rate": 2.817299902708436e-06, + "loss": 0.7013, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00028172999027084356, + "step": 4643 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3344, + "image_masks_loss": 0.0187, + "llm_learning_rate": 2.816476977659619e-06, + "loss": 0.7123, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00028164769776596187, + "step": 4644 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3343, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.815654017759652e-06, + "loss": 0.7271, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.0002815654017759652, + "step": 4645 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3343, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.814831023099161e-06, + "loss": 0.7905, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0002814831023099161, + "step": 4646 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3342, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.8140079937687774e-06, + "loss": 0.8236, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00028140079937687774, + "step": 4647 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3341, + "image_masks_loss": 0.017, + "llm_learning_rate": 2.813184929859134e-06, + "loss": 0.7651, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.0002813184929859134, + "step": 4648 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.334, + "image_masks_loss": 0.0271, + "llm_learning_rate": 2.8123618314608673e-06, + "loss": 0.7793, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00028123618314608676, + "step": 4649 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3339, + "image_masks_loss": 0.0226, + "llm_learning_rate": 2.8115386986646215e-06, + "loss": 0.6559, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00028115386986646214, + "step": 4650 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3339, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.81071553156104e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.000281071553156104, + "step": 4651 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3338, + "image_masks_loss": 0.0041, + "llm_learning_rate": 2.8098923302407716e-06, + "loss": 0.7991, + "output_text_mask_loss": 0.0633, + "predictor_learning_rate": 0.00028098923302407713, + "step": 4652 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3337, + "image_masks_loss": 0.0132, + "llm_learning_rate": 2.8090690947944697e-06, + "loss": 0.7276, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.000280906909479447, + "step": 4653 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3336, + "image_masks_loss": 0.0208, + "llm_learning_rate": 2.8082458253127917e-06, + "loss": 0.752, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002808245825312791, + "step": 4654 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3335, + "image_masks_loss": 0.0151, + "llm_learning_rate": 2.807422521886395e-06, + "loss": 0.756, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002807422521886395, + "step": 4655 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3335, + "image_masks_loss": 0.0058, + "llm_learning_rate": 2.8065991846059466e-06, + "loss": 0.8496, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00028065991846059463, + "step": 4656 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3334, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.805775813562113e-06, + "loss": 0.7367, + "output_text_mask_loss": 0.0085, + "predictor_learning_rate": 0.0002805775813562113, + "step": 4657 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3333, + "image_masks_loss": 0.0217, + "llm_learning_rate": 2.804952408845565e-06, + "loss": 0.8456, + "output_text_mask_loss": 0.0984, + "predictor_learning_rate": 0.00028049524088455646, + "step": 4658 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3332, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.8041289705469785e-06, + "loss": 0.7597, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00028041289705469784, + "step": 4659 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3332, + "image_masks_loss": 0.0587, + "llm_learning_rate": 2.803305498757033e-06, + "loss": 0.7808, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00028033054987570326, + "step": 4660 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3331, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.802481993566411e-06, + "loss": 0.681, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.0002802481993566411, + "step": 4661 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.333, + "image_masks_loss": 0.0086, + "llm_learning_rate": 2.8016584550657966e-06, + "loss": 0.7473, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00028016584550657965, + "step": 4662 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3329, + "image_masks_loss": 0.0199, + "llm_learning_rate": 2.800834883345882e-06, + "loss": 0.7394, + "output_text_mask_loss": 0.0843, + "predictor_learning_rate": 0.0002800834883345882, + "step": 4663 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3328, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.80001127849736e-06, + "loss": 0.7702, + "output_text_mask_loss": 0.0121, + "predictor_learning_rate": 0.00028000112784973594, + "step": 4664 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3328, + "image_masks_loss": 0.0433, + "llm_learning_rate": 2.7991876406109275e-06, + "loss": 0.7185, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00027991876406109276, + "step": 4665 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3327, + "image_masks_loss": 0.0025, + "llm_learning_rate": 2.798363969777286e-06, + "loss": 0.7685, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00027983639697772857, + "step": 4666 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3326, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.7975402660871394e-06, + "loss": 0.8352, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002797540266087139, + "step": 4667 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3325, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.7967165296311954e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0002796716529631195, + "step": 4668 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3324, + "image_masks_loss": 0.0248, + "llm_learning_rate": 2.7958927605001664e-06, + "loss": 0.7205, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002795892760500166, + "step": 4669 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3324, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.7950689587847684e-06, + "loss": 0.7762, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0002795068958784768, + "step": 4670 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3323, + "image_masks_loss": 0.0252, + "llm_learning_rate": 2.7942451245757186e-06, + "loss": 0.7024, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.00027942451245757183, + "step": 4671 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3322, + "image_masks_loss": 0.0156, + "llm_learning_rate": 2.7934212579637403e-06, + "loss": 0.7287, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00027934212579637405, + "step": 4672 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3321, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.79259735903956e-06, + "loss": 0.66, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00027925973590395595, + "step": 4673 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3321, + "image_masks_loss": 0.0148, + "llm_learning_rate": 2.791773427893907e-06, + "loss": 0.7726, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00027917734278939065, + "step": 4674 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.332, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.7909494646175134e-06, + "loss": 0.8356, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.00027909494646175133, + "step": 4675 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3319, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.790125469301117e-06, + "loss": 0.7516, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.0002790125469301117, + "step": 4676 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3318, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.7893014420354573e-06, + "loss": 0.7027, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00027893014420354574, + "step": 4677 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3317, + "image_masks_loss": 0.0244, + "llm_learning_rate": 2.7884773829112787e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0645, + "predictor_learning_rate": 0.00027884773829112787, + "step": 4678 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3317, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.7876532920193284e-06, + "loss": 0.709, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0002787653292019328, + "step": 4679 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3316, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.7868291694503572e-06, + "loss": 0.648, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00027868291694503573, + "step": 4680 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3315, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.7860050152951183e-06, + "loss": 0.8044, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.0002786005015295118, + "step": 4681 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3314, + "image_masks_loss": 0.0138, + "llm_learning_rate": 2.7851808296443704e-06, + "loss": 0.737, + "output_text_mask_loss": 0.0492, + "predictor_learning_rate": 0.00027851808296443703, + "step": 4682 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3313, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.7843566125888744e-06, + "loss": 0.776, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00027843566125888744, + "step": 4683 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3313, + "image_masks_loss": 0.0261, + "llm_learning_rate": 2.7835323642193944e-06, + "loss": 0.6959, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.00027835323642193944, + "step": 4684 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3312, + "image_masks_loss": 0.0225, + "llm_learning_rate": 2.782708084626699e-06, + "loss": 0.8391, + "output_text_mask_loss": 0.0383, + "predictor_learning_rate": 0.0002782708084626699, + "step": 4685 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3311, + "image_masks_loss": 0.0226, + "llm_learning_rate": 2.7818837739015597e-06, + "loss": 0.7118, + "output_text_mask_loss": 0.0068, + "predictor_learning_rate": 0.000278188377390156, + "step": 4686 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.331, + "image_masks_loss": 0.0237, + "llm_learning_rate": 2.7810594321347517e-06, + "loss": 0.6501, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.00027810594321347516, + "step": 4687 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.331, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.7802350594170522e-06, + "loss": 0.6938, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002780235059417052, + "step": 4688 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3309, + "image_masks_loss": 0.0515, + "llm_learning_rate": 2.779410655839245e-06, + "loss": 0.7604, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0002779410655839245, + "step": 4689 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3308, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.7785862214921133e-06, + "loss": 0.7344, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002778586221492113, + "step": 4690 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3307, + "image_masks_loss": 0.0087, + "llm_learning_rate": 2.777761756466446e-06, + "loss": 0.7865, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002777761756466446, + "step": 4691 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3306, + "image_masks_loss": 0.009, + "llm_learning_rate": 2.776937260853036e-06, + "loss": 0.7467, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00027769372608530356, + "step": 4692 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3306, + "image_masks_loss": 0.0202, + "llm_learning_rate": 2.7761127347426774e-06, + "loss": 0.6829, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00027761127347426774, + "step": 4693 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3305, + "image_masks_loss": 0.0259, + "llm_learning_rate": 2.7752881782261705e-06, + "loss": 0.7591, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.000277528817822617, + "step": 4694 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3304, + "image_masks_loss": 0.0243, + "llm_learning_rate": 2.7744635913943146e-06, + "loss": 0.7648, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.00027744635913943145, + "step": 4695 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3303, + "image_masks_loss": 0.0347, + "llm_learning_rate": 2.7736389743379176e-06, + "loss": 0.7317, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00027736389743379176, + "step": 4696 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3303, + "image_masks_loss": 0.0182, + "llm_learning_rate": 2.7728143271477877e-06, + "loss": 0.8177, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0002772814327147787, + "step": 4697 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3302, + "image_masks_loss": 0.0246, + "llm_learning_rate": 2.771989649914736e-06, + "loss": 0.7775, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0002771989649914736, + "step": 4698 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3301, + "image_masks_loss": 0.0035, + "llm_learning_rate": 2.7711649427295778e-06, + "loss": 0.6848, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0002771164942729578, + "step": 4699 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.33, + "image_masks_loss": 0.0131, + "llm_learning_rate": 2.770340205683133e-06, + "loss": 0.6868, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 0.0002770340205683133, + "step": 4700 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3299, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.7695154388662214e-06, + "loss": 0.6637, + "output_text_mask_loss": 0.0408, + "predictor_learning_rate": 0.00027695154388662215, + "step": 4701 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3299, + "image_masks_loss": 0.0058, + "llm_learning_rate": 2.76869064236967e-06, + "loss": 0.8241, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000276869064236967, + "step": 4702 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3298, + "image_masks_loss": 0.0275, + "llm_learning_rate": 2.767865816284307e-06, + "loss": 0.7128, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.0002767865816284307, + "step": 4703 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3297, + "image_masks_loss": 0.0136, + "llm_learning_rate": 2.7670409607009624e-06, + "loss": 0.739, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0002767040960700962, + "step": 4704 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3296, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.7662160757104733e-06, + "loss": 0.7004, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002766216075710473, + "step": 4705 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3296, + "image_masks_loss": 0.0099, + "llm_learning_rate": 2.765391161403677e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0002765391161403677, + "step": 4706 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3295, + "image_masks_loss": 0.0193, + "llm_learning_rate": 2.764566217871415e-06, + "loss": 0.808, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002764566217871415, + "step": 4707 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3294, + "image_masks_loss": 0.0237, + "llm_learning_rate": 2.7637412452045313e-06, + "loss": 0.7883, + "output_text_mask_loss": 0.0166, + "predictor_learning_rate": 0.0002763741245204531, + "step": 4708 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3293, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.762916243493875e-06, + "loss": 0.831, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00027629162434938747, + "step": 4709 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3292, + "image_masks_loss": 0.0173, + "llm_learning_rate": 2.7620912128302957e-06, + "loss": 0.7623, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00027620912128302954, + "step": 4710 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3292, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.761266153304648e-06, + "loss": 0.7383, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00027612661533046477, + "step": 4711 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3291, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.7604410650077902e-06, + "loss": 0.7734, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000276044106500779, + "step": 4712 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.329, + "image_masks_loss": 0.023, + "llm_learning_rate": 2.7596159480305818e-06, + "loss": 0.8091, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002759615948030582, + "step": 4713 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3289, + "image_masks_loss": 0.0048, + "llm_learning_rate": 2.7587908024638865e-06, + "loss": 0.6507, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0002758790802463886, + "step": 4714 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3289, + "image_masks_loss": 0.0278, + "llm_learning_rate": 2.757965628398572e-06, + "loss": 0.7207, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002757965628398572, + "step": 4715 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3288, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.7571404259255087e-06, + "loss": 0.7216, + "output_text_mask_loss": 0.0089, + "predictor_learning_rate": 0.00027571404259255083, + "step": 4716 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3287, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.7563151951355684e-06, + "loss": 0.8974, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0002756315195135568, + "step": 4717 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3286, + "image_masks_loss": 0.0178, + "llm_learning_rate": 2.755489936119628e-06, + "loss": 0.8344, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.0002755489936119628, + "step": 4718 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3285, + "image_masks_loss": 0.0138, + "llm_learning_rate": 2.7546646489685662e-06, + "loss": 0.7416, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002754664648968566, + "step": 4719 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3285, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.753839333773266e-06, + "loss": 0.6943, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0002753839333773266, + "step": 4720 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3284, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.753013990624614e-06, + "loss": 0.7417, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0002753013990624614, + "step": 4721 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3283, + "image_masks_loss": 0.017, + "llm_learning_rate": 2.7521886196134966e-06, + "loss": 0.7714, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.00027521886196134966, + "step": 4722 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3282, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.7513632208308076e-06, + "loss": 0.7372, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00027513632208308074, + "step": 4723 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3282, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.7505377943674398e-06, + "loss": 0.8147, + "output_text_mask_loss": 0.0607, + "predictor_learning_rate": 0.00027505377943674397, + "step": 4724 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3281, + "image_masks_loss": 0.015, + "llm_learning_rate": 2.749712340314293e-06, + "loss": 0.782, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00027497123403142927, + "step": 4725 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.328, + "image_masks_loss": 0.0212, + "llm_learning_rate": 2.7488868587622674e-06, + "loss": 0.7584, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00027488868587622673, + "step": 4726 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3279, + "image_masks_loss": 0.018, + "llm_learning_rate": 2.748061349802266e-06, + "loss": 0.7452, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00027480613498022654, + "step": 4727 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3279, + "image_masks_loss": 0.0173, + "llm_learning_rate": 2.747235813525196e-06, + "loss": 0.8079, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00027472358135251956, + "step": 4728 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3278, + "image_masks_loss": 0.0271, + "llm_learning_rate": 2.7464102500219676e-06, + "loss": 0.8042, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0002746410250021967, + "step": 4729 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3277, + "image_masks_loss": 0.0203, + "llm_learning_rate": 2.745584659383493e-06, + "loss": 0.7777, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00027455846593834926, + "step": 4730 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3276, + "image_masks_loss": 0.0124, + "llm_learning_rate": 2.7447590417006893e-06, + "loss": 0.7651, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002744759041700689, + "step": 4731 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3275, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.743933397064475e-06, + "loss": 0.7282, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0002743933397064475, + "step": 4732 + }, + { + "epoch": 0.48, + "gumbel_tau": 0.3275, + "image_masks_loss": 0.0039, + "llm_learning_rate": 2.743107725565771e-06, + "loss": 0.7512, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0002743107725565771, + "step": 4733 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3274, + "image_masks_loss": 0.036, + "llm_learning_rate": 2.7422820272955032e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0002742282027295503, + "step": 4734 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3273, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.7414563023445993e-06, + "loss": 0.7217, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00027414563023445994, + "step": 4735 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3272, + "image_masks_loss": 0.0332, + "llm_learning_rate": 2.7406305508039886e-06, + "loss": 0.781, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00027406305508039885, + "step": 4736 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3272, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.7398047727646064e-06, + "loss": 0.768, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00027398047727646065, + "step": 4737 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3271, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.7389789683173882e-06, + "loss": 0.6796, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002738978968317388, + "step": 4738 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.327, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.7381531375532743e-06, + "loss": 0.7432, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00027381531375532744, + "step": 4739 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3269, + "image_masks_loss": 0.0182, + "llm_learning_rate": 2.7373272805632056e-06, + "loss": 0.596, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00027373272805632053, + "step": 4740 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3268, + "image_masks_loss": 0.0052, + "llm_learning_rate": 2.736501397438128e-06, + "loss": 0.7829, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00027365013974381283, + "step": 4741 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3268, + "image_masks_loss": 0.0166, + "llm_learning_rate": 2.7356754882689905e-06, + "loss": 0.6647, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000273567548826899, + "step": 4742 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3267, + "image_masks_loss": 0.008, + "llm_learning_rate": 2.7348495531467427e-06, + "loss": 0.7371, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 0.00027348495531467426, + "step": 4743 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3266, + "image_masks_loss": 0.0252, + "llm_learning_rate": 2.73402359216234e-06, + "loss": 0.8094, + "output_text_mask_loss": 0.0793, + "predictor_learning_rate": 0.00027340235921623397, + "step": 4744 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3265, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.7331976054067376e-06, + "loss": 0.7391, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.00027331976054067374, + "step": 4745 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3265, + "image_masks_loss": 0.0143, + "llm_learning_rate": 2.7323715929708947e-06, + "loss": 0.7677, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0002732371592970895, + "step": 4746 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3264, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.731545554945775e-06, + "loss": 0.6811, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002731545554945775, + "step": 4747 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3263, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.730719491422343e-06, + "loss": 0.6475, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00027307194914223426, + "step": 4748 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3262, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.729893402491567e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00027298934024915666, + "step": 4749 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3262, + "image_masks_loss": 0.0263, + "llm_learning_rate": 2.729067288244416e-06, + "loss": 0.7528, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0002729067288244416, + "step": 4750 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3261, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.728241148771866e-06, + "loss": 0.6953, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0002728241148771866, + "step": 4751 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.326, + "image_masks_loss": 0.0381, + "llm_learning_rate": 2.7274149841648932e-06, + "loss": 0.8011, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0002727414984164893, + "step": 4752 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3259, + "image_masks_loss": 0.0234, + "llm_learning_rate": 2.7265887945144743e-06, + "loss": 0.7242, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0002726588794514474, + "step": 4753 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3258, + "image_masks_loss": 0.0022, + "llm_learning_rate": 2.725762579911594e-06, + "loss": 0.7772, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002725762579911594, + "step": 4754 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3258, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.724936340447235e-06, + "loss": 0.7976, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00027249363404472347, + "step": 4755 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3257, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.7241100762123846e-06, + "loss": 0.8366, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00027241100762123846, + "step": 4756 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3256, + "image_masks_loss": 0.0345, + "llm_learning_rate": 2.723283787298034e-06, + "loss": 0.7398, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002723283787298034, + "step": 4757 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3255, + "image_masks_loss": 0.0116, + "llm_learning_rate": 2.7224574737951755e-06, + "loss": 0.729, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002722457473795175, + "step": 4758 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3255, + "image_masks_loss": 0.0129, + "llm_learning_rate": 2.721631135794804e-06, + "loss": 0.8005, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 0.00027216311357948036, + "step": 4759 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3254, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.7208047733879183e-06, + "loss": 0.7033, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 0.0002720804773387918, + "step": 4760 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3253, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.7199783866655192e-06, + "loss": 0.7272, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0002719978386665519, + "step": 4761 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3252, + "image_masks_loss": 0.0257, + "llm_learning_rate": 2.71915197571861e-06, + "loss": 0.7759, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.000271915197571861, + "step": 4762 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3252, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.7183255406381976e-06, + "loss": 0.6961, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.00027183255406381974, + "step": 4763 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3251, + "image_masks_loss": 0.0089, + "llm_learning_rate": 2.71749908151529e-06, + "loss": 0.6817, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.000271749908151529, + "step": 4764 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.325, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.7166725984408994e-06, + "loss": 0.7634, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0002716672598440899, + "step": 4765 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3249, + "image_masks_loss": 0.018, + "llm_learning_rate": 2.715846091506039e-06, + "loss": 0.7247, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002715846091506039, + "step": 4766 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3248, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.7150195608017265e-06, + "loss": 0.7584, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00027150195608017266, + "step": 4767 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3248, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.7141930064189815e-06, + "loss": 0.7463, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002714193006418981, + "step": 4768 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3247, + "image_masks_loss": 0.0199, + "llm_learning_rate": 2.713366428448825e-06, + "loss": 0.7252, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 0.0002713366428448825, + "step": 4769 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3246, + "image_masks_loss": 0.0275, + "llm_learning_rate": 2.712539826982282e-06, + "loss": 0.7975, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.0002712539826982282, + "step": 4770 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3245, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.7117132021103803e-06, + "loss": 0.7559, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00027117132021103803, + "step": 4771 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3245, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.710886553924149e-06, + "loss": 0.7782, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002710886553924149, + "step": 4772 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3244, + "image_masks_loss": 0.0146, + "llm_learning_rate": 2.710059882514621e-06, + "loss": 0.7217, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00027100598825146205, + "step": 4773 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3243, + "image_masks_loss": 0.0271, + "llm_learning_rate": 2.7092331879728305e-06, + "loss": 0.8022, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00027092331879728307, + "step": 4774 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3242, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.708406470389815e-06, + "loss": 0.7449, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0002708406470389815, + "step": 4775 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3242, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.707579729856615e-06, + "loss": 0.7291, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002707579729856615, + "step": 4776 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3241, + "image_masks_loss": 0.0296, + "llm_learning_rate": 2.7067529664642735e-06, + "loss": 0.7165, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0002706752966464273, + "step": 4777 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.324, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.705926180303834e-06, + "loss": 0.7262, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0002705926180303834, + "step": 4778 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3239, + "image_masks_loss": 0.0043, + "llm_learning_rate": 2.7050993714663447e-06, + "loss": 0.691, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002705099371466345, + "step": 4779 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3239, + "image_masks_loss": 0.0183, + "llm_learning_rate": 2.704272540042856e-06, + "loss": 0.7468, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002704272540042856, + "step": 4780 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3238, + "image_masks_loss": 0.0055, + "llm_learning_rate": 2.703445686124421e-06, + "loss": 0.7352, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00027034456861244207, + "step": 4781 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3237, + "image_masks_loss": 0.0075, + "llm_learning_rate": 2.702618809802093e-06, + "loss": 0.8826, + "output_text_mask_loss": 0.1022, + "predictor_learning_rate": 0.00027026188098020924, + "step": 4782 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3236, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.7017919111669304e-06, + "loss": 0.7913, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00027017919111669306, + "step": 4783 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3235, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.7009649903099936e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00027009649903099935, + "step": 4784 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3235, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.7001380473223437e-06, + "loss": 0.787, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00027001380473223435, + "step": 4785 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3234, + "image_masks_loss": 0.0222, + "llm_learning_rate": 2.6993110822950465e-06, + "loss": 0.7158, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0002699311082295046, + "step": 4786 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3233, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.6984840953191693e-06, + "loss": 0.7199, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002698484095319169, + "step": 4787 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3232, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.6976570864857814e-06, + "loss": 0.7422, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002697657086485781, + "step": 4788 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3232, + "image_masks_loss": 0.0152, + "llm_learning_rate": 2.6968300558859537e-06, + "loss": 0.6922, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0002696830055885954, + "step": 4789 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3231, + "image_masks_loss": 0.0101, + "llm_learning_rate": 2.6960030036107634e-06, + "loss": 0.6855, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.0002696003003610763, + "step": 4790 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.323, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.6951759297512855e-06, + "loss": 0.7056, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.0002695175929751285, + "step": 4791 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3229, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.694348834398599e-06, + "loss": 0.644, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0002694348834398599, + "step": 4792 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3229, + "image_masks_loss": 0.019, + "llm_learning_rate": 2.693521717643786e-06, + "loss": 0.6462, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0002693521717643786, + "step": 4793 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3228, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.692694579577931e-06, + "loss": 0.7411, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00026926945795779304, + "step": 4794 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3227, + "image_masks_loss": 0.0176, + "llm_learning_rate": 2.6918674202921184e-06, + "loss": 0.6891, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0002691867420292118, + "step": 4795 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3226, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.6910402398774386e-06, + "loss": 0.85, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.00026910402398774386, + "step": 4796 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3226, + "image_masks_loss": 0.005, + "llm_learning_rate": 2.690213038424983e-06, + "loss": 0.7755, + "output_text_mask_loss": 0.0393, + "predictor_learning_rate": 0.0002690213038424983, + "step": 4797 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3225, + "image_masks_loss": 0.0178, + "llm_learning_rate": 2.689385816025843e-06, + "loss": 0.7313, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00026893858160258427, + "step": 4798 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3224, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.688558572771115e-06, + "loss": 0.7622, + "output_text_mask_loss": 0.0044, + "predictor_learning_rate": 0.00026885585727711147, + "step": 4799 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3223, + "image_masks_loss": 0.0105, + "llm_learning_rate": 2.687731308751898e-06, + "loss": 0.6936, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0002687731308751898, + "step": 4800 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3223, + "image_masks_loss": 0.0177, + "llm_learning_rate": 2.6869040240592904e-06, + "loss": 0.7323, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00026869040240592903, + "step": 4801 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3222, + "image_masks_loss": 0.0066, + "llm_learning_rate": 2.686076718784395e-06, + "loss": 0.8062, + "output_text_mask_loss": 0.042, + "predictor_learning_rate": 0.00026860767187843947, + "step": 4802 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3221, + "image_masks_loss": 0.0228, + "llm_learning_rate": 2.6852493930183176e-06, + "loss": 0.7459, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00026852493930183176, + "step": 4803 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.322, + "image_masks_loss": 0.0311, + "llm_learning_rate": 2.684422046852164e-06, + "loss": 0.7265, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0002684422046852164, + "step": 4804 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3219, + "image_masks_loss": 0.0099, + "llm_learning_rate": 2.6835946803770434e-06, + "loss": 0.771, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.0002683594680377043, + "step": 4805 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3219, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.6827672936840687e-06, + "loss": 0.8172, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 0.00026827672936840683, + "step": 4806 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3218, + "image_masks_loss": 0.0248, + "llm_learning_rate": 2.681939886864352e-06, + "loss": 0.8018, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002681939886864352, + "step": 4807 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3217, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.6811124600090094e-06, + "loss": 0.7869, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.00026811124600090093, + "step": 4808 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3216, + "image_masks_loss": 0.0301, + "llm_learning_rate": 2.68028501320916e-06, + "loss": 0.7647, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.000268028501320916, + "step": 4809 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3216, + "image_masks_loss": 0.016, + "llm_learning_rate": 2.679457546555923e-06, + "loss": 0.7727, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0002679457546555923, + "step": 4810 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3215, + "image_masks_loss": 0.016, + "llm_learning_rate": 2.678630060140421e-06, + "loss": 0.7102, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00026786300601404204, + "step": 4811 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3214, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.6778025540537787e-06, + "loss": 0.7179, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00026778025540537785, + "step": 4812 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3213, + "image_masks_loss": 0.0179, + "llm_learning_rate": 2.6769750283871237e-06, + "loss": 0.7715, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00026769750283871235, + "step": 4813 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3213, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.6761474832315844e-06, + "loss": 0.7512, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00026761474832315843, + "step": 4814 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3212, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.6753199186782914e-06, + "loss": 0.7724, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00026753199186782913, + "step": 4815 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3211, + "image_masks_loss": 0.0239, + "llm_learning_rate": 2.6744923348183786e-06, + "loss": 0.745, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002674492334818379, + "step": 4816 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.321, + "image_masks_loss": 0.0057, + "llm_learning_rate": 2.6736647317429813e-06, + "loss": 0.7685, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00026736647317429814, + "step": 4817 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.321, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.672837109543237e-06, + "loss": 0.762, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002672837109543237, + "step": 4818 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3209, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.6720094683102856e-06, + "loss": 0.6896, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002672009468310285, + "step": 4819 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3208, + "image_masks_loss": 0.0065, + "llm_learning_rate": 2.6711818081352682e-06, + "loss": 0.7794, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.0002671181808135268, + "step": 4820 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3207, + "image_masks_loss": 0.0132, + "llm_learning_rate": 2.670354129109328e-06, + "loss": 0.8117, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0002670354129109328, + "step": 4821 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3207, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.6695264313236126e-06, + "loss": 0.8191, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002669526431323612, + "step": 4822 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3206, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.668698714869268e-06, + "loss": 0.7693, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0002668698714869268, + "step": 4823 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3205, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.667870979837446e-06, + "loss": 0.6813, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0002667870979837446, + "step": 4824 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3204, + "image_masks_loss": 0.0094, + "llm_learning_rate": 2.6670432263192975e-06, + "loss": 0.7893, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.00026670432263192973, + "step": 4825 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3204, + "image_masks_loss": 0.0164, + "llm_learning_rate": 2.666215454405977e-06, + "loss": 0.6863, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00026662154544059767, + "step": 4826 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3203, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.6653876641886407e-06, + "loss": 0.8567, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00026653876641886406, + "step": 4827 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3202, + "image_masks_loss": 0.0311, + "llm_learning_rate": 2.664559855758446e-06, + "loss": 0.7639, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0002664559855758446, + "step": 4828 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3201, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.6637320292065537e-06, + "loss": 0.706, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00026637320292065536, + "step": 4829 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.3201, + "image_masks_loss": 0.0292, + "llm_learning_rate": 2.6629041846241255e-06, + "loss": 0.7814, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.00026629041846241253, + "step": 4830 + }, + { + "epoch": 0.49, + "gumbel_tau": 0.32, + "image_masks_loss": 0.0038, + "llm_learning_rate": 2.6620763221023247e-06, + "loss": 0.7437, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.0002662076322102325, + "step": 4831 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3199, + "image_masks_loss": 0.0151, + "llm_learning_rate": 2.6612484417323196e-06, + "loss": 0.7663, + "output_text_mask_loss": 0.0389, + "predictor_learning_rate": 0.000266124844173232, + "step": 4832 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3198, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.6604205436052767e-06, + "loss": 0.7476, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00026604205436052766, + "step": 4833 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3198, + "image_masks_loss": 0.0084, + "llm_learning_rate": 2.6595926278123655e-06, + "loss": 0.8104, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 0.00026595926278123657, + "step": 4834 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3197, + "image_masks_loss": 0.0248, + "llm_learning_rate": 2.658764694444759e-06, + "loss": 0.809, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00026587646944447593, + "step": 4835 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3196, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.6579367435936316e-06, + "loss": 0.7057, + "output_text_mask_loss": 0.0729, + "predictor_learning_rate": 0.0002657936743593631, + "step": 4836 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3195, + "image_masks_loss": 0.0053, + "llm_learning_rate": 2.657108775350157e-06, + "loss": 0.8102, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00026571087753501574, + "step": 4837 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3195, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.6562807898055147e-06, + "loss": 0.6871, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.0002656280789805515, + "step": 4838 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3194, + "image_masks_loss": 0.0157, + "llm_learning_rate": 2.655452787050884e-06, + "loss": 0.7287, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00026554527870508837, + "step": 4839 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3193, + "image_masks_loss": 0.0173, + "llm_learning_rate": 2.6546247671774455e-06, + "loss": 0.7198, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00026546247671774454, + "step": 4840 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3192, + "image_masks_loss": 0.0202, + "llm_learning_rate": 2.6537967302763836e-06, + "loss": 0.7193, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00026537967302763834, + "step": 4841 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3192, + "image_masks_loss": 0.0175, + "llm_learning_rate": 2.652968676438883e-06, + "loss": 0.8208, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00026529686764388826, + "step": 4842 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3191, + "image_masks_loss": 0.0026, + "llm_learning_rate": 2.6521406057561313e-06, + "loss": 0.7389, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0002652140605756131, + "step": 4843 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.319, + "image_masks_loss": 0.0193, + "llm_learning_rate": 2.651312518319317e-06, + "loss": 0.7683, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00026513125183193166, + "step": 4844 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3189, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.650484414219631e-06, + "loss": 0.7467, + "output_text_mask_loss": 0.0016, + "predictor_learning_rate": 0.00026504844142196304, + "step": 4845 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3188, + "image_masks_loss": 0.0214, + "llm_learning_rate": 2.6496562935482666e-06, + "loss": 0.737, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00026496562935482664, + "step": 4846 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3188, + "image_masks_loss": 0.0278, + "llm_learning_rate": 2.6488281563964174e-06, + "loss": 0.7458, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00026488281563964174, + "step": 4847 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3187, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.6480000028552798e-06, + "loss": 0.7985, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.000264800000285528, + "step": 4848 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3186, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.647171833016053e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0182, + "predictor_learning_rate": 0.00026471718330160527, + "step": 4849 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3185, + "image_masks_loss": 0.0176, + "llm_learning_rate": 2.6463436469699354e-06, + "loss": 0.7609, + "output_text_mask_loss": 0.0303, + "predictor_learning_rate": 0.00026463436469699355, + "step": 4850 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3185, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.64551544480813e-06, + "loss": 0.7213, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.000264551544480813, + "step": 4851 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3184, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.6446872266218394e-06, + "loss": 0.7757, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.0002644687226621839, + "step": 4852 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3183, + "image_masks_loss": 0.0059, + "llm_learning_rate": 2.6438589925022695e-06, + "loss": 0.6357, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002643858992502269, + "step": 4853 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3182, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.643030742540626e-06, + "loss": 0.6698, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00026430307425406255, + "step": 4854 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3182, + "image_masks_loss": 0.0305, + "llm_learning_rate": 2.6422024768281195e-06, + "loss": 0.7835, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00026422024768281194, + "step": 4855 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3181, + "image_masks_loss": 0.0393, + "llm_learning_rate": 2.641374195455959e-06, + "loss": 0.7614, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002641374195455959, + "step": 4856 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.318, + "image_masks_loss": 0.005, + "llm_learning_rate": 2.6405458985153575e-06, + "loss": 0.6973, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00026405458985153573, + "step": 4857 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3179, + "image_masks_loss": 0.0227, + "llm_learning_rate": 2.639717586097529e-06, + "loss": 0.6982, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00026397175860975287, + "step": 4858 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3179, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.638889258293688e-06, + "loss": 0.7897, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002638889258293688, + "step": 4859 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3178, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.638060915195052e-06, + "loss": 0.7427, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002638060915195052, + "step": 4860 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3177, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.6372325568928416e-06, + "loss": 0.776, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 0.00026372325568928414, + "step": 4861 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3176, + "image_masks_loss": 0.0244, + "llm_learning_rate": 2.636404183478276e-06, + "loss": 0.8821, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0002636404183478276, + "step": 4862 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3176, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.6355757950425776e-06, + "loss": 0.7024, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002635575795042577, + "step": 4863 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3175, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.634747391676971e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002634747391676971, + "step": 4864 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3174, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.633918973472682e-06, + "loss": 0.7759, + "output_text_mask_loss": 0.1213, + "predictor_learning_rate": 0.0002633918973472682, + "step": 4865 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3173, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.633090540520936e-06, + "loss": 0.6847, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.0002633090540520936, + "step": 4866 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3173, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.6322620929129646e-06, + "loss": 0.7084, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0002632262092912964, + "step": 4867 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3172, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.631433630739996e-06, + "loss": 0.7603, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002631433630739996, + "step": 4868 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3171, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.6306051540932635e-06, + "loss": 0.8066, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00026306051540932635, + "step": 4869 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.317, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.6297766630640005e-06, + "loss": 0.6912, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00026297766630640004, + "step": 4870 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.317, + "image_masks_loss": 0.0163, + "llm_learning_rate": 2.628948157743442e-06, + "loss": 0.7274, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002628948157743442, + "step": 4871 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3169, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.628119638222826e-06, + "loss": 0.7144, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0002628119638222826, + "step": 4872 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3168, + "image_masks_loss": 0.0087, + "llm_learning_rate": 2.627291104593389e-06, + "loss": 0.7113, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0002627291104593389, + "step": 4873 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3168, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.6264625569463737e-06, + "loss": 0.7589, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00026264625569463734, + "step": 4874 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3167, + "image_masks_loss": 0.0292, + "llm_learning_rate": 2.6256339953730186e-06, + "loss": 0.6613, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00026256339953730184, + "step": 4875 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3166, + "image_masks_loss": 0.0204, + "llm_learning_rate": 2.6248054199645678e-06, + "loss": 0.7888, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00026248054199645677, + "step": 4876 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3165, + "image_masks_loss": 0.0152, + "llm_learning_rate": 2.623976830812267e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0002623976830812267, + "step": 4877 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3165, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.623148228007361e-06, + "loss": 0.7248, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.0002623148228007361, + "step": 4878 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3164, + "image_masks_loss": 0.0315, + "llm_learning_rate": 2.6223196116410976e-06, + "loss": 0.7506, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00026223196116410976, + "step": 4879 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3163, + "image_masks_loss": 0.034, + "llm_learning_rate": 2.6214909818047266e-06, + "loss": 0.8272, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0002621490981804727, + "step": 4880 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3162, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.6206623385894982e-06, + "loss": 0.7691, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00026206623385894976, + "step": 4881 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3162, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.6198336820866645e-06, + "loss": 0.7419, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.00026198336820866644, + "step": 4882 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3161, + "image_masks_loss": 0.0178, + "llm_learning_rate": 2.6190050123874784e-06, + "loss": 0.7194, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002619005012387478, + "step": 4883 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.316, + "image_masks_loss": 0.0244, + "llm_learning_rate": 2.618176329583195e-06, + "loss": 0.7312, + "output_text_mask_loss": 0.0406, + "predictor_learning_rate": 0.0002618176329583195, + "step": 4884 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3159, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.6173476337650717e-06, + "loss": 0.7259, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00026173476337650715, + "step": 4885 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3159, + "image_masks_loss": 0.016, + "llm_learning_rate": 2.6165189250243643e-06, + "loss": 0.6597, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00026165189250243644, + "step": 4886 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3158, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.6156902034523346e-06, + "loss": 0.7664, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00026156902034523343, + "step": 4887 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3157, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.614861469140241e-06, + "loss": 0.8126, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.0002614861469140241, + "step": 4888 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3156, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.614032722179347e-06, + "loss": 0.7723, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002614032722179347, + "step": 4889 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3156, + "image_masks_loss": 0.0334, + "llm_learning_rate": 2.6132039626609157e-06, + "loss": 0.7115, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 0.00026132039626609155, + "step": 4890 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3155, + "image_masks_loss": 0.0091, + "llm_learning_rate": 2.6123751906762124e-06, + "loss": 0.7562, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00026123751906762123, + "step": 4891 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3154, + "image_masks_loss": 0.0086, + "llm_learning_rate": 2.6115464063165026e-06, + "loss": 0.7177, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00026115464063165025, + "step": 4892 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3153, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.6107176096730537e-06, + "loss": 0.8124, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.0002610717609673053, + "step": 4893 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3153, + "image_masks_loss": 0.0155, + "llm_learning_rate": 2.6098888008371348e-06, + "loss": 0.789, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00026098888008371345, + "step": 4894 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3152, + "image_masks_loss": 0.0142, + "llm_learning_rate": 2.6090599799000165e-06, + "loss": 0.7779, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.00026090599799000164, + "step": 4895 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3151, + "image_masks_loss": 0.022, + "llm_learning_rate": 2.6082311469529705e-06, + "loss": 0.8053, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.000260823114695297, + "step": 4896 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.315, + "image_masks_loss": 0.0075, + "llm_learning_rate": 2.6074023020872697e-06, + "loss": 0.7808, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00026074023020872695, + "step": 4897 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.315, + "image_masks_loss": 0.0158, + "llm_learning_rate": 2.6065734453941883e-06, + "loss": 0.7875, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0002606573445394188, + "step": 4898 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3149, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.6057445769650007e-06, + "loss": 0.7391, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002605744576965001, + "step": 4899 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3148, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.6049156968909855e-06, + "loss": 0.6794, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00026049156968909853, + "step": 4900 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3147, + "image_masks_loss": 0.0246, + "llm_learning_rate": 2.60408680526342e-06, + "loss": 0.6954, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000260408680526342, + "step": 4901 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3147, + "image_masks_loss": 0.0237, + "llm_learning_rate": 2.603257902173584e-06, + "loss": 0.7416, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00026032579021735834, + "step": 4902 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3146, + "image_masks_loss": 0.0141, + "llm_learning_rate": 2.602428987712757e-06, + "loss": 0.7152, + "output_text_mask_loss": 0.0546, + "predictor_learning_rate": 0.0002602428987712757, + "step": 4903 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3145, + "image_masks_loss": 0.0197, + "llm_learning_rate": 2.6016000619722226e-06, + "loss": 0.7094, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.00026016000619722226, + "step": 4904 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3144, + "image_masks_loss": 0.0372, + "llm_learning_rate": 2.600771125043262e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002600771125043262, + "step": 4905 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3144, + "image_masks_loss": 0.0022, + "llm_learning_rate": 2.5999421770171616e-06, + "loss": 0.7911, + "output_text_mask_loss": 0.0656, + "predictor_learning_rate": 0.0002599942177017162, + "step": 4906 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3143, + "image_masks_loss": 0.0184, + "llm_learning_rate": 2.599113217985206e-06, + "loss": 0.7498, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00025991132179852054, + "step": 4907 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3142, + "image_masks_loss": 0.0187, + "llm_learning_rate": 2.598284248038682e-06, + "loss": 0.7716, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002598284248038682, + "step": 4908 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3141, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.597455267268877e-06, + "loss": 0.7291, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025974552672688767, + "step": 4909 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3141, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.596626275767082e-06, + "loss": 0.804, + "output_text_mask_loss": 0.0916, + "predictor_learning_rate": 0.0002596626275767082, + "step": 4910 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.314, + "image_masks_loss": 0.0148, + "llm_learning_rate": 2.595797273624585e-06, + "loss": 0.7351, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0002595797273624585, + "step": 4911 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3139, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.59496826093268e-06, + "loss": 0.7648, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025949682609326797, + "step": 4912 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3138, + "image_masks_loss": 0.0196, + "llm_learning_rate": 2.5941392377826587e-06, + "loss": 0.7436, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00025941392377826586, + "step": 4913 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3138, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.5933102042658137e-06, + "loss": 0.6514, + "output_text_mask_loss": 0.0067, + "predictor_learning_rate": 0.00025933102042658134, + "step": 4914 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3137, + "image_masks_loss": 0.0239, + "llm_learning_rate": 2.592481160473442e-06, + "loss": 0.71, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00025924811604734416, + "step": 4915 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3136, + "image_masks_loss": 0.0188, + "llm_learning_rate": 2.591652106496838e-06, + "loss": 0.7151, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0002591652106496838, + "step": 4916 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3136, + "image_masks_loss": 0.005, + "llm_learning_rate": 2.5908230424273008e-06, + "loss": 0.7168, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.0002590823042427301, + "step": 4917 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3135, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.5899939683561277e-06, + "loss": 0.6987, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025899939683561276, + "step": 4918 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3134, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.589164884374618e-06, + "loss": 0.7439, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002589164884374618, + "step": 4919 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3133, + "image_masks_loss": 0.015, + "llm_learning_rate": 2.5883357905740735e-06, + "loss": 0.6533, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.00025883357905740736, + "step": 4920 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3133, + "image_masks_loss": 0.0263, + "llm_learning_rate": 2.5875066870457936e-06, + "loss": 0.7747, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00025875066870457937, + "step": 4921 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3132, + "image_masks_loss": 0.0043, + "llm_learning_rate": 2.5866775738810833e-06, + "loss": 0.7169, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.0002586677573881083, + "step": 4922 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3131, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.5858484511712455e-06, + "loss": 0.7268, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002585848451171245, + "step": 4923 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.313, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.585019319007585e-06, + "loss": 0.687, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0002585019319007585, + "step": 4924 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.313, + "image_masks_loss": 0.0238, + "llm_learning_rate": 2.584190177481408e-06, + "loss": 0.7754, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00025841901774814076, + "step": 4925 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3129, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.58336102668402e-06, + "loss": 0.711, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000258336102668402, + "step": 4926 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3128, + "image_masks_loss": 0.0059, + "llm_learning_rate": 2.5825318667067313e-06, + "loss": 0.8388, + "output_text_mask_loss": 0.0656, + "predictor_learning_rate": 0.0002582531866706731, + "step": 4927 + }, + { + "epoch": 0.5, + "gumbel_tau": 0.3127, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.5817026976408487e-06, + "loss": 0.8337, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0002581702697640849, + "step": 4928 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3127, + "image_masks_loss": 0.0119, + "llm_learning_rate": 2.5808735195776836e-06, + "loss": 0.6455, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002580873519577683, + "step": 4929 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3126, + "image_masks_loss": 0.0177, + "llm_learning_rate": 2.5800443326085467e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025800443326085465, + "step": 4930 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3125, + "image_masks_loss": 0.036, + "llm_learning_rate": 2.579215136824749e-06, + "loss": 0.6947, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00025792151368247485, + "step": 4931 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3124, + "image_masks_loss": 0.028, + "llm_learning_rate": 2.5783859323176043e-06, + "loss": 0.6942, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0002578385932317604, + "step": 4932 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3124, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.5775567191784257e-06, + "loss": 0.7219, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0002577556719178426, + "step": 4933 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3123, + "image_masks_loss": 0.0181, + "llm_learning_rate": 2.5767274974985292e-06, + "loss": 0.7001, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0002576727497498529, + "step": 4934 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3122, + "image_masks_loss": 0.0353, + "llm_learning_rate": 2.5758982673692284e-06, + "loss": 0.7457, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.0002575898267369228, + "step": 4935 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3122, + "image_masks_loss": 0.0259, + "llm_learning_rate": 2.5750690288818426e-06, + "loss": 0.7377, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00025750690288818426, + "step": 4936 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3121, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.5742397821276876e-06, + "loss": 0.6746, + "output_text_mask_loss": 0.0119, + "predictor_learning_rate": 0.00025742397821276877, + "step": 4937 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.312, + "image_masks_loss": 0.0148, + "llm_learning_rate": 2.5734105271980824e-06, + "loss": 0.7984, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00025734105271980824, + "step": 4938 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3119, + "image_masks_loss": 0.0162, + "llm_learning_rate": 2.5725812641843474e-06, + "loss": 0.8297, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.0002572581264184347, + "step": 4939 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3119, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.571751993177801e-06, + "loss": 0.7804, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002571751993177801, + "step": 4940 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3118, + "image_masks_loss": 0.0048, + "llm_learning_rate": 2.570922714269765e-06, + "loss": 0.8067, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0002570922714269765, + "step": 4941 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3117, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.570093427551562e-06, + "loss": 0.7877, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00025700934275515617, + "step": 4942 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3116, + "image_masks_loss": 0.0043, + "llm_learning_rate": 2.569264133114515e-06, + "loss": 0.7094, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0002569264133114515, + "step": 4943 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3116, + "image_masks_loss": 0.0435, + "llm_learning_rate": 2.5684348310499467e-06, + "loss": 0.639, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.0002568434831049947, + "step": 4944 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3115, + "image_masks_loss": 0.0094, + "llm_learning_rate": 2.567605521449183e-06, + "loss": 0.7839, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002567605521449183, + "step": 4945 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3114, + "image_masks_loss": 0.0205, + "llm_learning_rate": 2.5667762044035487e-06, + "loss": 0.7521, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002566776204403549, + "step": 4946 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3113, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.5659468800043704e-06, + "loss": 0.744, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.000256594688000437, + "step": 4947 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3113, + "image_masks_loss": 0.0094, + "llm_learning_rate": 2.565117548342974e-06, + "loss": 0.7364, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0002565117548342974, + "step": 4948 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3112, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.5642882095106885e-06, + "loss": 0.6679, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00025642882095106885, + "step": 4949 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3111, + "image_masks_loss": 0.0154, + "llm_learning_rate": 2.563458863598842e-06, + "loss": 0.7544, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002563458863598842, + "step": 4950 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.311, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.562629510698764e-06, + "loss": 0.7118, + "output_text_mask_loss": 0.0206, + "predictor_learning_rate": 0.0002562629510698764, + "step": 4951 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.311, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.561800150901785e-06, + "loss": 0.7595, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025618001509017846, + "step": 4952 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3109, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.560970784299236e-06, + "loss": 0.7052, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002560970784299236, + "step": 4953 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3108, + "image_masks_loss": 0.0234, + "llm_learning_rate": 2.560141410982447e-06, + "loss": 0.7533, + "output_text_mask_loss": 0.1221, + "predictor_learning_rate": 0.00025601414109824473, + "step": 4954 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3108, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.5593120310427535e-06, + "loss": 0.6757, + "output_text_mask_loss": 0.0094, + "predictor_learning_rate": 0.0002559312031042753, + "step": 4955 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3107, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.5584826445714867e-06, + "loss": 0.7367, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00025584826445714865, + "step": 4956 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3106, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.5576532516599806e-06, + "loss": 0.7445, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025576532516599805, + "step": 4957 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3105, + "image_masks_loss": 0.0298, + "llm_learning_rate": 2.55682385239957e-06, + "loss": 0.7315, + "output_text_mask_loss": 0.0622, + "predictor_learning_rate": 0.000255682385239957, + "step": 4958 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3105, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.55599444688159e-06, + "loss": 0.7787, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.000255599444688159, + "step": 4959 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3104, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.5551650351973765e-06, + "loss": 0.7277, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025551650351973763, + "step": 4960 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3103, + "image_masks_loss": 0.0205, + "llm_learning_rate": 2.5543356174382673e-06, + "loss": 0.7611, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0002554335617438267, + "step": 4961 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3102, + "image_masks_loss": 0.0435, + "llm_learning_rate": 2.5535061936955984e-06, + "loss": 0.7344, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.0002553506193695598, + "step": 4962 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3102, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.5526767640607087e-06, + "loss": 0.6118, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025526767640607085, + "step": 4963 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3101, + "image_masks_loss": 0.0288, + "llm_learning_rate": 2.551847328624936e-06, + "loss": 0.6552, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00025518473286249355, + "step": 4964 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.31, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.5510178874796197e-06, + "loss": 0.7655, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00025510178874796195, + "step": 4965 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3099, + "image_masks_loss": 0.0256, + "llm_learning_rate": 2.550188440716101e-06, + "loss": 0.6958, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00025501884407161005, + "step": 4966 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3099, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.5493589884257187e-06, + "loss": 0.7468, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00025493589884257185, + "step": 4967 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3098, + "image_masks_loss": 0.0027, + "llm_learning_rate": 2.548529530699815e-06, + "loss": 0.6701, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00025485295306998146, + "step": 4968 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3097, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.547700067629731e-06, + "loss": 0.757, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0002547700067629731, + "step": 4969 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3097, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.5468705993068088e-06, + "loss": 0.7519, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002546870599306809, + "step": 4970 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3096, + "image_masks_loss": 0.0105, + "llm_learning_rate": 2.546041125822392e-06, + "loss": 0.817, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002546041125822392, + "step": 4971 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3095, + "image_masks_loss": 0.0211, + "llm_learning_rate": 2.5452116472678244e-06, + "loss": 0.7388, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.00025452116472678246, + "step": 4972 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3094, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.5443821637344495e-06, + "loss": 0.7406, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00025443821637344494, + "step": 4973 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3094, + "image_masks_loss": 0.0089, + "llm_learning_rate": 2.5435526753136115e-06, + "loss": 0.8067, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0002543552675313611, + "step": 4974 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3093, + "image_masks_loss": 0.0233, + "llm_learning_rate": 2.5427231820966565e-06, + "loss": 0.8265, + "output_text_mask_loss": 0.0668, + "predictor_learning_rate": 0.00025427231820966564, + "step": 4975 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3092, + "image_masks_loss": 0.0221, + "llm_learning_rate": 2.5418936841749285e-06, + "loss": 0.8004, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 0.00025418936841749284, + "step": 4976 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3091, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.541064181639776e-06, + "loss": 0.7984, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00025410641816397755, + "step": 4977 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3091, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.5402346745825436e-06, + "loss": 0.7225, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002540234674582543, + "step": 4978 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.309, + "image_masks_loss": 0.0158, + "llm_learning_rate": 2.5394051630945793e-06, + "loss": 0.7148, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00025394051630945794, + "step": 4979 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3089, + "image_masks_loss": 0.0144, + "llm_learning_rate": 2.53857564726723e-06, + "loss": 0.7311, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000253857564726723, + "step": 4980 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3089, + "image_masks_loss": 0.0133, + "llm_learning_rate": 2.5377461271918453e-06, + "loss": 0.583, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002537746127191845, + "step": 4981 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3088, + "image_masks_loss": 0.0261, + "llm_learning_rate": 2.536916602959773e-06, + "loss": 0.7446, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0002536916602959773, + "step": 4982 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3087, + "image_masks_loss": 0.0171, + "llm_learning_rate": 2.536087074662361e-06, + "loss": 0.6845, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.0002536087074662361, + "step": 4983 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3086, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.5352575423909616e-06, + "loss": 0.7343, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00025352575423909614, + "step": 4984 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3086, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.534428006236923e-06, + "loss": 0.7936, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0002534428006236923, + "step": 4985 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3085, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.533598466291595e-06, + "loss": 0.724, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00025335984662915945, + "step": 4986 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3084, + "image_masks_loss": 0.0211, + "llm_learning_rate": 2.5327689226463287e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.00025327689226463287, + "step": 4987 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3083, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.5319393753924766e-06, + "loss": 0.7319, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025319393753924765, + "step": 4988 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3083, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.5311098246213883e-06, + "loss": 0.7474, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0002531109824621388, + "step": 4989 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3082, + "image_masks_loss": 0.0098, + "llm_learning_rate": 2.5302802704244172e-06, + "loss": 0.7594, + "output_text_mask_loss": 0.0303, + "predictor_learning_rate": 0.0002530280270424417, + "step": 4990 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3081, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.529450712892916e-06, + "loss": 0.8525, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 0.00025294507128929155, + "step": 4991 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3081, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.528621152118236e-06, + "loss": 0.7209, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.00025286211521182355, + "step": 4992 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.308, + "image_masks_loss": 0.0163, + "llm_learning_rate": 2.5277915881917303e-06, + "loss": 0.7524, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.000252779158819173, + "step": 4993 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3079, + "image_masks_loss": 0.0217, + "llm_learning_rate": 2.526962021204754e-06, + "loss": 0.6946, + "output_text_mask_loss": 0.0043, + "predictor_learning_rate": 0.00025269620212047537, + "step": 4994 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3078, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.52613245124866e-06, + "loss": 0.7289, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.000252613245124866, + "step": 4995 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3078, + "image_masks_loss": 0.0347, + "llm_learning_rate": 2.525302878414802e-06, + "loss": 0.6649, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0002525302878414802, + "step": 4996 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3077, + "image_masks_loss": 0.0175, + "llm_learning_rate": 2.5244733027945343e-06, + "loss": 0.7689, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00025244733027945346, + "step": 4997 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3076, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.5236437244792128e-06, + "loss": 0.749, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0002523643724479213, + "step": 4998 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3075, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.5228141435601907e-06, + "loss": 0.7926, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.00025228141435601904, + "step": 4999 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3075, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.5219845601288245e-06, + "loss": 0.7409, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00025219845601288243, + "step": 5000 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3074, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.5211549742764703e-06, + "loss": 0.7996, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.000252115497427647, + "step": 5001 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3073, + "image_masks_loss": 0.0142, + "llm_learning_rate": 2.5203253860944825e-06, + "loss": 0.7127, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00025203253860944824, + "step": 5002 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3073, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.5194957956742178e-06, + "loss": 0.6833, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00025194957956742173, + "step": 5003 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3072, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.5186662031070337e-06, + "loss": 0.7433, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00025186662031070336, + "step": 5004 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3071, + "image_masks_loss": 0.0101, + "llm_learning_rate": 2.5178366084842852e-06, + "loss": 0.7311, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002517836608484285, + "step": 5005 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.307, + "image_masks_loss": 0.0063, + "llm_learning_rate": 2.517007011897329e-06, + "loss": 0.7255, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.0002517007011897329, + "step": 5006 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.307, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.5161774134375233e-06, + "loss": 0.7495, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002516177413437523, + "step": 5007 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3069, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.515347813196225e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002515347813196225, + "step": 5008 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3068, + "image_masks_loss": 0.0246, + "llm_learning_rate": 2.514518211264791e-06, + "loss": 0.8367, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00025145182112647907, + "step": 5009 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3067, + "image_masks_loss": 0.0257, + "llm_learning_rate": 2.513688607734579e-06, + "loss": 0.7917, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0002513688607734579, + "step": 5010 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3067, + "image_masks_loss": 0.0183, + "llm_learning_rate": 2.5128590026969474e-06, + "loss": 0.7675, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00025128590026969475, + "step": 5011 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3066, + "image_masks_loss": 0.0154, + "llm_learning_rate": 2.512029396243254e-06, + "loss": 0.7254, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0002512029396243254, + "step": 5012 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3065, + "image_masks_loss": 0.0089, + "llm_learning_rate": 2.5111997884648566e-06, + "loss": 0.6877, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00025111997884648564, + "step": 5013 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3065, + "image_masks_loss": 0.009, + "llm_learning_rate": 2.510370179453113e-06, + "loss": 0.687, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002510370179453113, + "step": 5014 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3064, + "image_masks_loss": 0.0076, + "llm_learning_rate": 2.5095405692993825e-06, + "loss": 0.7121, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 0.0002509540569299382, + "step": 5015 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3063, + "image_masks_loss": 0.0226, + "llm_learning_rate": 2.5087109580950235e-06, + "loss": 0.7598, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.00025087109580950237, + "step": 5016 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3062, + "image_masks_loss": 0.0149, + "llm_learning_rate": 2.5078813459313938e-06, + "loss": 0.7906, + "output_text_mask_loss": 0.0774, + "predictor_learning_rate": 0.00025078813459313937, + "step": 5017 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3062, + "image_masks_loss": 0.0254, + "llm_learning_rate": 2.5070517328998534e-06, + "loss": 0.6504, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 0.0002507051732899853, + "step": 5018 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3061, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.5062221190917596e-06, + "loss": 0.716, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.00025062221190917596, + "step": 5019 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.306, + "image_masks_loss": 0.0048, + "llm_learning_rate": 2.505392504598473e-06, + "loss": 0.6383, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0002505392504598473, + "step": 5020 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.306, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.5045628895113518e-06, + "loss": 0.7364, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.00025045628895113513, + "step": 5021 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3059, + "image_masks_loss": 0.0201, + "llm_learning_rate": 2.5037332739217545e-06, + "loss": 0.8152, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.00025037332739217546, + "step": 5022 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3058, + "image_masks_loss": 0.0169, + "llm_learning_rate": 2.5029036579210414e-06, + "loss": 0.6681, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025029036579210414, + "step": 5023 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3057, + "image_masks_loss": 0.017, + "llm_learning_rate": 2.502074041600571e-06, + "loss": 0.8089, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00025020740416005705, + "step": 5024 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3057, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.5012444250517016e-06, + "loss": 0.7163, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00025012444250517015, + "step": 5025 + }, + { + "epoch": 0.51, + "gumbel_tau": 0.3056, + "image_masks_loss": 0.0203, + "llm_learning_rate": 2.5004148083657943e-06, + "loss": 0.7464, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00025004148083657943, + "step": 5026 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3055, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.4995851916342074e-06, + "loss": 0.708, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0002499585191634207, + "step": 5027 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3054, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.4987555749482992e-06, + "loss": 0.7244, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.0002498755574948299, + "step": 5028 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3054, + "image_masks_loss": 0.0098, + "llm_learning_rate": 2.4979259583994304e-06, + "loss": 0.7479, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00024979259583994307, + "step": 5029 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3053, + "image_masks_loss": 0.0212, + "llm_learning_rate": 2.49709634207896e-06, + "loss": 0.7351, + "output_text_mask_loss": 0.0024, + "predictor_learning_rate": 0.000249709634207896, + "step": 5030 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3052, + "image_masks_loss": 0.0129, + "llm_learning_rate": 2.4962667260782463e-06, + "loss": 0.7338, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0002496266726078246, + "step": 5031 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3052, + "image_masks_loss": 0.0217, + "llm_learning_rate": 2.495437110488649e-06, + "loss": 0.763, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0002495437110488649, + "step": 5032 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3051, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.4946074954015275e-06, + "loss": 0.7536, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0002494607495401527, + "step": 5033 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.305, + "image_masks_loss": 0.0148, + "llm_learning_rate": 2.4937778809082408e-06, + "loss": 0.8001, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00024937778809082405, + "step": 5034 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3049, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.4929482671001474e-06, + "loss": 0.696, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00024929482671001474, + "step": 5035 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3049, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.4921186540686066e-06, + "loss": 0.6863, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00024921186540686064, + "step": 5036 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3048, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.4912890419049773e-06, + "loss": 0.7967, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002491289041904977, + "step": 5037 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3047, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.4904594307006184e-06, + "loss": 0.7398, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.0002490459430700618, + "step": 5038 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3047, + "image_masks_loss": 0.0234, + "llm_learning_rate": 2.4896298205468873e-06, + "loss": 0.739, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.0002489629820546887, + "step": 5039 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3046, + "image_masks_loss": 0.0115, + "llm_learning_rate": 2.4888002115351442e-06, + "loss": 0.8094, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.0002488800211535144, + "step": 5040 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3045, + "image_masks_loss": 0.0178, + "llm_learning_rate": 2.4879706037567468e-06, + "loss": 0.7316, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.00024879706037567466, + "step": 5041 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3044, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.4871409973030525e-06, + "loss": 0.7237, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0002487140997303052, + "step": 5042 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3044, + "image_masks_loss": 0.0176, + "llm_learning_rate": 2.4863113922654212e-06, + "loss": 0.6689, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.0002486311392265421, + "step": 5043 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3043, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.4854817887352095e-06, + "loss": 0.7483, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00024854817887352094, + "step": 5044 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3042, + "image_masks_loss": 0.0065, + "llm_learning_rate": 2.4846521868037755e-06, + "loss": 0.8172, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.00024846521868037753, + "step": 5045 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3042, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.483822586562478e-06, + "loss": 0.8194, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0002483822586562478, + "step": 5046 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3041, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.482992988102672e-06, + "loss": 0.7349, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00024829929881026717, + "step": 5047 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.304, + "image_masks_loss": 0.0248, + "llm_learning_rate": 2.482163391515716e-06, + "loss": 0.8564, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.0002482163391515716, + "step": 5048 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3039, + "image_masks_loss": 0.0043, + "llm_learning_rate": 2.481333796892967e-06, + "loss": 0.7279, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0002481333796892967, + "step": 5049 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3039, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.480504204325783e-06, + "loss": 0.755, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.0002480504204325783, + "step": 5050 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3038, + "image_masks_loss": 0.0256, + "llm_learning_rate": 2.479674613905518e-06, + "loss": 0.8003, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.0002479674613905518, + "step": 5051 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3037, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.4788450257235305e-06, + "loss": 0.8115, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00024788450257235306, + "step": 5052 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3037, + "image_masks_loss": 0.0224, + "llm_learning_rate": 2.478015439871176e-06, + "loss": 0.7535, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002478015439871176, + "step": 5053 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3036, + "image_masks_loss": 0.0148, + "llm_learning_rate": 2.47718585643981e-06, + "loss": 0.8668, + "output_text_mask_loss": 0.0507, + "predictor_learning_rate": 0.000247718585643981, + "step": 5054 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3035, + "image_masks_loss": 0.0203, + "llm_learning_rate": 2.476356275520788e-06, + "loss": 0.8573, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.0002476356275520788, + "step": 5055 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3034, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.475526697205466e-06, + "loss": 0.7188, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002475526697205466, + "step": 5056 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3034, + "image_masks_loss": 0.0101, + "llm_learning_rate": 2.474697121585199e-06, + "loss": 0.7164, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00024746971215851987, + "step": 5057 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3033, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.4738675487513404e-06, + "loss": 0.7388, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.000247386754875134, + "step": 5058 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3032, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.4730379787952464e-06, + "loss": 0.6948, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00024730379787952464, + "step": 5059 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3032, + "image_masks_loss": 0.0166, + "llm_learning_rate": 2.4722084118082697e-06, + "loss": 0.6639, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00024722084118082696, + "step": 5060 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3031, + "image_masks_loss": 0.0089, + "llm_learning_rate": 2.471378847881765e-06, + "loss": 0.7964, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 0.00024713788478817646, + "step": 5061 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.303, + "image_masks_loss": 0.0098, + "llm_learning_rate": 2.4705492871070845e-06, + "loss": 0.8133, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00024705492871070846, + "step": 5062 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3029, + "image_masks_loss": 0.0086, + "llm_learning_rate": 2.4697197295755827e-06, + "loss": 0.7337, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002469719729575583, + "step": 5063 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3029, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.4688901753786125e-06, + "loss": 0.6794, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.0002468890175378613, + "step": 5064 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3028, + "image_masks_loss": 0.0166, + "llm_learning_rate": 2.468060624607525e-06, + "loss": 0.6984, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00024680606246075247, + "step": 5065 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3027, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.4672310773536725e-06, + "loss": 0.7672, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 0.00024672310773536725, + "step": 5066 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3027, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.4664015337084064e-06, + "loss": 0.7445, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002466401533708406, + "step": 5067 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3026, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.4655719937630785e-06, + "loss": 0.7629, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00024655719937630784, + "step": 5068 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3025, + "image_masks_loss": 0.0175, + "llm_learning_rate": 2.464742457609039e-06, + "loss": 0.7942, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0002464742457609039, + "step": 5069 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3024, + "image_masks_loss": 0.0151, + "llm_learning_rate": 2.463912925337639e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0002463912925337639, + "step": 5070 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3024, + "image_masks_loss": 0.0099, + "llm_learning_rate": 2.463083397040228e-06, + "loss": 0.826, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00024630833970402277, + "step": 5071 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3023, + "image_masks_loss": 0.0138, + "llm_learning_rate": 2.4622538728081555e-06, + "loss": 0.7585, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0002462253872808155, + "step": 5072 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3022, + "image_masks_loss": 0.0028, + "llm_learning_rate": 2.4614243527327704e-06, + "loss": 0.694, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00024614243527327703, + "step": 5073 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3022, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.4605948369054216e-06, + "loss": 0.7518, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002460594836905421, + "step": 5074 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3021, + "image_masks_loss": 0.0226, + "llm_learning_rate": 2.4597653254174573e-06, + "loss": 0.7481, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0002459765325417457, + "step": 5075 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.302, + "image_masks_loss": 0.0175, + "llm_learning_rate": 2.458935818360225e-06, + "loss": 0.7827, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00024589358183602246, + "step": 5076 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3019, + "image_masks_loss": 0.0119, + "llm_learning_rate": 2.458106315825072e-06, + "loss": 0.7668, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00024581063158250717, + "step": 5077 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3019, + "image_masks_loss": 0.0158, + "llm_learning_rate": 2.457276817903344e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00024572768179033437, + "step": 5078 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3018, + "image_masks_loss": 0.0176, + "llm_learning_rate": 2.456447324686389e-06, + "loss": 0.6904, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.00024564473246863885, + "step": 5079 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3017, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.455617836265551e-06, + "loss": 0.878, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002455617836265551, + "step": 5080 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3017, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.454788352732175e-06, + "loss": 0.6808, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0002454788352732175, + "step": 5081 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3016, + "image_masks_loss": 0.0142, + "llm_learning_rate": 2.453958874177608e-06, + "loss": 0.7737, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002453958874177608, + "step": 5082 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3015, + "image_masks_loss": 0.0288, + "llm_learning_rate": 2.453129400693192e-06, + "loss": 0.7662, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0002453129400693192, + "step": 5083 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3014, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.4522999323702703e-06, + "loss": 0.7069, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00024522999323702704, + "step": 5084 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3014, + "image_masks_loss": 0.0229, + "llm_learning_rate": 2.4514704693001864e-06, + "loss": 0.7937, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 0.0002451470469300186, + "step": 5085 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3013, + "image_masks_loss": 0.0233, + "llm_learning_rate": 2.4506410115742825e-06, + "loss": 0.7998, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00024506410115742827, + "step": 5086 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3012, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.4498115592839e-06, + "loss": 0.6955, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00024498115592839, + "step": 5087 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3012, + "image_masks_loss": 0.0137, + "llm_learning_rate": 2.4489821125203807e-06, + "loss": 0.8025, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.00024489821125203806, + "step": 5088 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3011, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.448152671375065e-06, + "loss": 0.7044, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0002448152671375065, + "step": 5089 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.301, + "image_masks_loss": 0.0041, + "llm_learning_rate": 2.447323235939292e-06, + "loss": 0.7636, + "output_text_mask_loss": 0.053, + "predictor_learning_rate": 0.0002447323235939292, + "step": 5090 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3009, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.446493806304402e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0176, + "predictor_learning_rate": 0.0002446493806304402, + "step": 5091 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3009, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.445664382561733e-06, + "loss": 0.7112, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.0002445664382561733, + "step": 5092 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3008, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.444834964802624e-06, + "loss": 0.7663, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0002444834964802624, + "step": 5093 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3007, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.4440055531184105e-06, + "loss": 0.7675, + "output_text_mask_loss": 0.0227, + "predictor_learning_rate": 0.00024440055531184103, + "step": 5094 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3007, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.4431761476004305e-06, + "loss": 0.7634, + "output_text_mask_loss": 0.0334, + "predictor_learning_rate": 0.00024431761476004303, + "step": 5095 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3006, + "image_masks_loss": 0.0214, + "llm_learning_rate": 2.4423467483400203e-06, + "loss": 0.7669, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.000244234674834002, + "step": 5096 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3005, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.4415173554285138e-06, + "loss": 0.7768, + "output_text_mask_loss": 0.0423, + "predictor_learning_rate": 0.00024415173554285136, + "step": 5097 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3004, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.440687968957247e-06, + "loss": 0.7647, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00024406879689572468, + "step": 5098 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3004, + "image_masks_loss": 0.0265, + "llm_learning_rate": 2.439858589017553e-06, + "loss": 0.6709, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00024398585890175525, + "step": 5099 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3003, + "image_masks_loss": 0.0075, + "llm_learning_rate": 2.439029215700765e-06, + "loss": 0.7658, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0002439029215700765, + "step": 5100 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3002, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.4381998490982164e-06, + "loss": 0.6709, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.00024381998490982163, + "step": 5101 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3002, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.4373704893012373e-06, + "loss": 0.6957, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002437370489301237, + "step": 5102 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3001, + "image_masks_loss": 0.0055, + "llm_learning_rate": 2.436541136401159e-06, + "loss": 0.8017, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.0002436541136401159, + "step": 5103 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.4357117904893127e-06, + "loss": 0.7339, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00024357117904893124, + "step": 5104 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.3, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.434882451657027e-06, + "loss": 0.7099, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002434882451657027, + "step": 5105 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2999, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.4340531199956305e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.00024340531199956304, + "step": 5106 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2998, + "image_masks_loss": 0.0141, + "llm_learning_rate": 2.433223795596452e-06, + "loss": 0.844, + "output_text_mask_loss": 0.0568, + "predictor_learning_rate": 0.00024332237955964516, + "step": 5107 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2997, + "image_masks_loss": 0.0044, + "llm_learning_rate": 2.4323944785508173e-06, + "loss": 0.763, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.00024323944785508173, + "step": 5108 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2997, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.4315651689500538e-06, + "loss": 0.7287, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00024315651689500538, + "step": 5109 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2996, + "image_masks_loss": 0.0418, + "llm_learning_rate": 2.4307358668854854e-06, + "loss": 0.7357, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00024307358668854853, + "step": 5110 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2995, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.4299065724484382e-06, + "loss": 0.8582, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002429906572448438, + "step": 5111 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2995, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.4290772857302354e-06, + "loss": 0.762, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.00024290772857302355, + "step": 5112 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2994, + "image_masks_loss": 0.0076, + "llm_learning_rate": 2.4282480068222e-06, + "loss": 0.7713, + "output_text_mask_loss": 0.0212, + "predictor_learning_rate": 0.00024282480068221995, + "step": 5113 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2993, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.4274187358156535e-06, + "loss": 0.7252, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00024274187358156533, + "step": 5114 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2992, + "image_masks_loss": 0.0159, + "llm_learning_rate": 2.426589472801918e-06, + "loss": 0.8142, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00024265894728019177, + "step": 5115 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2992, + "image_masks_loss": 0.0159, + "llm_learning_rate": 2.4257602178723128e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00024257602178723127, + "step": 5116 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2991, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.4249309711181578e-06, + "loss": 0.8303, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.00024249309711181575, + "step": 5117 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.299, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.4241017326307716e-06, + "loss": 0.8375, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00024241017326307712, + "step": 5118 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.299, + "image_masks_loss": 0.019, + "llm_learning_rate": 2.4232725025014716e-06, + "loss": 0.7781, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00024232725025014716, + "step": 5119 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2989, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.422443280821575e-06, + "loss": 0.6354, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00024224432808215752, + "step": 5120 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2988, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.421614067682397e-06, + "loss": 0.6048, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00024216140676823968, + "step": 5121 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2987, + "image_masks_loss": 0.0029, + "llm_learning_rate": 2.4207848631752516e-06, + "loss": 0.8468, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00024207848631752516, + "step": 5122 + }, + { + "epoch": 0.52, + "gumbel_tau": 0.2987, + "image_masks_loss": 0.0259, + "llm_learning_rate": 2.4199556673914545e-06, + "loss": 0.7375, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 0.00024199556673914544, + "step": 5123 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2986, + "image_masks_loss": 0.0181, + "llm_learning_rate": 2.4191264804223173e-06, + "loss": 0.7495, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0002419126480422317, + "step": 5124 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2985, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.418297302359152e-06, + "loss": 0.7334, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002418297302359152, + "step": 5125 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2985, + "image_masks_loss": 0.0151, + "llm_learning_rate": 2.4174681332932696e-06, + "loss": 0.772, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00024174681332932694, + "step": 5126 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2984, + "image_masks_loss": 0.0024, + "llm_learning_rate": 2.4166389733159802e-06, + "loss": 0.797, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00024166389733159803, + "step": 5127 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2983, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.4158098225185935e-06, + "loss": 0.7136, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0002415809822518593, + "step": 5128 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2983, + "image_masks_loss": 0.0147, + "llm_learning_rate": 2.4149806809924155e-06, + "loss": 0.6813, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002414980680992415, + "step": 5129 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2982, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.414151548828755e-06, + "loss": 0.8023, + "output_text_mask_loss": 0.041, + "predictor_learning_rate": 0.00024141515488287546, + "step": 5130 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2981, + "image_masks_loss": 0.0193, + "llm_learning_rate": 2.413322426118917e-06, + "loss": 0.7765, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00024133224261189168, + "step": 5131 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.298, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.412493312954207e-06, + "loss": 0.6628, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00024124933129542067, + "step": 5132 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.298, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.411664209425927e-06, + "loss": 0.7327, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002411664209425927, + "step": 5133 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2979, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.4108351156253825e-06, + "loss": 0.7944, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.00024108351156253822, + "step": 5134 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2978, + "image_masks_loss": 0.0158, + "llm_learning_rate": 2.4100060316438727e-06, + "loss": 0.7947, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00024100060316438725, + "step": 5135 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2978, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.409176957572699e-06, + "loss": 0.7448, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0002409176957572699, + "step": 5136 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2977, + "image_masks_loss": 0.0084, + "llm_learning_rate": 2.4083478935031616e-06, + "loss": 0.8553, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00024083478935031617, + "step": 5137 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2976, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.4075188395265594e-06, + "loss": 0.6131, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0002407518839526559, + "step": 5138 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2976, + "image_masks_loss": 0.0091, + "llm_learning_rate": 2.406689795734187e-06, + "loss": 0.7849, + "output_text_mask_loss": 0.0462, + "predictor_learning_rate": 0.0002406689795734187, + "step": 5139 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2975, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.405860762217343e-06, + "loss": 0.6992, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.0002405860762217343, + "step": 5140 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2974, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.4050317390673212e-06, + "loss": 0.7973, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002405031739067321, + "step": 5141 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2973, + "image_masks_loss": 0.015, + "llm_learning_rate": 2.4042027263754153e-06, + "loss": 0.7872, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.0002404202726375415, + "step": 5142 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2973, + "image_masks_loss": 0.0192, + "llm_learning_rate": 2.403373724232919e-06, + "loss": 0.7196, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0002403373724232919, + "step": 5143 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2972, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.4025447327311236e-06, + "loss": 0.819, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.00024025447327311234, + "step": 5144 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2971, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.4017157519613187e-06, + "loss": 0.7472, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 0.00024017157519613187, + "step": 5145 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2971, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.4008867820147946e-06, + "loss": 0.6858, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00024008867820147944, + "step": 5146 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.297, + "image_masks_loss": 0.0236, + "llm_learning_rate": 2.400057822982839e-06, + "loss": 0.7793, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00024000578229828388, + "step": 5147 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2969, + "image_masks_loss": 0.0037, + "llm_learning_rate": 2.3992288749567383e-06, + "loss": 0.7381, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00023992288749567382, + "step": 5148 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2969, + "image_masks_loss": 0.0054, + "llm_learning_rate": 2.398399938027778e-06, + "loss": 0.7856, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.0002398399938027778, + "step": 5149 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2968, + "image_masks_loss": 0.0059, + "llm_learning_rate": 2.3975710122872433e-06, + "loss": 0.7709, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.00023975710122872433, + "step": 5150 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2967, + "image_masks_loss": 0.025, + "llm_learning_rate": 2.396742097826417e-06, + "loss": 0.7983, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023967420978264167, + "step": 5151 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2966, + "image_masks_loss": 0.0086, + "llm_learning_rate": 2.3959131947365803e-06, + "loss": 0.6955, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000239591319473658, + "step": 5152 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2966, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.3950843031090145e-06, + "loss": 0.7265, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00023950843031090145, + "step": 5153 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2965, + "image_masks_loss": 0.0162, + "llm_learning_rate": 2.3942554230349997e-06, + "loss": 0.6867, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00023942554230349994, + "step": 5154 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2964, + "image_masks_loss": 0.0232, + "llm_learning_rate": 2.3934265546058126e-06, + "loss": 0.7807, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00023934265546058126, + "step": 5155 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2964, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.3925976979127307e-06, + "loss": 0.6828, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00023925976979127304, + "step": 5156 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2963, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.3917688530470308e-06, + "loss": 0.6175, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00023917688530470304, + "step": 5157 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2962, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.3909400200999844e-06, + "loss": 0.653, + "output_text_mask_loss": 0.079, + "predictor_learning_rate": 0.00023909400200999842, + "step": 5158 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2962, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.390111199162866e-06, + "loss": 0.6978, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002390111199162866, + "step": 5159 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2961, + "image_masks_loss": 0.0147, + "llm_learning_rate": 2.3892823903269475e-06, + "loss": 0.7809, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00023892823903269475, + "step": 5160 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.296, + "image_masks_loss": 0.0299, + "llm_learning_rate": 2.3884535936834986e-06, + "loss": 0.8643, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00023884535936834984, + "step": 5161 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2959, + "image_masks_loss": 0.0148, + "llm_learning_rate": 2.3876248093237884e-06, + "loss": 0.7303, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00023876248093237886, + "step": 5162 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2959, + "image_masks_loss": 0.0277, + "llm_learning_rate": 2.3867960373390847e-06, + "loss": 0.7159, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.00023867960373390846, + "step": 5163 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2958, + "image_masks_loss": 0.0159, + "llm_learning_rate": 2.3859672778206537e-06, + "loss": 0.7343, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.00023859672778206534, + "step": 5164 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2957, + "image_masks_loss": 0.0076, + "llm_learning_rate": 2.3851385308597592e-06, + "loss": 0.8476, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00023851385308597592, + "step": 5165 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2957, + "image_masks_loss": 0.0059, + "llm_learning_rate": 2.3843097965476662e-06, + "loss": 0.7301, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00023843097965476663, + "step": 5166 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2956, + "image_masks_loss": 0.0042, + "llm_learning_rate": 2.383481074975636e-06, + "loss": 0.7541, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0002383481074975636, + "step": 5167 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2955, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.3826523662349295e-06, + "loss": 0.8536, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00023826523662349292, + "step": 5168 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2955, + "image_masks_loss": 0.0141, + "llm_learning_rate": 2.3818236704168053e-06, + "loss": 0.6857, + "output_text_mask_loss": 0.0389, + "predictor_learning_rate": 0.00023818236704168053, + "step": 5169 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2954, + "image_masks_loss": 0.0154, + "llm_learning_rate": 2.3809949876125224e-06, + "loss": 0.7395, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00023809949876125221, + "step": 5170 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2953, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.3801663179133368e-06, + "loss": 0.6935, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00023801663179133365, + "step": 5171 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2952, + "image_masks_loss": 0.0202, + "llm_learning_rate": 2.3793376614105018e-06, + "loss": 0.7475, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 0.00023793376614105016, + "step": 5172 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2952, + "image_masks_loss": 0.0216, + "llm_learning_rate": 2.3785090181952734e-06, + "loss": 0.7801, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00023785090181952732, + "step": 5173 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2951, + "image_masks_loss": 0.0196, + "llm_learning_rate": 2.3776803883589024e-06, + "loss": 0.8155, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.00023776803883589022, + "step": 5174 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.295, + "image_masks_loss": 0.0058, + "llm_learning_rate": 2.37685177199264e-06, + "loss": 0.7991, + "output_text_mask_loss": 0.0908, + "predictor_learning_rate": 0.000237685177199264, + "step": 5175 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.295, + "image_masks_loss": 0.0076, + "llm_learning_rate": 2.376023169187734e-06, + "loss": 0.7042, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0002376023169187734, + "step": 5176 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2949, + "image_masks_loss": 0.0359, + "llm_learning_rate": 2.3751945800354326e-06, + "loss": 0.7027, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00023751945800354324, + "step": 5177 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2948, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.3743660046269827e-06, + "loss": 0.7931, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00023743660046269825, + "step": 5178 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2948, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.373537443053628e-06, + "loss": 0.7375, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00023735374430536277, + "step": 5179 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2947, + "image_masks_loss": 0.0185, + "llm_learning_rate": 2.3727088954066117e-06, + "loss": 0.7326, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.00023727088954066114, + "step": 5180 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2946, + "image_masks_loss": 0.0053, + "llm_learning_rate": 2.3718803617771747e-06, + "loss": 0.6842, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023718803617771744, + "step": 5181 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2946, + "image_masks_loss": 0.0212, + "llm_learning_rate": 2.3710518422565583e-06, + "loss": 0.7699, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.0002371051842256558, + "step": 5182 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2945, + "image_masks_loss": 0.0254, + "llm_learning_rate": 2.3702233369360003e-06, + "loss": 0.7078, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00023702233369360002, + "step": 5183 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2944, + "image_masks_loss": 0.0171, + "llm_learning_rate": 2.369394845906737e-06, + "loss": 0.8319, + "output_text_mask_loss": 0.0296, + "predictor_learning_rate": 0.00023693948459067366, + "step": 5184 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2943, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.3685663692600043e-06, + "loss": 0.7316, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00023685663692600044, + "step": 5185 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2943, + "image_masks_loss": 0.0094, + "llm_learning_rate": 2.367737907087036e-06, + "loss": 0.7914, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0002367737907087036, + "step": 5186 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2942, + "image_masks_loss": 0.02, + "llm_learning_rate": 2.3669094594790647e-06, + "loss": 0.6931, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00023669094594790644, + "step": 5187 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2941, + "image_masks_loss": 0.0063, + "llm_learning_rate": 2.3660810265273188e-06, + "loss": 0.6602, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023660810265273185, + "step": 5188 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2941, + "image_masks_loss": 0.0254, + "llm_learning_rate": 2.3652526083230292e-06, + "loss": 0.7614, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002365252608323029, + "step": 5189 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.294, + "image_masks_loss": 0.0201, + "llm_learning_rate": 2.3644242049574224e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00023644242049574224, + "step": 5190 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2939, + "image_masks_loss": 0.0063, + "llm_learning_rate": 2.363595816521724e-06, + "loss": 0.7712, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0002363595816521724, + "step": 5191 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2939, + "image_masks_loss": 0.0131, + "llm_learning_rate": 2.3627674431071584e-06, + "loss": 0.831, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00023627674431071584, + "step": 5192 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2938, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.361939084804948e-06, + "loss": 0.7397, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 0.00023619390848049476, + "step": 5193 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2937, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.3611107417063133e-06, + "loss": 0.6743, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 0.00023611107417063128, + "step": 5194 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2936, + "image_masks_loss": 0.0197, + "llm_learning_rate": 2.3602824139024722e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00023602824139024723, + "step": 5195 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2936, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.3594541014846437e-06, + "loss": 0.7415, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00023594541014846436, + "step": 5196 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2935, + "image_masks_loss": 0.0217, + "llm_learning_rate": 2.358625804544042e-06, + "loss": 0.7843, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00023586258045440415, + "step": 5197 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2934, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.3577975231718813e-06, + "loss": 0.7669, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00023577975231718813, + "step": 5198 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2934, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.3569692574593746e-06, + "loss": 0.7626, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.00023569692574593746, + "step": 5199 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2933, + "image_masks_loss": 0.0201, + "llm_learning_rate": 2.3561410074977314e-06, + "loss": 0.692, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023561410074977312, + "step": 5200 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2932, + "image_masks_loss": 0.0043, + "llm_learning_rate": 2.355312773378161e-06, + "loss": 0.7951, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0002355312773378161, + "step": 5201 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2932, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.354484555191871e-06, + "loss": 0.7685, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00023544845551918707, + "step": 5202 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2931, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.3536563530300654e-06, + "loss": 0.7805, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00023536563530300654, + "step": 5203 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.293, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.3528281669839474e-06, + "loss": 0.8129, + "output_text_mask_loss": 0.0694, + "predictor_learning_rate": 0.00023528281669839475, + "step": 5204 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.293, + "image_masks_loss": 0.0098, + "llm_learning_rate": 2.3519999971447206e-06, + "loss": 0.6991, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.00023519999971447202, + "step": 5205 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2929, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.3511718436035835e-06, + "loss": 0.7416, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023511718436035835, + "step": 5206 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2928, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.350343706451734e-06, + "loss": 0.7635, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002350343706451734, + "step": 5207 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2927, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.3495155857803696e-06, + "loss": 0.8015, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00023495155857803694, + "step": 5208 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2927, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.3486874816806836e-06, + "loss": 0.673, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00023486874816806835, + "step": 5209 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2926, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.347859394243869e-06, + "loss": 0.8112, + "output_text_mask_loss": 0.0024, + "predictor_learning_rate": 0.0002347859394243869, + "step": 5210 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2925, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.3470313235611173e-06, + "loss": 0.7778, + "output_text_mask_loss": 0.0393, + "predictor_learning_rate": 0.0002347031323561117, + "step": 5211 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2925, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.3462032697236177e-06, + "loss": 0.8077, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00023462032697236176, + "step": 5212 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2924, + "image_masks_loss": 0.0084, + "llm_learning_rate": 2.3453752328225554e-06, + "loss": 0.7229, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023453752328225552, + "step": 5213 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2923, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.3445472129491173e-06, + "loss": 0.7613, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00023445472129491173, + "step": 5214 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2923, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.343719210194486e-06, + "loss": 0.6246, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0002343719210194486, + "step": 5215 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2922, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.3428912246498432e-06, + "loss": 0.8208, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.00023428912246498433, + "step": 5216 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2921, + "image_masks_loss": 0.0214, + "llm_learning_rate": 2.3420632564063692e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023420632564063692, + "step": 5217 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2921, + "image_masks_loss": 0.0196, + "llm_learning_rate": 2.3412353055552413e-06, + "loss": 0.8012, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00023412353055552414, + "step": 5218 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.292, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.3404073721876353e-06, + "loss": 0.7299, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00023404073721876352, + "step": 5219 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2919, + "image_masks_loss": 0.0209, + "llm_learning_rate": 2.339579456394724e-06, + "loss": 0.7901, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0002339579456394724, + "step": 5220 + }, + { + "epoch": 0.53, + "gumbel_tau": 0.2919, + "image_masks_loss": 0.0191, + "llm_learning_rate": 2.3387515582676812e-06, + "loss": 0.7839, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.0002338751558267681, + "step": 5221 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2918, + "image_masks_loss": 0.0207, + "llm_learning_rate": 2.3379236778976757e-06, + "loss": 0.6899, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00023379236778976755, + "step": 5222 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2917, + "image_masks_loss": 0.0055, + "llm_learning_rate": 2.3370958153758753e-06, + "loss": 0.7166, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002337095815375875, + "step": 5223 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2916, + "image_masks_loss": 0.0111, + "llm_learning_rate": 2.336267970793447e-06, + "loss": 0.7207, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002336267970793447, + "step": 5224 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2916, + "image_masks_loss": 0.0087, + "llm_learning_rate": 2.335440144241555e-06, + "loss": 0.6696, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 0.00023354401442415547, + "step": 5225 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2915, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.33461233581136e-06, + "loss": 0.7238, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023346123358113603, + "step": 5226 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2914, + "image_masks_loss": 0.0057, + "llm_learning_rate": 2.333784545594023e-06, + "loss": 0.698, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00023337845455940229, + "step": 5227 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2914, + "image_masks_loss": 0.0133, + "llm_learning_rate": 2.3329567736807025e-06, + "loss": 0.7895, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023329567736807025, + "step": 5228 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2913, + "image_masks_loss": 0.0175, + "llm_learning_rate": 2.3321290201625544e-06, + "loss": 0.8113, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.00023321290201625542, + "step": 5229 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2912, + "image_masks_loss": 0.0089, + "llm_learning_rate": 2.3313012851307314e-06, + "loss": 0.6949, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.00023313012851307313, + "step": 5230 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2912, + "image_masks_loss": 0.0027, + "llm_learning_rate": 2.3304735686763886e-06, + "loss": 0.7563, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023304735686763885, + "step": 5231 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2911, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.3296458708906728e-06, + "loss": 0.7678, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.00023296458708906725, + "step": 5232 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.291, + "image_masks_loss": 0.0052, + "llm_learning_rate": 2.328818191864733e-06, + "loss": 0.6869, + "output_text_mask_loss": 0.0576, + "predictor_learning_rate": 0.00023288181918647332, + "step": 5233 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.291, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.3279905316897157e-06, + "loss": 0.7837, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.00023279905316897154, + "step": 5234 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2909, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.327162890456764e-06, + "loss": 0.7174, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.0002327162890456764, + "step": 5235 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2908, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.326335268257019e-06, + "loss": 0.7039, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.0002326335268257019, + "step": 5236 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2908, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.325507665181622e-06, + "loss": 0.7298, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0002325507665181622, + "step": 5237 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2907, + "image_masks_loss": 0.0116, + "llm_learning_rate": 2.3246800813217095e-06, + "loss": 0.7632, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00023246800813217093, + "step": 5238 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2906, + "image_masks_loss": 0.0221, + "llm_learning_rate": 2.323852516768416e-06, + "loss": 0.7951, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.0002323852516768416, + "step": 5239 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2905, + "image_masks_loss": 0.0212, + "llm_learning_rate": 2.323024971612877e-06, + "loss": 0.7158, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023230249716128768, + "step": 5240 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2905, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.3221974459462217e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00023221974459462216, + "step": 5241 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2904, + "image_masks_loss": 0.0057, + "llm_learning_rate": 2.32136993985958e-06, + "loss": 0.7118, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00023213699398595797, + "step": 5242 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2903, + "image_masks_loss": 0.0132, + "llm_learning_rate": 2.3205424534440776e-06, + "loss": 0.6464, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00023205424534440775, + "step": 5243 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2903, + "image_masks_loss": 0.0096, + "llm_learning_rate": 2.3197149867908407e-06, + "loss": 0.7513, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00023197149867908406, + "step": 5244 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2902, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.318887539990991e-06, + "loss": 0.6429, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00023188875399909908, + "step": 5245 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2901, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.3180601131356484e-06, + "loss": 0.702, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023180601131356481, + "step": 5246 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2901, + "image_masks_loss": 0.0143, + "llm_learning_rate": 2.3172327063159317e-06, + "loss": 0.738, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00023172327063159315, + "step": 5247 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.29, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.3164053196229566e-06, + "loss": 0.7781, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00023164053196229564, + "step": 5248 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2899, + "image_masks_loss": 0.018, + "llm_learning_rate": 2.315577953147837e-06, + "loss": 0.7377, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002315577953147837, + "step": 5249 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2899, + "image_masks_loss": 0.0065, + "llm_learning_rate": 2.3147506069816837e-06, + "loss": 0.7641, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023147506069816834, + "step": 5250 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2898, + "image_masks_loss": 0.0191, + "llm_learning_rate": 2.313923281215606e-06, + "loss": 0.7838, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.00023139232812156057, + "step": 5251 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2897, + "image_masks_loss": 0.0046, + "llm_learning_rate": 2.313095975940711e-06, + "loss": 0.7558, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00023130959759407106, + "step": 5252 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2897, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.312268691248103e-06, + "loss": 0.6639, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002312268691248103, + "step": 5253 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2896, + "image_masks_loss": 0.0187, + "llm_learning_rate": 2.3114414272288857e-06, + "loss": 0.7846, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00023114414272288857, + "step": 5254 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2895, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.3106141839741576e-06, + "loss": 0.7132, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00023106141839741574, + "step": 5255 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2895, + "image_masks_loss": 0.0193, + "llm_learning_rate": 2.309786961575018e-06, + "loss": 0.7043, + "output_text_mask_loss": 0.0037, + "predictor_learning_rate": 0.00023097869615750177, + "step": 5256 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2894, + "image_masks_loss": 0.0228, + "llm_learning_rate": 2.308959760122562e-06, + "loss": 0.7754, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00023089597601225615, + "step": 5257 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2893, + "image_masks_loss": 0.019, + "llm_learning_rate": 2.3081325797078825e-06, + "loss": 0.694, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023081325797078822, + "step": 5258 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2892, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.3073054204220703e-06, + "loss": 0.7197, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.000230730542042207, + "step": 5259 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2892, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.3064782823562147e-06, + "loss": 0.6201, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023064782823562145, + "step": 5260 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2891, + "image_masks_loss": 0.0275, + "llm_learning_rate": 2.305651165601402e-06, + "loss": 0.7144, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.0002305651165601402, + "step": 5261 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.289, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.3048240702487158e-06, + "loss": 0.7314, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00023048240702487155, + "step": 5262 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.289, + "image_masks_loss": 0.0059, + "llm_learning_rate": 2.303996996389237e-06, + "loss": 0.7711, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00023039969963892367, + "step": 5263 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2889, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.303169944114046e-06, + "loss": 0.7431, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002303169944114046, + "step": 5264 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2888, + "image_masks_loss": 0.0225, + "llm_learning_rate": 2.3023429135142194e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00023023429135142194, + "step": 5265 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2888, + "image_masks_loss": 0.0236, + "llm_learning_rate": 2.3015159046808307e-06, + "loss": 0.7272, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.00023015159046808306, + "step": 5266 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2887, + "image_masks_loss": 0.0222, + "llm_learning_rate": 2.3006889177049535e-06, + "loss": 0.7285, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00023006889177049533, + "step": 5267 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2886, + "image_masks_loss": 0.0176, + "llm_learning_rate": 2.299861952677657e-06, + "loss": 0.8703, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00022998619526776572, + "step": 5268 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2886, + "image_masks_loss": 0.0029, + "llm_learning_rate": 2.2990350096900076e-06, + "loss": 0.7668, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.00022990350096900076, + "step": 5269 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2885, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.2982080888330704e-06, + "loss": 0.7666, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.00022982080888330703, + "step": 5270 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2884, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.2973811901979084e-06, + "loss": 0.7256, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002297381190197908, + "step": 5271 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2884, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.29655431387558e-06, + "loss": 0.7628, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.000229655431387558, + "step": 5272 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2883, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.2957274599571443e-06, + "loss": 0.815, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.00022957274599571444, + "step": 5273 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2882, + "image_masks_loss": 0.0098, + "llm_learning_rate": 2.2949006285336557e-06, + "loss": 0.7128, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.00022949006285336558, + "step": 5274 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2882, + "image_masks_loss": 0.008, + "llm_learning_rate": 2.2940738196961664e-06, + "loss": 0.7485, + "output_text_mask_loss": 0.0713, + "predictor_learning_rate": 0.00022940738196961663, + "step": 5275 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2881, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.2932470335357273e-06, + "loss": 0.7021, + "output_text_mask_loss": 0.0889, + "predictor_learning_rate": 0.00022932470335357272, + "step": 5276 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.288, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.2924202701433852e-06, + "loss": 0.6934, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00022924202701433852, + "step": 5277 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.288, + "image_masks_loss": 0.0317, + "llm_learning_rate": 2.2915935296101855e-06, + "loss": 0.6597, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00022915935296101856, + "step": 5278 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2879, + "image_masks_loss": 0.0147, + "llm_learning_rate": 2.29076681202717e-06, + "loss": 0.768, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.000229076681202717, + "step": 5279 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2878, + "image_masks_loss": 0.0204, + "llm_learning_rate": 2.2899401174853796e-06, + "loss": 0.7684, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00022899401174853796, + "step": 5280 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2878, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.2891134460758515e-06, + "loss": 0.7099, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022891134460758513, + "step": 5281 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2877, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.2882867978896197e-06, + "loss": 0.774, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00022882867978896195, + "step": 5282 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2876, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.287460173017718e-06, + "loss": 0.7045, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002287460173017718, + "step": 5283 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2875, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.2866335715511755e-06, + "loss": 0.7254, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 0.00022866335715511754, + "step": 5284 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2875, + "image_masks_loss": 0.0324, + "llm_learning_rate": 2.2858069935810193e-06, + "loss": 0.763, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0002285806993581019, + "step": 5285 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2874, + "image_masks_loss": 0.0111, + "llm_learning_rate": 2.2849804391982748e-06, + "loss": 0.6995, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022849804391982746, + "step": 5286 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2873, + "image_masks_loss": 0.0246, + "llm_learning_rate": 2.2841539084939623e-06, + "loss": 0.7191, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0002284153908493962, + "step": 5287 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2873, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.283327401559102e-06, + "loss": 0.7497, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00022833274015591015, + "step": 5288 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2872, + "image_masks_loss": 0.0136, + "llm_learning_rate": 2.282500918484711e-06, + "loss": 0.9232, + "output_text_mask_loss": 0.0607, + "predictor_learning_rate": 0.0002282500918484711, + "step": 5289 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2871, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.2816744593618037e-06, + "loss": 0.7146, + "output_text_mask_loss": 0.0188, + "predictor_learning_rate": 0.00022816744593618036, + "step": 5290 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2871, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.2808480242813907e-06, + "loss": 0.731, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00022808480242813902, + "step": 5291 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.287, + "image_masks_loss": 0.0181, + "llm_learning_rate": 2.2800216133344816e-06, + "loss": 0.7522, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00022800216133344814, + "step": 5292 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2869, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.2791952266120825e-06, + "loss": 0.7433, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022791952266120824, + "step": 5293 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2869, + "image_masks_loss": 0.0237, + "llm_learning_rate": 2.278368864205197e-06, + "loss": 0.8275, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0002278368864205197, + "step": 5294 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2868, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.2775425262048254e-06, + "loss": 0.6567, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.00022775425262048252, + "step": 5295 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2867, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.276716212701967e-06, + "loss": 0.7098, + "output_text_mask_loss": 0.0383, + "predictor_learning_rate": 0.00022767162127019667, + "step": 5296 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2867, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.275889923787616e-06, + "loss": 0.7806, + "output_text_mask_loss": 0.103, + "predictor_learning_rate": 0.0002275889923787616, + "step": 5297 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2866, + "image_masks_loss": 0.0178, + "llm_learning_rate": 2.2750636595527657e-06, + "loss": 0.7309, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022750636595527654, + "step": 5298 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2865, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.2742374200884065e-06, + "loss": 0.6684, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022742374200884064, + "step": 5299 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2865, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.2734112054855256e-06, + "loss": 0.8501, + "output_text_mask_loss": 0.0778, + "predictor_learning_rate": 0.00022734112054855255, + "step": 5300 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2864, + "image_masks_loss": 0.0096, + "llm_learning_rate": 2.2725850158351076e-06, + "loss": 0.78, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00022725850158351074, + "step": 5301 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2863, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.2717588512281334e-06, + "loss": 0.7518, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00022717588512281335, + "step": 5302 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2863, + "image_masks_loss": 0.0075, + "llm_learning_rate": 2.2709327117555838e-06, + "loss": 0.7654, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00022709327117555837, + "step": 5303 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2862, + "image_masks_loss": 0.008, + "llm_learning_rate": 2.2701065975084343e-06, + "loss": 0.6867, + "output_text_mask_loss": 0.0364, + "predictor_learning_rate": 0.0002270106597508434, + "step": 5304 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2861, + "image_masks_loss": 0.0054, + "llm_learning_rate": 2.2692805085776583e-06, + "loss": 0.7053, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.0002269280508577658, + "step": 5305 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2861, + "image_masks_loss": 0.0193, + "llm_learning_rate": 2.268454445054226e-06, + "loss": 0.7946, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 0.0002268454445054226, + "step": 5306 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.286, + "image_masks_loss": 0.0143, + "llm_learning_rate": 2.267628407029106e-06, + "loss": 0.763, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0002267628407029106, + "step": 5307 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2859, + "image_masks_loss": 0.0148, + "llm_learning_rate": 2.2668023945932636e-06, + "loss": 0.7418, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022668023945932636, + "step": 5308 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2859, + "image_masks_loss": 0.0066, + "llm_learning_rate": 2.265976407837661e-06, + "loss": 0.6771, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002265976407837661, + "step": 5309 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2858, + "image_masks_loss": 0.0235, + "llm_learning_rate": 2.2651504468532577e-06, + "loss": 0.8043, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00022651504468532577, + "step": 5310 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2857, + "image_masks_loss": 0.0086, + "llm_learning_rate": 2.26432451173101e-06, + "loss": 0.7822, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00022643245117310098, + "step": 5311 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2857, + "image_masks_loss": 0.0152, + "llm_learning_rate": 2.2634986025618723e-06, + "loss": 0.6914, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.0002263498602561872, + "step": 5312 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2856, + "image_masks_loss": 0.0046, + "llm_learning_rate": 2.262672719436795e-06, + "loss": 0.7369, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00022626727194367948, + "step": 5313 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2855, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.2618468624467265e-06, + "loss": 0.692, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022618468624467262, + "step": 5314 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2855, + "image_masks_loss": 0.0257, + "llm_learning_rate": 2.261021031682612e-06, + "loss": 0.8171, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00022610210316826118, + "step": 5315 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2854, + "image_masks_loss": 0.0101, + "llm_learning_rate": 2.260195227235394e-06, + "loss": 0.6832, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.0002260195227235394, + "step": 5316 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2853, + "image_masks_loss": 0.016, + "llm_learning_rate": 2.259369449196012e-06, + "loss": 0.754, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00022593694491960116, + "step": 5317 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2852, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.258543697655401e-06, + "loss": 0.8098, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.0002258543697655401, + "step": 5318 + }, + { + "epoch": 0.54, + "gumbel_tau": 0.2852, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.257717972704497e-06, + "loss": 0.8473, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00022577179727044968, + "step": 5319 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2851, + "image_masks_loss": 0.0049, + "llm_learning_rate": 2.2568922744342294e-06, + "loss": 0.7898, + "output_text_mask_loss": 0.103, + "predictor_learning_rate": 0.0002256892274434229, + "step": 5320 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.285, + "image_masks_loss": 0.0052, + "llm_learning_rate": 2.256066602935525e-06, + "loss": 0.7313, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002256066602935525, + "step": 5321 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.285, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.255240958299311e-06, + "loss": 0.7379, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00022552409582993107, + "step": 5322 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2849, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.254415340616508e-06, + "loss": 0.8289, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.00022544153406165078, + "step": 5323 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2848, + "image_masks_loss": 0.002, + "llm_learning_rate": 2.253589749978034e-06, + "loss": 0.7923, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.00022535897499780338, + "step": 5324 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2848, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.2527641864748057e-06, + "loss": 0.7795, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.00022527641864748056, + "step": 5325 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2847, + "image_masks_loss": 0.0207, + "llm_learning_rate": 2.251938650197736e-06, + "loss": 0.6981, + "output_text_mask_loss": 0.0257, + "predictor_learning_rate": 0.00022519386501977358, + "step": 5326 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2846, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.251113141237734e-06, + "loss": 0.6749, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00022511131412377336, + "step": 5327 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2846, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.2502876596857076e-06, + "loss": 0.8262, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00022502876596857074, + "step": 5328 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2845, + "image_masks_loss": 0.0159, + "llm_learning_rate": 2.2494622056325606e-06, + "loss": 0.748, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 0.00022494622056325607, + "step": 5329 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2844, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.248636779169193e-06, + "loss": 0.7385, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002248636779169193, + "step": 5330 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2844, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.2478113803865038e-06, + "loss": 0.7502, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.00022478113803865035, + "step": 5331 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2843, + "image_masks_loss": 0.0043, + "llm_learning_rate": 2.246986009375387e-06, + "loss": 0.7073, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0002246986009375387, + "step": 5332 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2842, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.2461606662267342e-06, + "loss": 0.6555, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.0002246160666226734, + "step": 5333 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2842, + "image_masks_loss": 0.0089, + "llm_learning_rate": 2.245335351031434e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.0002245335351031434, + "step": 5334 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2841, + "image_masks_loss": 0.0029, + "llm_learning_rate": 2.2445100638803726e-06, + "loss": 0.6905, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00022445100638803726, + "step": 5335 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.284, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.243684804864432e-06, + "loss": 0.747, + "output_text_mask_loss": 0.0067, + "predictor_learning_rate": 0.0002243684804864432, + "step": 5336 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.284, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.2428595740744917e-06, + "loss": 0.7004, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00022428595740744915, + "step": 5337 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2839, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.2420343716014274e-06, + "loss": 0.67, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022420343716014274, + "step": 5338 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2838, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.2412091975361135e-06, + "loss": 0.7153, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00022412091975361131, + "step": 5339 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2838, + "image_masks_loss": 0.0206, + "llm_learning_rate": 2.2403840519694187e-06, + "loss": 0.7784, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00022403840519694186, + "step": 5340 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2837, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.23955893499221e-06, + "loss": 0.8438, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000223955893499221, + "step": 5341 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2836, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.238733846695353e-06, + "loss": 0.7635, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.0002238733846695353, + "step": 5342 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2836, + "image_masks_loss": 0.0044, + "llm_learning_rate": 2.237908787169705e-06, + "loss": 0.7586, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00022379087871697052, + "step": 5343 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2835, + "image_masks_loss": 0.0277, + "llm_learning_rate": 2.2370837565061264e-06, + "loss": 0.817, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.00022370837565061262, + "step": 5344 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2834, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.2362587547954696e-06, + "loss": 0.7355, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.00022362587547954693, + "step": 5345 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2834, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.2354337821285857e-06, + "loss": 0.6989, + "output_text_mask_loss": 0.0084, + "predictor_learning_rate": 0.00022354337821285856, + "step": 5346 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2833, + "image_masks_loss": 0.0146, + "llm_learning_rate": 2.2346088385963237e-06, + "loss": 0.7339, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00022346088385963235, + "step": 5347 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2832, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.2337839242895275e-06, + "loss": 0.7288, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002233783924289527, + "step": 5348 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2832, + "image_masks_loss": 0.0053, + "llm_learning_rate": 2.2329590392990384e-06, + "loss": 0.7328, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0002232959039299038, + "step": 5349 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2831, + "image_masks_loss": 0.0045, + "llm_learning_rate": 2.232134183715694e-06, + "loss": 0.7402, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002232134183715694, + "step": 5350 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.283, + "image_masks_loss": 0.0214, + "llm_learning_rate": 2.2313093576303307e-06, + "loss": 0.6917, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00022313093576303307, + "step": 5351 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.283, + "image_masks_loss": 0.004, + "llm_learning_rate": 2.2304845611337794e-06, + "loss": 0.7081, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00022304845611337794, + "step": 5352 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2829, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.229659794316868e-06, + "loss": 0.7684, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00022296597943168679, + "step": 5353 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2828, + "image_masks_loss": 0.0099, + "llm_learning_rate": 2.2288350572704226e-06, + "loss": 0.7685, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00022288350572704225, + "step": 5354 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2828, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.228010350085265e-06, + "loss": 0.8046, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002228010350085265, + "step": 5355 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2827, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.2271856728522136e-06, + "loss": 0.6805, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.00022271856728522133, + "step": 5356 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2826, + "image_masks_loss": 0.0045, + "llm_learning_rate": 2.2263610256620824e-06, + "loss": 0.7507, + "output_text_mask_loss": 0.0135, + "predictor_learning_rate": 0.00022263610256620823, + "step": 5357 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2826, + "image_masks_loss": 0.0199, + "llm_learning_rate": 2.2255364086056854e-06, + "loss": 0.6922, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00022255364086056853, + "step": 5358 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2825, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.2247118217738307e-06, + "loss": 0.701, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022247118217738303, + "step": 5359 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2824, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.2238872652573234e-06, + "loss": 0.7499, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.00022238872652573233, + "step": 5360 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2824, + "image_masks_loss": 0.0031, + "llm_learning_rate": 2.2230627391469653e-06, + "loss": 0.8393, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0002223062739146965, + "step": 5361 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2823, + "image_masks_loss": 0.0267, + "llm_learning_rate": 2.2222382435335545e-06, + "loss": 0.7607, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022222382435335545, + "step": 5362 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2822, + "image_masks_loss": 0.0277, + "llm_learning_rate": 2.221413778507888e-06, + "loss": 0.6357, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00022214137785078877, + "step": 5363 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2822, + "image_masks_loss": 0.009, + "llm_learning_rate": 2.220589344160756e-06, + "loss": 0.7664, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002220589344160756, + "step": 5364 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2821, + "image_masks_loss": 0.0022, + "llm_learning_rate": 2.2197649405829486e-06, + "loss": 0.7488, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022197649405829485, + "step": 5365 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.282, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.218940567865249e-06, + "loss": 0.7108, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022189405678652488, + "step": 5366 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.282, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.2181162260984407e-06, + "loss": 0.7588, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022181162260984406, + "step": 5367 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2819, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.2172919153733017e-06, + "loss": 0.7031, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00022172919153733015, + "step": 5368 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2818, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.216467635780606e-06, + "loss": 0.7014, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00022164676357806057, + "step": 5369 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2818, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.2156433874111265e-06, + "loss": 0.826, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00022156433874111263, + "step": 5370 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2817, + "image_masks_loss": 0.0105, + "llm_learning_rate": 2.21481917035563e-06, + "loss": 0.7554, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.000221481917035563, + "step": 5371 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2816, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.213994984704882e-06, + "loss": 0.7789, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00022139949847048822, + "step": 5372 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2816, + "image_masks_loss": 0.0137, + "llm_learning_rate": 2.213170830549643e-06, + "loss": 0.7448, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.0002213170830549643, + "step": 5373 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2815, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.2123467079806716e-06, + "loss": 0.7576, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00022123467079806713, + "step": 5374 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2814, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.2115226170887213e-06, + "loss": 0.7131, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00022115226170887212, + "step": 5375 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2814, + "image_masks_loss": 0.0026, + "llm_learning_rate": 2.2106985579645423e-06, + "loss": 0.6949, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00022106985579645422, + "step": 5376 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2813, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.2098745306988834e-06, + "loss": 0.7037, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00022098745306988832, + "step": 5377 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2812, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.209050535382487e-06, + "loss": 0.7269, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 0.00022090505353824868, + "step": 5378 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2812, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.2082265721060944e-06, + "loss": 0.8059, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00022082265721060944, + "step": 5379 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2811, + "image_masks_loss": 0.0033, + "llm_learning_rate": 2.207402640960441e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00022074026409604408, + "step": 5380 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.281, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.2065787420362606e-06, + "loss": 0.7211, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00022065787420362605, + "step": 5381 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.281, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.205754875424282e-06, + "loss": 0.6489, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0002205754875424282, + "step": 5382 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2809, + "image_masks_loss": 0.0119, + "llm_learning_rate": 2.204931041215233e-06, + "loss": 0.6727, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00022049310412152327, + "step": 5383 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2808, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.204107239499834e-06, + "loss": 0.7291, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00022041072394998342, + "step": 5384 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2808, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.203283470368805e-06, + "loss": 0.7134, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.00022032834703688048, + "step": 5385 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2807, + "image_masks_loss": 0.0101, + "llm_learning_rate": 2.202459733912862e-06, + "loss": 0.6835, + "output_text_mask_loss": 0.0462, + "predictor_learning_rate": 0.00022024597339128616, + "step": 5386 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2806, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.2016360302227153e-06, + "loss": 0.8069, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00022016360302227152, + "step": 5387 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2806, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.2008123593890734e-06, + "loss": 0.7313, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00022008123593890734, + "step": 5388 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2805, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.1999887215026405e-06, + "loss": 0.7756, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021999887215026404, + "step": 5389 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2804, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.199165116654119e-06, + "loss": 0.7355, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021991651166541185, + "step": 5390 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2804, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.198341544934204e-06, + "loss": 0.8317, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00021983415449342039, + "step": 5391 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2803, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.19751800643359e-06, + "loss": 0.7839, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021975180064335897, + "step": 5392 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2802, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.196694501242967e-06, + "loss": 0.8349, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.0002196694501242967, + "step": 5393 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2802, + "image_masks_loss": 0.0162, + "llm_learning_rate": 2.195871029453022e-06, + "loss": 0.5792, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021958710294530214, + "step": 5394 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2801, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.1950475911544354e-06, + "loss": 0.7323, + "output_text_mask_loss": 0.0732, + "predictor_learning_rate": 0.00021950475911544355, + "step": 5395 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.28, + "image_masks_loss": 0.0133, + "llm_learning_rate": 2.1942241864378876e-06, + "loss": 0.7373, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00021942241864378876, + "step": 5396 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.28, + "image_masks_loss": 0.0041, + "llm_learning_rate": 2.193400815394054e-06, + "loss": 0.7196, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.00021934008153940538, + "step": 5397 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2799, + "image_masks_loss": 0.0048, + "llm_learning_rate": 2.192577478113606e-06, + "loss": 0.8039, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00021925774781136057, + "step": 5398 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2799, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.19175417468721e-06, + "loss": 0.763, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000219175417468721, + "step": 5399 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2798, + "image_masks_loss": 0.0338, + "llm_learning_rate": 2.190930905205531e-06, + "loss": 0.8064, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.0002190930905205531, + "step": 5400 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2797, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.190107669759229e-06, + "loss": 0.7387, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00021901076697592288, + "step": 5401 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2797, + "image_masks_loss": 0.0203, + "llm_learning_rate": 2.1892844684389608e-06, + "loss": 0.815, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00021892844684389607, + "step": 5402 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2796, + "image_masks_loss": 0.0076, + "llm_learning_rate": 2.1884613013353793e-06, + "loss": 0.6369, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002188461301335379, + "step": 5403 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2795, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.187638168539133e-06, + "loss": 0.7783, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.00021876381685391328, + "step": 5404 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2795, + "image_masks_loss": 0.0091, + "llm_learning_rate": 2.186815070140867e-06, + "loss": 0.7526, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00021868150701408667, + "step": 5405 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2794, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.1859920062312234e-06, + "loss": 0.7494, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00021859920062312232, + "step": 5406 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2793, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.1851689769008394e-06, + "loss": 0.7733, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00021851689769008392, + "step": 5407 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2793, + "image_masks_loss": 0.0193, + "llm_learning_rate": 2.1843459822403487e-06, + "loss": 0.7144, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021843459822403486, + "step": 5408 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2792, + "image_masks_loss": 0.0065, + "llm_learning_rate": 2.183523022340382e-06, + "loss": 0.7921, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00021835230223403817, + "step": 5409 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2791, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.182700097291565e-06, + "loss": 0.7421, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00021827000972915648, + "step": 5410 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2791, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.18187720718452e-06, + "loss": 0.8243, + "output_text_mask_loss": 0.0067, + "predictor_learning_rate": 0.00021818772071845196, + "step": 5411 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.279, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.181054352109865e-06, + "loss": 0.7481, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00021810543521098646, + "step": 5412 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2789, + "image_masks_loss": 0.0111, + "llm_learning_rate": 2.1802315321582155e-06, + "loss": 0.7717, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00021802315321582156, + "step": 5413 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2789, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.179408747420182e-06, + "loss": 0.7399, + "output_text_mask_loss": 0.0146, + "predictor_learning_rate": 0.0002179408747420182, + "step": 5414 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2788, + "image_masks_loss": 0.036, + "llm_learning_rate": 2.1785859979863713e-06, + "loss": 0.7698, + "output_text_mask_loss": 0.004, + "predictor_learning_rate": 0.00021785859979863708, + "step": 5415 + }, + { + "epoch": 0.55, + "gumbel_tau": 0.2787, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.1777632839473873e-06, + "loss": 0.6164, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021777632839473874, + "step": 5416 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2787, + "image_masks_loss": 0.0515, + "llm_learning_rate": 2.176940605393827e-06, + "loss": 0.6601, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021769406053938272, + "step": 5417 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2786, + "image_masks_loss": 0.0319, + "llm_learning_rate": 2.1761179624162874e-06, + "loss": 0.8302, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.00021761179624162874, + "step": 5418 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2785, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.175295355105359e-06, + "loss": 0.7011, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.0002175295355105359, + "step": 5419 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2785, + "image_masks_loss": 0.0056, + "llm_learning_rate": 2.1744727835516295e-06, + "loss": 0.754, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 0.00021744727835516296, + "step": 5420 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2784, + "image_masks_loss": 0.0187, + "llm_learning_rate": 2.1736502478456816e-06, + "loss": 0.8109, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00021736502478456816, + "step": 5421 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2783, + "image_masks_loss": 0.0131, + "llm_learning_rate": 2.172827748078096e-06, + "loss": 0.6576, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00021728277480780958, + "step": 5422 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2783, + "image_masks_loss": 0.0132, + "llm_learning_rate": 2.1720052843394474e-06, + "loss": 0.7352, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.00021720052843394474, + "step": 5423 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2782, + "image_masks_loss": 0.0176, + "llm_learning_rate": 2.1711828567203068e-06, + "loss": 0.6749, + "output_text_mask_loss": 0.0026, + "predictor_learning_rate": 0.00021711828567203069, + "step": 5424 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2781, + "image_masks_loss": 0.015, + "llm_learning_rate": 2.170360465311243e-06, + "loss": 0.7376, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.0002170360465311243, + "step": 5425 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2781, + "image_masks_loss": 0.0193, + "llm_learning_rate": 2.169538110202819e-06, + "loss": 0.842, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.0002169538110202819, + "step": 5426 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.278, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.1687157914855946e-06, + "loss": 0.7216, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 0.00021687157914855945, + "step": 5427 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2779, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.1678935092501243e-06, + "loss": 0.7973, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0002167893509250124, + "step": 5428 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2779, + "image_masks_loss": 0.0105, + "llm_learning_rate": 2.1670712635869617e-06, + "loss": 0.6381, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00021670712635869614, + "step": 5429 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2778, + "image_masks_loss": 0.0143, + "llm_learning_rate": 2.166249054586652e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.0553, + "predictor_learning_rate": 0.0002166249054586652, + "step": 5430 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2777, + "image_masks_loss": 0.0047, + "llm_learning_rate": 2.16542688233974e-06, + "loss": 0.725, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000216542688233974, + "step": 5431 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2777, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.1646047469367656e-06, + "loss": 0.7506, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 0.00021646047469367656, + "step": 5432 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2776, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.163782648468263e-06, + "loss": 0.7414, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00021637826484682634, + "step": 5433 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2775, + "image_masks_loss": 0.0196, + "llm_learning_rate": 2.162960587024765e-06, + "loss": 0.7376, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021629605870247646, + "step": 5434 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2775, + "image_masks_loss": 0.009, + "llm_learning_rate": 2.1621385626967976e-06, + "loss": 0.8098, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00021621385626967974, + "step": 5435 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2774, + "image_masks_loss": 0.0158, + "llm_learning_rate": 2.161316575574884e-06, + "loss": 0.757, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00021613165755748838, + "step": 5436 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2774, + "image_masks_loss": 0.0216, + "llm_learning_rate": 2.160494625749542e-06, + "loss": 0.8009, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00021604946257495423, + "step": 5437 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2773, + "image_masks_loss": 0.0076, + "llm_learning_rate": 2.15967271331129e-06, + "loss": 0.7557, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021596727133112897, + "step": 5438 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2772, + "image_masks_loss": 0.0087, + "llm_learning_rate": 2.1588508383506364e-06, + "loss": 0.7761, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00021588508383506362, + "step": 5439 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2772, + "image_masks_loss": 0.0254, + "llm_learning_rate": 2.1580290009580875e-06, + "loss": 0.7862, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.00021580290009580872, + "step": 5440 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2771, + "image_masks_loss": 0.005, + "llm_learning_rate": 2.1572072012241474e-06, + "loss": 0.7478, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00021572072012241472, + "step": 5441 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.277, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.156385439239314e-06, + "loss": 0.7686, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00021563854392393138, + "step": 5442 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.277, + "image_masks_loss": 0.022, + "llm_learning_rate": 2.155563715094081e-06, + "loss": 0.7992, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.0002155563715094081, + "step": 5443 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2769, + "image_masks_loss": 0.0149, + "llm_learning_rate": 2.154742028878939e-06, + "loss": 0.8007, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.00021547420288789388, + "step": 5444 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2768, + "image_masks_loss": 0.0166, + "llm_learning_rate": 2.153920380684374e-06, + "loss": 0.7484, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.0002153920380684374, + "step": 5445 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2768, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.1530987706008676e-06, + "loss": 0.6438, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00021530987706008673, + "step": 5446 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2767, + "image_masks_loss": 0.0378, + "llm_learning_rate": 2.1522771987188966e-06, + "loss": 0.604, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00021522771987188963, + "step": 5447 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2766, + "image_masks_loss": 0.0214, + "llm_learning_rate": 2.1514556651289354e-06, + "loss": 0.7725, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0002151455665128935, + "step": 5448 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2766, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.150634169921453e-06, + "loss": 0.6574, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.00021506341699214528, + "step": 5449 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2765, + "image_masks_loss": 0.0159, + "llm_learning_rate": 2.149812713186914e-06, + "loss": 0.7322, + "output_text_mask_loss": 0.0477, + "predictor_learning_rate": 0.00021498127131869138, + "step": 5450 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2764, + "image_masks_loss": 0.0091, + "llm_learning_rate": 2.148991295015778e-06, + "loss": 0.789, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 0.0002148991295015778, + "step": 5451 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2764, + "image_masks_loss": 0.0216, + "llm_learning_rate": 2.1481699154985037e-06, + "loss": 0.7561, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.00021481699154985036, + "step": 5452 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2763, + "image_masks_loss": 0.0157, + "llm_learning_rate": 2.147348574725542e-06, + "loss": 0.6721, + "output_text_mask_loss": 0.0927, + "predictor_learning_rate": 0.0002147348574725542, + "step": 5453 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2762, + "image_masks_loss": 0.0065, + "llm_learning_rate": 2.1465272727873404e-06, + "loss": 0.7346, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00021465272727873403, + "step": 5454 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2762, + "image_masks_loss": 0.005, + "llm_learning_rate": 2.145706009774343e-06, + "loss": 0.7094, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00021457060097743427, + "step": 5455 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2761, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.144884785776988e-06, + "loss": 0.7235, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00021448847857769883, + "step": 5456 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.276, + "image_masks_loss": 0.0116, + "llm_learning_rate": 2.1440636008857123e-06, + "loss": 0.8198, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00021440636008857123, + "step": 5457 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.276, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.1432424551909454e-06, + "loss": 0.7757, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00021432424551909453, + "step": 5458 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2759, + "image_masks_loss": 0.0188, + "llm_learning_rate": 2.1424213487831145e-06, + "loss": 0.7853, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.00021424213487831144, + "step": 5459 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2759, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.14160028175264e-06, + "loss": 0.7616, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.000214160028175264, + "step": 5460 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2758, + "image_masks_loss": 0.0259, + "llm_learning_rate": 2.140779254189942e-06, + "loss": 0.7518, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 0.0002140779254189942, + "step": 5461 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2757, + "image_masks_loss": 0.0136, + "llm_learning_rate": 2.139958266185432e-06, + "loss": 0.7981, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002139958266185432, + "step": 5462 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2757, + "image_masks_loss": 0.023, + "llm_learning_rate": 2.1391373178295194e-06, + "loss": 0.7246, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00021391373178295193, + "step": 5463 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2756, + "image_masks_loss": 0.0225, + "llm_learning_rate": 2.1383164092126096e-06, + "loss": 0.7967, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00021383164092126096, + "step": 5464 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2755, + "image_masks_loss": 0.0046, + "llm_learning_rate": 2.1374955404251027e-06, + "loss": 0.857, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 0.00021374955404251025, + "step": 5465 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2755, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.136674711557394e-06, + "loss": 0.7025, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002136674711557394, + "step": 5466 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2754, + "image_masks_loss": 0.0183, + "llm_learning_rate": 2.135853922699875e-06, + "loss": 0.7417, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.00021358539226998748, + "step": 5467 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2753, + "image_masks_loss": 0.0216, + "llm_learning_rate": 2.1350331739429334e-06, + "loss": 0.7016, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00021350331739429336, + "step": 5468 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2753, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.134212465376952e-06, + "loss": 0.741, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002134212465376952, + "step": 5469 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2752, + "image_masks_loss": 0.0263, + "llm_learning_rate": 2.1333917970923084e-06, + "loss": 0.7165, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002133391797092308, + "step": 5470 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2751, + "image_masks_loss": 0.0087, + "llm_learning_rate": 2.1325711691793766e-06, + "loss": 0.7212, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.00021325711691793763, + "step": 5471 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2751, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.1317505817285265e-06, + "loss": 0.8562, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00021317505817285265, + "step": 5472 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.275, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.1309300348301225e-06, + "loss": 0.6663, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00021309300348301225, + "step": 5473 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2749, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.130109528574525e-06, + "loss": 0.762, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0002130109528574525, + "step": 5474 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2749, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.1292890630520903e-06, + "loss": 0.645, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000212928906305209, + "step": 5475 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2748, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.1284686383531687e-06, + "loss": 0.6905, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.00021284686383531688, + "step": 5476 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2747, + "image_masks_loss": 0.0179, + "llm_learning_rate": 2.1276482545681093e-06, + "loss": 0.741, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.00021276482545681092, + "step": 5477 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2747, + "image_masks_loss": 0.0282, + "llm_learning_rate": 2.126827911787253e-06, + "loss": 0.7599, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002126827911787253, + "step": 5478 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2746, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.126007610100938e-06, + "loss": 0.779, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0002126007610100938, + "step": 5479 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2746, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.1251873495994983e-06, + "loss": 0.7622, + "output_text_mask_loss": 0.0595, + "predictor_learning_rate": 0.00021251873495994983, + "step": 5480 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2745, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.1243671303732628e-06, + "loss": 0.6995, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002124367130373263, + "step": 5481 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2744, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.123546952512556e-06, + "loss": 0.749, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00021235469525125557, + "step": 5482 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2744, + "image_masks_loss": 0.0244, + "llm_learning_rate": 2.122726816107696e-06, + "loss": 0.6501, + "output_text_mask_loss": 0.0402, + "predictor_learning_rate": 0.00021227268161076957, + "step": 5483 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2743, + "image_masks_loss": 0.0058, + "llm_learning_rate": 2.121906721249e-06, + "loss": 0.637, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0002121906721249, + "step": 5484 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2742, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.1210866680267783e-06, + "loss": 0.7473, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00021210866680267782, + "step": 5485 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2742, + "image_masks_loss": 0.0162, + "llm_learning_rate": 2.120266656531336e-06, + "loss": 0.6665, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00021202666565313359, + "step": 5486 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2741, + "image_masks_loss": 0.028, + "llm_learning_rate": 2.119446686852976e-06, + "loss": 0.7596, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00021194466868529758, + "step": 5487 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.274, + "image_masks_loss": 0.0152, + "llm_learning_rate": 2.118626759081994e-06, + "loss": 0.7961, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00021186267590819942, + "step": 5488 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.274, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.1178068733086835e-06, + "loss": 0.7454, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.00021178068733086834, + "step": 5489 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2739, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.116987029623331e-06, + "loss": 0.7145, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0002116987029623331, + "step": 5490 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2738, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.1161672281162204e-06, + "loss": 0.7395, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00021161672281162203, + "step": 5491 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2738, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.1153474688776288e-06, + "loss": 0.6253, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00021153474688776286, + "step": 5492 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2737, + "image_masks_loss": 0.0022, + "llm_learning_rate": 2.114527751997831e-06, + "loss": 0.7198, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002114527751997831, + "step": 5493 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2736, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.113708077567096e-06, + "loss": 0.7454, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0002113708077567096, + "step": 5494 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2736, + "image_masks_loss": 0.0049, + "llm_learning_rate": 2.1128884456756875e-06, + "loss": 0.7221, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00021128884456756875, + "step": 5495 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2735, + "image_masks_loss": 0.0196, + "llm_learning_rate": 2.1120688564138656e-06, + "loss": 0.8145, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00021120688564138653, + "step": 5496 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2735, + "image_masks_loss": 0.0133, + "llm_learning_rate": 2.1112493098718856e-06, + "loss": 0.718, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00021112493098718855, + "step": 5497 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2734, + "image_masks_loss": 0.0171, + "llm_learning_rate": 2.1104298061399977e-06, + "loss": 0.7876, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00021104298061399974, + "step": 5498 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2733, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.1096103453084464e-06, + "loss": 0.8133, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00021096103453084462, + "step": 5499 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2733, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.108790927467474e-06, + "loss": 0.8562, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00021087909274674737, + "step": 5500 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2732, + "image_masks_loss": 0.0194, + "llm_learning_rate": 2.1079715527073163e-06, + "loss": 0.6845, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002107971552707316, + "step": 5501 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2731, + "image_masks_loss": 0.0063, + "llm_learning_rate": 2.1071522211182044e-06, + "loss": 0.7716, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00021071522211182044, + "step": 5502 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2731, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.1063329327903644e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00021063329327903642, + "step": 5503 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.273, + "image_masks_loss": 0.0175, + "llm_learning_rate": 2.1055136878140188e-06, + "loss": 0.6503, + "output_text_mask_loss": 0.0587, + "predictor_learning_rate": 0.00021055136878140188, + "step": 5504 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2729, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.104694486279385e-06, + "loss": 0.6632, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.00021046944862793848, + "step": 5505 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2729, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.1038753282766746e-06, + "loss": 0.7784, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00021038753282766743, + "step": 5506 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2728, + "image_masks_loss": 0.0043, + "llm_learning_rate": 2.1030562138960953e-06, + "loss": 0.6774, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 0.0002103056213896095, + "step": 5507 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2727, + "image_masks_loss": 0.0033, + "llm_learning_rate": 2.1022371432278503e-06, + "loss": 0.6806, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000210223714322785, + "step": 5508 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2727, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.1014181163621375e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00021014181163621373, + "step": 5509 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2726, + "image_masks_loss": 0.0136, + "llm_learning_rate": 2.100599133389149e-06, + "loss": 0.703, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00021005991333891488, + "step": 5510 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2726, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.0997801943990734e-06, + "loss": 0.6519, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020997801943990733, + "step": 5511 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2725, + "image_masks_loss": 0.028, + "llm_learning_rate": 2.098961299482094e-06, + "loss": 0.6916, + "output_text_mask_loss": 0.0166, + "predictor_learning_rate": 0.0002098961299482094, + "step": 5512 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2724, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.09814244872839e-06, + "loss": 0.7694, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00020981424487283897, + "step": 5513 + }, + { + "epoch": 0.56, + "gumbel_tau": 0.2724, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.0973236422281346e-06, + "loss": 0.6969, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020973236422281344, + "step": 5514 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2723, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.096504880071496e-06, + "loss": 0.7457, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 0.00020965048800714958, + "step": 5515 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2722, + "image_masks_loss": 0.02, + "llm_learning_rate": 2.0956861623486393e-06, + "loss": 0.7357, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00020956861623486393, + "step": 5516 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2722, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.0948674891497234e-06, + "loss": 0.6781, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.00020948674891497232, + "step": 5517 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2721, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.0940488605649015e-06, + "loss": 0.7357, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00020940488605649014, + "step": 5518 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.272, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.093230276684323e-06, + "loss": 0.7865, + "output_text_mask_loss": 0.0265, + "predictor_learning_rate": 0.00020932302766843227, + "step": 5519 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.272, + "image_masks_loss": 0.0142, + "llm_learning_rate": 2.092411737598133e-06, + "loss": 0.6335, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00020924117375981327, + "step": 5520 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2719, + "image_masks_loss": 0.0054, + "llm_learning_rate": 2.0915932433964705e-06, + "loss": 0.776, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000209159324339647, + "step": 5521 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2718, + "image_masks_loss": 0.0181, + "llm_learning_rate": 2.0907747941694686e-06, + "loss": 0.7144, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00020907747941694685, + "step": 5522 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2718, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.089956390007259e-06, + "loss": 0.7689, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00020899563900072588, + "step": 5523 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2717, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.089138030999965e-06, + "loss": 0.8364, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.00020891380309999647, + "step": 5524 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2717, + "image_masks_loss": 0.018, + "llm_learning_rate": 2.0883197172377056e-06, + "loss": 0.733, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00020883197172377055, + "step": 5525 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2716, + "image_masks_loss": 0.009, + "llm_learning_rate": 2.0875014488105962e-06, + "loss": 0.7116, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0002087501448810596, + "step": 5526 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2715, + "image_masks_loss": 0.0138, + "llm_learning_rate": 2.0866832258087474e-06, + "loss": 0.6787, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002086683225808747, + "step": 5527 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2715, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.0858650483222605e-06, + "loss": 0.7924, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00020858650483222602, + "step": 5528 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2714, + "image_masks_loss": 0.0211, + "llm_learning_rate": 2.0850469164412376e-06, + "loss": 0.8467, + "output_text_mask_loss": 0.066, + "predictor_learning_rate": 0.00020850469164412376, + "step": 5529 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2713, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.0842288302557727e-06, + "loss": 0.8269, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.00020842288302557724, + "step": 5530 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2713, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.0834107898559542e-06, + "loss": 0.7592, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.00020834107898559542, + "step": 5531 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2712, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.0825927953318683e-06, + "loss": 0.7733, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00020825927953318683, + "step": 5532 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2711, + "image_masks_loss": 0.0056, + "llm_learning_rate": 2.0817748467735936e-06, + "loss": 0.7196, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00020817748467735935, + "step": 5533 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2711, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.080956944271204e-06, + "loss": 0.713, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002080956944271204, + "step": 5534 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.271, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.0801390879147687e-06, + "loss": 0.7209, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00020801390879147686, + "step": 5535 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.271, + "image_masks_loss": 0.0105, + "llm_learning_rate": 2.0793212777943527e-06, + "loss": 0.7938, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020793212777943525, + "step": 5536 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2709, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.0785035140000147e-06, + "loss": 0.8135, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00020785035140000147, + "step": 5537 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2708, + "image_masks_loss": 0.0149, + "llm_learning_rate": 2.0776857966218083e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00020776857966218082, + "step": 5538 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2708, + "image_masks_loss": 0.016, + "llm_learning_rate": 2.076868125749783e-06, + "loss": 0.8106, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00020768681257497828, + "step": 5539 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2707, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.076050501473982e-06, + "loss": 0.7848, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0002076050501473982, + "step": 5540 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2706, + "image_masks_loss": 0.0189, + "llm_learning_rate": 2.0752329238844444e-06, + "loss": 0.8134, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00020752329238844441, + "step": 5541 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2706, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.0744153930712023e-06, + "loss": 0.7511, + "output_text_mask_loss": 0.1289, + "predictor_learning_rate": 0.00020744153930712022, + "step": 5542 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2705, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.0735979091242863e-06, + "loss": 0.6962, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0002073597909124286, + "step": 5543 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2704, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.072780472133718e-06, + "loss": 0.7495, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020727804721337175, + "step": 5544 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2704, + "image_masks_loss": 0.0086, + "llm_learning_rate": 2.0719630821895147e-06, + "loss": 0.6863, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020719630821895146, + "step": 5545 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2703, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.0711457393816925e-06, + "loss": 0.7941, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00020711457393816923, + "step": 5546 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2702, + "image_masks_loss": 0.0084, + "llm_learning_rate": 2.0703284438002545e-06, + "loss": 0.6579, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00020703284438002546, + "step": 5547 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2702, + "image_masks_loss": 0.0194, + "llm_learning_rate": 2.0695111955352066e-06, + "loss": 0.7847, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.00020695111955352065, + "step": 5548 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2701, + "image_masks_loss": 0.0144, + "llm_learning_rate": 2.0686939946765445e-06, + "loss": 0.7527, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.00020686939946765442, + "step": 5549 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2701, + "image_masks_loss": 0.0075, + "llm_learning_rate": 2.06787684131426e-06, + "loss": 0.7653, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 0.00020678768413142599, + "step": 5550 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.27, + "image_masks_loss": 0.0219, + "llm_learning_rate": 2.0670597355383397e-06, + "loss": 0.7902, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00020670597355383397, + "step": 5551 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2699, + "image_masks_loss": 0.0116, + "llm_learning_rate": 2.066242677438767e-06, + "loss": 0.7887, + "output_text_mask_loss": 0.008, + "predictor_learning_rate": 0.00020662426774387666, + "step": 5552 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2699, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.065425667105516e-06, + "loss": 0.8002, + "output_text_mask_loss": 0.0782, + "predictor_learning_rate": 0.00020654256671055158, + "step": 5553 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2698, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.064608704628558e-06, + "loss": 0.7859, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0002064608704628558, + "step": 5554 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2697, + "image_masks_loss": 0.0147, + "llm_learning_rate": 2.0637917900978604e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0362, + "predictor_learning_rate": 0.000206379179009786, + "step": 5555 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2697, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.062974923603382e-06, + "loss": 0.7603, + "output_text_mask_loss": 0.0334, + "predictor_learning_rate": 0.0002062974923603382, + "step": 5556 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2696, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.062158105235078e-06, + "loss": 0.7621, + "output_text_mask_loss": 0.0317, + "predictor_learning_rate": 0.0002062158105235078, + "step": 5557 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2695, + "image_masks_loss": 0.0041, + "llm_learning_rate": 2.0613413350828993e-06, + "loss": 0.7105, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020613413350828992, + "step": 5558 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2695, + "image_masks_loss": 0.0057, + "llm_learning_rate": 2.0605246132367896e-06, + "loss": 0.8088, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.00020605246132367894, + "step": 5559 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2694, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.0597079397866887e-06, + "loss": 0.8042, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020597079397866887, + "step": 5560 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2694, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.0588913148225295e-06, + "loss": 0.7358, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.00020588913148225294, + "step": 5561 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2693, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.058074738434242e-06, + "loss": 0.7566, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00020580747384342418, + "step": 5562 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2692, + "image_masks_loss": 0.0084, + "llm_learning_rate": 2.0572582107117482e-06, + "loss": 0.7953, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00020572582107117484, + "step": 5563 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2692, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.0564417317449674e-06, + "loss": 0.7706, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00020564417317449674, + "step": 5564 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2691, + "image_masks_loss": 0.0048, + "llm_learning_rate": 2.05562530162381e-06, + "loss": 0.7589, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 0.00020556253016238102, + "step": 5565 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.269, + "image_masks_loss": 0.0035, + "llm_learning_rate": 2.054808920438185e-06, + "loss": 0.718, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00020548089204381847, + "step": 5566 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.269, + "image_masks_loss": 0.0201, + "llm_learning_rate": 2.053992588277992e-06, + "loss": 0.7549, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0002053992588277992, + "step": 5567 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2689, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.0531763052331297e-06, + "loss": 0.6466, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020531763052331295, + "step": 5568 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2688, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.0523600713934876e-06, + "loss": 0.7168, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020523600713934876, + "step": 5569 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2688, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.0515438868489514e-06, + "loss": 0.7467, + "output_text_mask_loss": 0.0992, + "predictor_learning_rate": 0.00020515438868489512, + "step": 5570 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2687, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.0507277516894015e-06, + "loss": 0.7779, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00020507277516894012, + "step": 5571 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2687, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.0499116660047123e-06, + "loss": 0.7658, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0002049911666004712, + "step": 5572 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2686, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.0490956298847527e-06, + "loss": 0.8216, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00020490956298847526, + "step": 5573 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2685, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.0482796434193867e-06, + "loss": 0.6127, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00020482796434193864, + "step": 5574 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2685, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.0474637066984724e-06, + "loss": 0.7082, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020474637066984725, + "step": 5575 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2684, + "image_masks_loss": 0.0158, + "llm_learning_rate": 2.0466478198118632e-06, + "loss": 0.7979, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0002046647819811863, + "step": 5576 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2683, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.0458319828494056e-06, + "loss": 0.685, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00020458319828494054, + "step": 5577 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2683, + "image_masks_loss": 0.0075, + "llm_learning_rate": 2.045016195900942e-06, + "loss": 0.7944, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020450161959009416, + "step": 5578 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2682, + "image_masks_loss": 0.0057, + "llm_learning_rate": 2.044200459056308e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0002044200459056308, + "step": 5579 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2682, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.043384772405335e-06, + "loss": 0.6546, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020433847724053348, + "step": 5580 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2681, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.042569136037847e-06, + "loss": 0.7211, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002042569136037847, + "step": 5581 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.268, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.0417535500436658e-06, + "loss": 0.7307, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00020417535500436655, + "step": 5582 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.268, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.0409380145126043e-06, + "loss": 0.8097, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00020409380145126042, + "step": 5583 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2679, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.040122529534471e-06, + "loss": 0.7493, + "output_text_mask_loss": 0.0183, + "predictor_learning_rate": 0.00020401225295344709, + "step": 5584 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2678, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.039307095199069e-06, + "loss": 0.7744, + "output_text_mask_loss": 0.0036, + "predictor_learning_rate": 0.0002039307095199069, + "step": 5585 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2678, + "image_masks_loss": 0.0159, + "llm_learning_rate": 2.0384917115961953e-06, + "loss": 0.7209, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00020384917115961952, + "step": 5586 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2677, + "image_masks_loss": 0.0042, + "llm_learning_rate": 2.037676378815643e-06, + "loss": 0.6725, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00020376763788156426, + "step": 5587 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2676, + "image_masks_loss": 0.0096, + "llm_learning_rate": 2.0368610969471974e-06, + "loss": 0.7567, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.0002036861096947197, + "step": 5588 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2676, + "image_masks_loss": 0.0116, + "llm_learning_rate": 2.0360458660806395e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0002036045866080639, + "step": 5589 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2675, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.0352306863057435e-06, + "loss": 0.7268, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0002035230686305743, + "step": 5590 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2675, + "image_masks_loss": 0.0044, + "llm_learning_rate": 2.03441555771228e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.000203441555771228, + "step": 5591 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2674, + "image_masks_loss": 0.0133, + "llm_learning_rate": 2.033600480390013e-06, + "loss": 0.7849, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00020336004803900128, + "step": 5592 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2673, + "image_masks_loss": 0.0042, + "llm_learning_rate": 2.032785454428699e-06, + "loss": 0.7588, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00020327854544286987, + "step": 5593 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2673, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.0319704799180918e-06, + "loss": 0.7858, + "output_text_mask_loss": 0.0114, + "predictor_learning_rate": 0.00020319704799180916, + "step": 5594 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2672, + "image_masks_loss": 0.0099, + "llm_learning_rate": 2.031155556947938e-06, + "loss": 0.729, + "output_text_mask_loss": 0.1038, + "predictor_learning_rate": 0.00020311555569479378, + "step": 5595 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2671, + "image_masks_loss": 0.0149, + "llm_learning_rate": 2.0303406856079786e-06, + "loss": 0.7526, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00020303406856079788, + "step": 5596 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2671, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.0295258659879487e-06, + "loss": 0.6658, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00020295258659879485, + "step": 5597 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.267, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.0287110981775787e-06, + "loss": 0.812, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020287110981775786, + "step": 5598 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.267, + "image_masks_loss": 0.0209, + "llm_learning_rate": 2.0278963822665925e-06, + "loss": 0.7441, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020278963822665922, + "step": 5599 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2669, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.0270817183447073e-06, + "loss": 0.7755, + "output_text_mask_loss": 0.0214, + "predictor_learning_rate": 0.0002027081718344707, + "step": 5600 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2668, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.0262671065016386e-06, + "loss": 0.7051, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.00020262671065016387, + "step": 5601 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2668, + "image_masks_loss": 0.0129, + "llm_learning_rate": 2.0254525468270895e-06, + "loss": 0.7515, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00020254525468270897, + "step": 5602 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2667, + "image_masks_loss": 0.0133, + "llm_learning_rate": 2.0246380394107636e-06, + "loss": 0.744, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.00020246380394107637, + "step": 5603 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2666, + "image_masks_loss": 0.0238, + "llm_learning_rate": 2.023823584342356e-06, + "loss": 0.7192, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00020238235843423558, + "step": 5604 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2666, + "image_masks_loss": 0.0171, + "llm_learning_rate": 2.0230091817115556e-06, + "loss": 0.6837, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.00020230091817115555, + "step": 5605 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2665, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.022194831608046e-06, + "loss": 0.7612, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00020221948316080456, + "step": 5606 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2664, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.021380534121506e-06, + "loss": 0.7956, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.0002021380534121506, + "step": 5607 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2664, + "image_masks_loss": 0.0049, + "llm_learning_rate": 2.0205662893416086e-06, + "loss": 0.7688, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00020205662893416082, + "step": 5608 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2663, + "image_masks_loss": 0.004, + "llm_learning_rate": 2.0197520973580183e-06, + "loss": 0.806, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002019752097358018, + "step": 5609 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2663, + "image_masks_loss": 0.0041, + "llm_learning_rate": 2.0189379582603967e-06, + "loss": 0.728, + "output_text_mask_loss": 0.0036, + "predictor_learning_rate": 0.00020189379582603967, + "step": 5610 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2662, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.0181238721383992e-06, + "loss": 0.7435, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020181238721383993, + "step": 5611 + }, + { + "epoch": 0.57, + "gumbel_tau": 0.2661, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.0173098390816747e-06, + "loss": 0.8119, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00020173098390816743, + "step": 5612 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2661, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.0164958591798645e-06, + "loss": 0.6248, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00020164958591798644, + "step": 5613 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.266, + "image_masks_loss": 0.0138, + "llm_learning_rate": 2.015681932522608e-06, + "loss": 0.753, + "output_text_mask_loss": 0.0443, + "predictor_learning_rate": 0.0002015681932522608, + "step": 5614 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2659, + "image_masks_loss": 0.0019, + "llm_learning_rate": 2.0148680591995362e-06, + "loss": 0.8073, + "output_text_mask_loss": 0.0458, + "predictor_learning_rate": 0.00020148680591995362, + "step": 5615 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2659, + "image_masks_loss": 0.0076, + "llm_learning_rate": 2.0140542393002736e-06, + "loss": 0.7698, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00020140542393002733, + "step": 5616 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2658, + "image_masks_loss": 0.0111, + "llm_learning_rate": 2.013240472914441e-06, + "loss": 0.6803, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0002013240472914441, + "step": 5617 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2658, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.012426760131652e-06, + "loss": 0.7152, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00020124267601316517, + "step": 5618 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2657, + "image_masks_loss": 0.015, + "llm_learning_rate": 2.011613101041514e-06, + "loss": 0.7357, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00020116131010415136, + "step": 5619 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2656, + "image_masks_loss": 0.003, + "llm_learning_rate": 2.0107994957336295e-06, + "loss": 0.6621, + "output_text_mask_loss": 0.0336, + "predictor_learning_rate": 0.00020107994957336295, + "step": 5620 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2656, + "image_masks_loss": 0.0054, + "llm_learning_rate": 2.0099859442975937e-06, + "loss": 0.6905, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00020099859442975938, + "step": 5621 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2655, + "image_masks_loss": 0.0196, + "llm_learning_rate": 2.009172446822997e-06, + "loss": 0.7959, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.00020091724468229968, + "step": 5622 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2654, + "image_masks_loss": 0.0031, + "llm_learning_rate": 2.008359003399424e-06, + "loss": 0.7296, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0002008359003399424, + "step": 5623 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2654, + "image_masks_loss": 0.0115, + "llm_learning_rate": 2.007545614116453e-06, + "loss": 0.6737, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.00020075456141164526, + "step": 5624 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2653, + "image_masks_loss": 0.0075, + "llm_learning_rate": 2.0067322790636546e-06, + "loss": 0.7175, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.00020067322790636547, + "step": 5625 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2653, + "image_masks_loss": 0.008, + "llm_learning_rate": 2.0059189983305976e-06, + "loss": 0.688, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00020059189983305975, + "step": 5626 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2652, + "image_masks_loss": 0.0059, + "llm_learning_rate": 2.0051057720068405e-06, + "loss": 0.8137, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00020051057720068405, + "step": 5627 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2651, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.0042926001819383e-06, + "loss": 0.6796, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00020042926001819384, + "step": 5628 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2651, + "image_masks_loss": 0.0072, + "llm_learning_rate": 2.003479482945439e-06, + "loss": 0.7462, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00020034794829454388, + "step": 5629 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.265, + "image_masks_loss": 0.0035, + "llm_learning_rate": 2.002666420386885e-06, + "loss": 0.7261, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.0002002666420386885, + "step": 5630 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2649, + "image_masks_loss": 0.0024, + "llm_learning_rate": 2.001853412595813e-06, + "loss": 0.736, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00020018534125958127, + "step": 5631 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2649, + "image_masks_loss": 0.0082, + "llm_learning_rate": 2.0010404596617516e-06, + "loss": 0.7761, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00020010404596617514, + "step": 5632 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2648, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.000227561674227e-06, + "loss": 0.7837, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0002000227561674227, + "step": 5633 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2648, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.999414718722757e-06, + "loss": 0.8049, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 0.00019994147187227569, + "step": 5634 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2647, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.9986019308968527e-06, + "loss": 0.6771, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00019986019308968525, + "step": 5635 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2646, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.99778919828602e-06, + "loss": 0.7178, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.000199778919828602, + "step": 5636 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2646, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.99697652097976e-06, + "loss": 0.6994, + "output_text_mask_loss": 0.0439, + "predictor_learning_rate": 0.000199697652097976, + "step": 5637 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2645, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.9961638990675666e-06, + "loss": 0.7511, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00019961638990675666, + "step": 5638 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2644, + "image_masks_loss": 0.0248, + "llm_learning_rate": 1.9953513326389263e-06, + "loss": 0.7358, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0001995351332638926, + "step": 5639 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2644, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.9945388217833215e-06, + "loss": 0.8129, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00019945388217833214, + "step": 5640 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2643, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.9937263665902267e-06, + "loss": 0.6911, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00019937263665902266, + "step": 5641 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2643, + "image_masks_loss": 0.0184, + "llm_learning_rate": 1.9929139671491125e-06, + "loss": 0.754, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.00019929139671491126, + "step": 5642 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2642, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.9921016235494423e-06, + "loss": 0.7553, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 0.00019921016235494423, + "step": 5643 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2641, + "image_masks_loss": 0.0167, + "llm_learning_rate": 1.991289335880672e-06, + "loss": 0.7491, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.0001991289335880672, + "step": 5644 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2641, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.990477104232253e-06, + "loss": 0.7698, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.0001990477104232253, + "step": 5645 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.264, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.9896649286936306e-06, + "loss": 0.6782, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00019896649286936304, + "step": 5646 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2639, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.988852809354243e-06, + "loss": 0.7461, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0001988852809354243, + "step": 5647 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2639, + "image_masks_loss": 0.0171, + "llm_learning_rate": 1.988040746303522e-06, + "loss": 0.8162, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00019880407463035218, + "step": 5648 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2638, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.9872287396308947e-06, + "loss": 0.7299, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.00019872287396308948, + "step": 5649 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2638, + "image_masks_loss": 0.0186, + "llm_learning_rate": 1.9864167894257814e-06, + "loss": 0.71, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019864167894257813, + "step": 5650 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2637, + "image_masks_loss": 0.015, + "llm_learning_rate": 1.9856048957775946e-06, + "loss": 0.6416, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00019856048957775943, + "step": 5651 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2636, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.984793058775742e-06, + "loss": 0.7492, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0001984793058775742, + "step": 5652 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2636, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.983981278509626e-06, + "loss": 0.8077, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00019839812785096258, + "step": 5653 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2635, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.9831695550686407e-06, + "loss": 0.7333, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019831695550686406, + "step": 5654 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2634, + "image_masks_loss": 0.0135, + "llm_learning_rate": 1.982357888542175e-06, + "loss": 0.7181, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001982357888542175, + "step": 5655 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2634, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.981546279019612e-06, + "loss": 0.8189, + "output_text_mask_loss": 0.0094, + "predictor_learning_rate": 0.0001981546279019612, + "step": 5656 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2633, + "image_masks_loss": 0.0095, + "llm_learning_rate": 1.9807347265903285e-06, + "loss": 0.6839, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.00019807347265903284, + "step": 5657 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2633, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.9799232313436925e-06, + "loss": 0.7637, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.00019799232313436925, + "step": 5658 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2632, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.979111793369069e-06, + "loss": 0.6561, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001979111793369069, + "step": 5659 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2631, + "image_masks_loss": 0.0196, + "llm_learning_rate": 1.9783004127558147e-06, + "loss": 0.6959, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00019783004127558147, + "step": 5660 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2631, + "image_masks_loss": 0.0322, + "llm_learning_rate": 1.9774890895932808e-06, + "loss": 0.7961, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.00019774890895932806, + "step": 5661 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.263, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.9766778239708126e-06, + "loss": 0.6848, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00019766778239708124, + "step": 5662 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.263, + "image_masks_loss": 0.0418, + "llm_learning_rate": 1.975866615977748e-06, + "loss": 0.7178, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001975866615977748, + "step": 5663 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2629, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.9750554657034186e-06, + "loss": 0.7026, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00019750554657034186, + "step": 5664 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2628, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.9742443732371514e-06, + "loss": 0.8043, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001974244373237151, + "step": 5665 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2628, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.9734333386682646e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00019734333386682646, + "step": 5666 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2627, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.9726223620860712e-06, + "loss": 0.7591, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.0001972622362086071, + "step": 5667 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2626, + "image_masks_loss": 0.0169, + "llm_learning_rate": 1.971811443579878e-06, + "loss": 0.6661, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.0001971811443579878, + "step": 5668 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2626, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.9710005832389852e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00019710005832389852, + "step": 5669 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2625, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.970189781152687e-06, + "loss": 0.7232, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.00019701897811526866, + "step": 5670 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2625, + "image_masks_loss": 0.0045, + "llm_learning_rate": 1.969379037410269e-06, + "loss": 0.6732, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019693790374102687, + "step": 5671 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2624, + "image_masks_loss": 0.0023, + "llm_learning_rate": 1.9685683521010142e-06, + "loss": 0.843, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0001968568352101014, + "step": 5672 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2623, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.967757725314196e-06, + "loss": 0.8339, + "output_text_mask_loss": 0.0568, + "predictor_learning_rate": 0.00019677577253141958, + "step": 5673 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2623, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.966947157139083e-06, + "loss": 0.7666, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001966947157139083, + "step": 5674 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2622, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.9661366476649367e-06, + "loss": 0.6432, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019661366476649367, + "step": 5675 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2621, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.9653261969810113e-06, + "loss": 0.7665, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00019653261969810112, + "step": 5676 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2621, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.9645158051765556e-06, + "loss": 0.6213, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00019645158051765556, + "step": 5677 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.262, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.963705472340813e-06, + "loss": 0.7474, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00019637054723408126, + "step": 5678 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.262, + "image_masks_loss": 0.0192, + "llm_learning_rate": 1.9628951985630177e-06, + "loss": 0.7016, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00019628951985630174, + "step": 5679 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2619, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.962084983932399e-06, + "loss": 0.718, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001962084983932399, + "step": 5680 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2618, + "image_masks_loss": 0.0195, + "llm_learning_rate": 1.961274828538181e-06, + "loss": 0.7623, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0001961274828538181, + "step": 5681 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2618, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.9604647324695786e-06, + "loss": 0.7141, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00019604647324695786, + "step": 5682 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2617, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.9596546958158017e-06, + "loss": 0.7557, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019596546958158015, + "step": 5683 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2617, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.958844718666053e-06, + "loss": 0.7427, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00019588447186660526, + "step": 5684 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2616, + "image_masks_loss": 0.0232, + "llm_learning_rate": 1.9580348011095295e-06, + "loss": 0.8044, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.00019580348011095293, + "step": 5685 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2615, + "image_masks_loss": 0.032, + "llm_learning_rate": 1.9572249432354206e-06, + "loss": 0.7383, + "output_text_mask_loss": 0.0387, + "predictor_learning_rate": 0.00019572249432354207, + "step": 5686 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2615, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.95641514513291e-06, + "loss": 0.7312, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019564151451329095, + "step": 5687 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2614, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.9556054068911747e-06, + "loss": 0.7608, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019556054068911743, + "step": 5688 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2613, + "image_masks_loss": 0.0767, + "llm_learning_rate": 1.954795728599384e-06, + "loss": 0.7754, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0001954795728599384, + "step": 5689 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2613, + "image_masks_loss": 0.0534, + "llm_learning_rate": 1.953986110346703e-06, + "loss": 0.6774, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00019539861103467026, + "step": 5690 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2612, + "image_masks_loss": 0.0265, + "llm_learning_rate": 1.9531765522222863e-06, + "loss": 0.6504, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019531765522222863, + "step": 5691 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2612, + "image_masks_loss": 0.0184, + "llm_learning_rate": 1.9523670543152867e-06, + "loss": 0.6799, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019523670543152865, + "step": 5692 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2611, + "image_masks_loss": 0.033, + "llm_learning_rate": 1.951557616714847e-06, + "loss": 0.7005, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001951557616714847, + "step": 5693 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.261, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.9507482395101044e-06, + "loss": 0.7698, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00019507482395101044, + "step": 5694 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.261, + "image_masks_loss": 0.0595, + "llm_learning_rate": 1.9499389227901887e-06, + "loss": 0.7298, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00019499389227901887, + "step": 5695 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2609, + "image_masks_loss": 0.0549, + "llm_learning_rate": 1.9491296666442236e-06, + "loss": 0.7629, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019491296666442236, + "step": 5696 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2609, + "image_masks_loss": 0.0202, + "llm_learning_rate": 1.948320471161327e-06, + "loss": 0.7803, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0001948320471161327, + "step": 5697 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2608, + "image_masks_loss": 0.0271, + "llm_learning_rate": 1.947511336430609e-06, + "loss": 0.7725, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0001947511336430609, + "step": 5698 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2607, + "image_masks_loss": 0.0301, + "llm_learning_rate": 1.9467022625411735e-06, + "loss": 0.8192, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001946702262541173, + "step": 5699 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2607, + "image_masks_loss": 0.0131, + "llm_learning_rate": 1.9458932495821165e-06, + "loss": 0.6976, + "output_text_mask_loss": 0.0267, + "predictor_learning_rate": 0.00019458932495821163, + "step": 5700 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2606, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.9450842976425295e-06, + "loss": 0.8007, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 0.00019450842976425291, + "step": 5701 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2605, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.9442754068114954e-06, + "loss": 0.7769, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00019442754068114954, + "step": 5702 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2605, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.9434665771780915e-06, + "loss": 0.72, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019434665771780912, + "step": 5703 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2604, + "image_masks_loss": 0.0332, + "llm_learning_rate": 1.9426578088313877e-06, + "loss": 0.7941, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.00019426578088313875, + "step": 5704 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2604, + "image_masks_loss": 0.0181, + "llm_learning_rate": 1.9418491018604477e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019418491018604475, + "step": 5705 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2603, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.9410404563543278e-06, + "loss": 0.7387, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 0.00019410404563543278, + "step": 5706 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2602, + "image_masks_loss": 0.0271, + "llm_learning_rate": 1.9402318724020773e-06, + "loss": 0.7217, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00019402318724020772, + "step": 5707 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2602, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.9394233500927406e-06, + "loss": 0.5781, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019394233500927405, + "step": 5708 + }, + { + "epoch": 0.58, + "gumbel_tau": 0.2601, + "image_masks_loss": 0.0237, + "llm_learning_rate": 1.9386148895153527e-06, + "loss": 0.7364, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019386148895153527, + "step": 5709 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2601, + "image_masks_loss": 0.0406, + "llm_learning_rate": 1.9378064907589436e-06, + "loss": 0.6749, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00019378064907589436, + "step": 5710 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.26, + "image_masks_loss": 0.0162, + "llm_learning_rate": 1.9369981539125364e-06, + "loss": 0.7732, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00019369981539125365, + "step": 5711 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2599, + "image_masks_loss": 0.034, + "llm_learning_rate": 1.936189879065147e-06, + "loss": 0.7819, + "output_text_mask_loss": 0.0427, + "predictor_learning_rate": 0.0001936189879065147, + "step": 5712 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2599, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.9353816663057835e-06, + "loss": 0.7109, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019353816663057833, + "step": 5713 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2598, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.934573515723448e-06, + "loss": 0.7786, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.00019345735157234482, + "step": 5714 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2597, + "image_masks_loss": 0.0303, + "llm_learning_rate": 1.933765427407137e-06, + "loss": 0.8105, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0001933765427407137, + "step": 5715 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2597, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.932957401445838e-06, + "loss": 0.7993, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0001932957401445838, + "step": 5716 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2596, + "image_masks_loss": 0.0254, + "llm_learning_rate": 1.932149437928533e-06, + "loss": 0.8023, + "output_text_mask_loss": 0.0372, + "predictor_learning_rate": 0.0001932149437928533, + "step": 5717 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2596, + "image_masks_loss": 0.0151, + "llm_learning_rate": 1.931341536944197e-06, + "loss": 0.7198, + "output_text_mask_loss": 0.0257, + "predictor_learning_rate": 0.00019313415369441966, + "step": 5718 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2595, + "image_masks_loss": 0.0188, + "llm_learning_rate": 1.930533698581797e-06, + "loss": 0.6785, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 0.00019305336985817967, + "step": 5719 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2594, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.929725922930295e-06, + "loss": 0.7729, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001929725922930295, + "step": 5720 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2594, + "image_masks_loss": 0.0164, + "llm_learning_rate": 1.9289182100786445e-06, + "loss": 0.6356, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019289182100786444, + "step": 5721 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2593, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.928110560115793e-06, + "loss": 0.7413, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001928110560115793, + "step": 5722 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2593, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.92730297313068e-06, + "loss": 0.7039, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.000192730297313068, + "step": 5723 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2592, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.9264954492122394e-06, + "loss": 0.7632, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00019264954492122395, + "step": 5724 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2591, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.9256879884493977e-06, + "loss": 0.7264, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019256879884493975, + "step": 5725 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2591, + "image_masks_loss": 0.0147, + "llm_learning_rate": 1.924880590931074e-06, + "loss": 0.7778, + "output_text_mask_loss": 0.025, + "predictor_learning_rate": 0.00019248805909310738, + "step": 5726 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.259, + "image_masks_loss": 0.0227, + "llm_learning_rate": 1.9240732567461796e-06, + "loss": 0.6905, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00019240732567461796, + "step": 5727 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.259, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.923265985983622e-06, + "loss": 0.6314, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001923265985983622, + "step": 5728 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2589, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.9224587787322985e-06, + "loss": 0.6954, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019224587787322983, + "step": 5729 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2588, + "image_masks_loss": 0.0139, + "llm_learning_rate": 1.9216516350811e-06, + "loss": 0.7694, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00019216516350811, + "step": 5730 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2588, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.9208445551189135e-06, + "loss": 0.8044, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019208445551189136, + "step": 5731 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2587, + "image_masks_loss": 0.0238, + "llm_learning_rate": 1.920037538934613e-06, + "loss": 0.7215, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00019200375389346127, + "step": 5732 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2586, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.919230586617071e-06, + "loss": 0.7144, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001919230586617071, + "step": 5733 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2586, + "image_masks_loss": 0.0192, + "llm_learning_rate": 1.9184236982551504e-06, + "loss": 0.7305, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00019184236982551503, + "step": 5734 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2585, + "image_masks_loss": 0.0222, + "llm_learning_rate": 1.9176168739377076e-06, + "loss": 0.7435, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019176168739377077, + "step": 5735 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2585, + "image_masks_loss": 0.0216, + "llm_learning_rate": 1.9168101137535914e-06, + "loss": 0.8323, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00019168101137535912, + "step": 5736 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2584, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.9160034177916446e-06, + "loss": 0.7754, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.00019160034177916445, + "step": 5737 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2583, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.9151967861407026e-06, + "loss": 0.772, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00019151967861407026, + "step": 5738 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2583, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.9143902188895927e-06, + "loss": 0.744, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00019143902188895927, + "step": 5739 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2582, + "image_masks_loss": 0.0194, + "llm_learning_rate": 1.9135837161271367e-06, + "loss": 0.5655, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.00019135837161271367, + "step": 5740 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2582, + "image_masks_loss": 0.0155, + "llm_learning_rate": 1.912777277942148e-06, + "loss": 0.7279, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0001912777277942148, + "step": 5741 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2581, + "image_masks_loss": 0.0374, + "llm_learning_rate": 1.9119709044234334e-06, + "loss": 0.6726, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019119709044234336, + "step": 5742 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.258, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.911164595659792e-06, + "loss": 0.766, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.0001911164595659792, + "step": 5743 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.258, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.9103583517400177e-06, + "loss": 0.7319, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.00019103583517400175, + "step": 5744 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2579, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.909552172752895e-06, + "loss": 0.7382, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.00019095521727528948, + "step": 5745 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2579, + "image_masks_loss": 0.0182, + "llm_learning_rate": 1.9087460587872015e-06, + "loss": 0.7268, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.00019087460587872014, + "step": 5746 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2578, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.9079400099317097e-06, + "loss": 0.7625, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00019079400099317095, + "step": 5747 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2577, + "image_masks_loss": 0.0025, + "llm_learning_rate": 1.9071340262751825e-06, + "loss": 0.7693, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.00019071340262751823, + "step": 5748 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2577, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.9063281079063772e-06, + "loss": 0.837, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0001906328107906377, + "step": 5749 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2576, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.9055222549140435e-06, + "loss": 0.7538, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00019055222549140433, + "step": 5750 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2575, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.9047164673869226e-06, + "loss": 0.7065, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00019047164673869225, + "step": 5751 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2575, + "image_masks_loss": 0.0214, + "llm_learning_rate": 1.9039107454137496e-06, + "loss": 0.7589, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00019039107454137493, + "step": 5752 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2574, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.9031050890832536e-06, + "loss": 0.7667, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00019031050890832535, + "step": 5753 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2574, + "image_masks_loss": 0.0216, + "llm_learning_rate": 1.902299498484155e-06, + "loss": 0.7644, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019022994984841548, + "step": 5754 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2573, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.9014939737051663e-06, + "loss": 0.7563, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00019014939737051662, + "step": 5755 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2572, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.9006885148349946e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00019006885148349946, + "step": 5756 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2572, + "image_masks_loss": 0.0234, + "llm_learning_rate": 1.899883121962339e-06, + "loss": 0.7348, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018998831219623388, + "step": 5757 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2571, + "image_masks_loss": 0.0309, + "llm_learning_rate": 1.8990777951758904e-06, + "loss": 0.7084, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00018990777951758902, + "step": 5758 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2571, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.8982725345643332e-06, + "loss": 0.6935, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0001898272534564333, + "step": 5759 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.257, + "image_masks_loss": 0.0173, + "llm_learning_rate": 1.897467340216345e-06, + "loss": 0.7397, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0001897467340216345, + "step": 5760 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2569, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.8966622122205957e-06, + "loss": 0.6817, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018966622122205956, + "step": 5761 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2569, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.8958571506657468e-06, + "loss": 0.7614, + "output_text_mask_loss": 0.119, + "predictor_learning_rate": 0.00018958571506657467, + "step": 5762 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2568, + "image_masks_loss": 0.0178, + "llm_learning_rate": 1.8950521556404548e-06, + "loss": 0.7792, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.00018950521556404545, + "step": 5763 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2568, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.8942472272333668e-06, + "loss": 0.7835, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.00018942472272333666, + "step": 5764 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2567, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.8934423655331232e-06, + "loss": 0.704, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00018934423655331232, + "step": 5765 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2566, + "image_masks_loss": 0.0257, + "llm_learning_rate": 1.892637570628357e-06, + "loss": 0.7987, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001892637570628357, + "step": 5766 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2566, + "image_masks_loss": 0.0368, + "llm_learning_rate": 1.8918328426076948e-06, + "loss": 0.7764, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00018918328426076949, + "step": 5767 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2565, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.8910281815597554e-06, + "loss": 0.7439, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018910281815597552, + "step": 5768 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2565, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.8902235875731485e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.0542, + "predictor_learning_rate": 0.00018902235875731482, + "step": 5769 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2564, + "image_masks_loss": 0.0252, + "llm_learning_rate": 1.8894190607364782e-06, + "loss": 0.7654, + "output_text_mask_loss": 0.0383, + "predictor_learning_rate": 0.0001889419060736478, + "step": 5770 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2563, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.8886146011383405e-06, + "loss": 0.7775, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00018886146011383404, + "step": 5771 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2563, + "image_masks_loss": 0.028, + "llm_learning_rate": 1.887810208867325e-06, + "loss": 0.7582, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018878102088673248, + "step": 5772 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2562, + "image_masks_loss": 0.0219, + "llm_learning_rate": 1.8870058840120133e-06, + "loss": 0.73, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.00018870058840120133, + "step": 5773 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2562, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.886201626660979e-06, + "loss": 0.6256, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 0.00018862016266609787, + "step": 5774 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2561, + "image_masks_loss": 0.0171, + "llm_learning_rate": 1.885397436902788e-06, + "loss": 0.7181, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00018853974369027878, + "step": 5775 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.256, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.884593314826001e-06, + "loss": 0.7548, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00018845933148260008, + "step": 5776 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.256, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.8837892605191688e-06, + "loss": 0.6882, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00018837892605191688, + "step": 5777 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2559, + "image_masks_loss": 0.016, + "llm_learning_rate": 1.8829852740708352e-06, + "loss": 0.7196, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0001882985274070835, + "step": 5778 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2559, + "image_masks_loss": 0.0116, + "llm_learning_rate": 1.882181355569538e-06, + "loss": 0.6485, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018821813555695377, + "step": 5779 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2558, + "image_masks_loss": 0.0317, + "llm_learning_rate": 1.8813775051038064e-06, + "loss": 0.6298, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00018813775051038063, + "step": 5780 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2557, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.8805737227621615e-06, + "loss": 0.7045, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018805737227621614, + "step": 5781 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2557, + "image_masks_loss": 0.0185, + "llm_learning_rate": 1.8797700086331178e-06, + "loss": 0.7914, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00018797700086331176, + "step": 5782 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2556, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.8789663628051824e-06, + "loss": 0.6189, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 0.00018789663628051824, + "step": 5783 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2556, + "image_masks_loss": 0.0336, + "llm_learning_rate": 1.8781627853668547e-06, + "loss": 0.702, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018781627853668546, + "step": 5784 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2555, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.8773592764066253e-06, + "loss": 0.7622, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00018773592764066252, + "step": 5785 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2554, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.8765558360129802e-06, + "loss": 0.7877, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.000187655583601298, + "step": 5786 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2554, + "image_masks_loss": 0.0181, + "llm_learning_rate": 1.8757524642743955e-06, + "loss": 0.7777, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.00018757524642743954, + "step": 5787 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2553, + "image_masks_loss": 0.038, + "llm_learning_rate": 1.8749491612793394e-06, + "loss": 0.8328, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00018749491612793392, + "step": 5788 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2553, + "image_masks_loss": 0.0168, + "llm_learning_rate": 1.8741459271162737e-06, + "loss": 0.7487, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.00018741459271162737, + "step": 5789 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2552, + "image_masks_loss": 0.0299, + "llm_learning_rate": 1.8733427618736528e-06, + "loss": 0.7698, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018733427618736526, + "step": 5790 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2551, + "image_masks_loss": 0.0174, + "llm_learning_rate": 1.872539665639922e-06, + "loss": 0.7472, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.0001872539665639922, + "step": 5791 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2551, + "image_masks_loss": 0.0222, + "llm_learning_rate": 1.8717366385035218e-06, + "loss": 0.7121, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00018717366385035216, + "step": 5792 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.255, + "image_masks_loss": 0.0114, + "llm_learning_rate": 1.8709336805528819e-06, + "loss": 0.7357, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018709336805528817, + "step": 5793 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2549, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.8701307918764261e-06, + "loss": 0.806, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0001870130791876426, + "step": 5794 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2549, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.8693279725625708e-06, + "loss": 0.7659, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.00018693279725625706, + "step": 5795 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2548, + "image_masks_loss": 0.0278, + "llm_learning_rate": 1.868525222699724e-06, + "loss": 0.8339, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00018685252226997238, + "step": 5796 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2548, + "image_masks_loss": 0.0311, + "llm_learning_rate": 1.8677225423762863e-06, + "loss": 0.7122, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00018677225423762862, + "step": 5797 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2547, + "image_masks_loss": 0.0231, + "llm_learning_rate": 1.8669199316806497e-06, + "loss": 0.7809, + "output_text_mask_loss": 0.0452, + "predictor_learning_rate": 0.00018669199316806497, + "step": 5798 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2546, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.8661173907012012e-06, + "loss": 0.8006, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00018661173907012012, + "step": 5799 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2546, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.8653149195263177e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00018653149195263174, + "step": 5800 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2545, + "image_masks_loss": 0.0257, + "llm_learning_rate": 1.864512518244368e-06, + "loss": 0.757, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.00018645125182443679, + "step": 5801 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2545, + "image_masks_loss": 0.0261, + "llm_learning_rate": 1.863710186943716e-06, + "loss": 0.7035, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00018637101869437158, + "step": 5802 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2544, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.8629079257127156e-06, + "loss": 0.6981, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.00018629079257127153, + "step": 5803 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2543, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.8621057346397133e-06, + "loss": 0.6825, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00018621057346397132, + "step": 5804 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2543, + "image_masks_loss": 0.0034, + "llm_learning_rate": 1.861303613813049e-06, + "loss": 0.802, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.0001861303613813049, + "step": 5805 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2542, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.8605015633210529e-06, + "loss": 0.7068, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0001860501563321053, + "step": 5806 + }, + { + "epoch": 0.59, + "gumbel_tau": 0.2542, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.8596995832520487e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00018596995832520486, + "step": 5807 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2541, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.8588976736943525e-06, + "loss": 0.7695, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00018588976736943526, + "step": 5808 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.254, + "image_masks_loss": 0.0292, + "llm_learning_rate": 1.8580958347362732e-06, + "loss": 0.8023, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00018580958347362732, + "step": 5809 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.254, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.8572940664661094e-06, + "loss": 0.7782, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00018572940664661093, + "step": 5810 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2539, + "image_masks_loss": 0.0239, + "llm_learning_rate": 1.8564923689721554e-06, + "loss": 0.7502, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.00018564923689721552, + "step": 5811 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2539, + "image_masks_loss": 0.0179, + "llm_learning_rate": 1.855690742342695e-06, + "loss": 0.6643, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.0001855690742342695, + "step": 5812 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2538, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.8548891866660052e-06, + "loss": 0.7362, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.0001854889186666005, + "step": 5813 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2537, + "image_masks_loss": 0.0153, + "llm_learning_rate": 1.8540877020303543e-06, + "loss": 0.7279, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 0.00018540877020303543, + "step": 5814 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2537, + "image_masks_loss": 0.0215, + "llm_learning_rate": 1.853286288524005e-06, + "loss": 0.7128, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00018532862885240048, + "step": 5815 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2536, + "image_masks_loss": 0.0257, + "llm_learning_rate": 1.8524849462352101e-06, + "loss": 0.7668, + "output_text_mask_loss": 0.0824, + "predictor_learning_rate": 0.00018524849462352101, + "step": 5816 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2536, + "image_masks_loss": 0.0231, + "llm_learning_rate": 1.851683675252215e-06, + "loss": 0.7863, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00018516836752522148, + "step": 5817 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2535, + "image_masks_loss": 0.0045, + "llm_learning_rate": 1.850882475663258e-06, + "loss": 0.8134, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00018508824756632578, + "step": 5818 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2535, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.8500813475565685e-06, + "loss": 0.7377, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 0.00018500813475565684, + "step": 5819 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2534, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.8492802910203688e-06, + "loss": 0.6844, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018492802910203686, + "step": 5820 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2533, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.8484793061428725e-06, + "loss": 0.6363, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 0.00018484793061428724, + "step": 5821 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2533, + "image_masks_loss": 0.011, + "llm_learning_rate": 1.8476783930122862e-06, + "loss": 0.6788, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.00018476783930122863, + "step": 5822 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2532, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.8468775517168087e-06, + "loss": 0.7245, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00018468775517168086, + "step": 5823 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2532, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.8460767823446302e-06, + "loss": 0.7832, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.000184607678234463, + "step": 5824 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2531, + "image_masks_loss": 0.015, + "llm_learning_rate": 1.8452760849839327e-06, + "loss": 0.7416, + "output_text_mask_loss": 0.0225, + "predictor_learning_rate": 0.00018452760849839325, + "step": 5825 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.253, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.8444754597228904e-06, + "loss": 0.7704, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.00018444754597228902, + "step": 5826 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.253, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.8436749066496711e-06, + "loss": 0.6456, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0001843674906649671, + "step": 5827 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2529, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.8428744258524335e-06, + "loss": 0.6865, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00018428744258524332, + "step": 5828 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2529, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.8420740174193273e-06, + "loss": 0.589, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.0001842074017419327, + "step": 5829 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2528, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.8412736814384952e-06, + "loss": 0.7919, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001841273681438495, + "step": 5830 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2527, + "image_masks_loss": 0.0191, + "llm_learning_rate": 1.8404734179980731e-06, + "loss": 0.7935, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00018404734179980732, + "step": 5831 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2527, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.8396732271861875e-06, + "loss": 0.7071, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.00018396732271861875, + "step": 5832 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2526, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.8388731090909561e-06, + "loss": 0.6721, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0001838873109090956, + "step": 5833 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2526, + "image_masks_loss": 0.0153, + "llm_learning_rate": 1.8380730638004913e-06, + "loss": 0.7717, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018380730638004912, + "step": 5834 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2525, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.8372730914028953e-06, + "loss": 0.8206, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.00018372730914028952, + "step": 5835 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2524, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.8364731919862623e-06, + "loss": 0.8513, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00018364731919862622, + "step": 5836 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2524, + "image_masks_loss": 0.0224, + "llm_learning_rate": 1.8356733656386788e-06, + "loss": 0.737, + "output_text_mask_loss": 0.0237, + "predictor_learning_rate": 0.00018356733656386788, + "step": 5837 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2523, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.8348736124482246e-06, + "loss": 0.7284, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00018348736124482247, + "step": 5838 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2523, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.8340739325029699e-06, + "loss": 0.7292, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018340739325029698, + "step": 5839 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2522, + "image_masks_loss": 0.0115, + "llm_learning_rate": 1.8332743258909764e-06, + "loss": 0.7353, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00018332743258909763, + "step": 5840 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2521, + "image_masks_loss": 0.0047, + "llm_learning_rate": 1.8324747927002996e-06, + "loss": 0.7923, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00018324747927002993, + "step": 5841 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2521, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.831675333018986e-06, + "loss": 0.762, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00018316753330189858, + "step": 5842 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.252, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.8308759469350728e-06, + "loss": 0.6914, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00018308759469350727, + "step": 5843 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.252, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.8300766345365911e-06, + "loss": 0.6929, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.0001830076634536591, + "step": 5844 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2519, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.8292773959115623e-06, + "loss": 0.6645, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00018292773959115621, + "step": 5845 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2518, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.8284782311480004e-06, + "loss": 0.7301, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00018284782311480004, + "step": 5846 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2518, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.8276791403339116e-06, + "loss": 0.7238, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.00018276791403339115, + "step": 5847 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2517, + "image_masks_loss": 0.0209, + "llm_learning_rate": 1.8268801235572935e-06, + "loss": 0.6474, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00018268801235572934, + "step": 5848 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2517, + "image_masks_loss": 0.0174, + "llm_learning_rate": 1.8260811809061352e-06, + "loss": 0.7075, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001826081180906135, + "step": 5849 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2516, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.8252823124684185e-06, + "loss": 0.7752, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.00018252823124684185, + "step": 5850 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2515, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.824483518332117e-06, + "loss": 0.6436, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00018244835183321167, + "step": 5851 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2515, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.8236847985851947e-06, + "loss": 0.7148, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00018236847985851946, + "step": 5852 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2514, + "image_masks_loss": 0.0173, + "llm_learning_rate": 1.822886153315608e-06, + "loss": 0.7061, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0001822886153315608, + "step": 5853 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2514, + "image_masks_loss": 0.0192, + "llm_learning_rate": 1.8220875826113072e-06, + "loss": 0.7636, + "output_text_mask_loss": 0.0648, + "predictor_learning_rate": 0.0001822087582611307, + "step": 5854 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2513, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.8212890865602317e-06, + "loss": 0.6866, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00018212890865602316, + "step": 5855 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2512, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.8204906652503136e-06, + "loss": 0.7182, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 0.00018204906652503132, + "step": 5856 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2512, + "image_masks_loss": 0.0145, + "llm_learning_rate": 1.8196923187694771e-06, + "loss": 0.7032, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001819692318769477, + "step": 5857 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2511, + "image_masks_loss": 0.0207, + "llm_learning_rate": 1.818894047205638e-06, + "loss": 0.7613, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00018188940472056378, + "step": 5858 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2511, + "image_masks_loss": 0.0181, + "llm_learning_rate": 1.8180958506467033e-06, + "loss": 0.8664, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00018180958506467033, + "step": 5859 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.251, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.817297729180572e-06, + "loss": 0.7772, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0001817297729180572, + "step": 5860 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.251, + "image_masks_loss": 0.0257, + "llm_learning_rate": 1.816499682895137e-06, + "loss": 0.7443, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00018164996828951368, + "step": 5861 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2509, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.815701711878278e-06, + "loss": 0.658, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.0001815701711878278, + "step": 5862 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2508, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.8149038162178711e-06, + "loss": 0.6964, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001814903816217871, + "step": 5863 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2508, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.8141059960017818e-06, + "loss": 0.7249, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00018141059960017817, + "step": 5864 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2507, + "image_masks_loss": 0.0463, + "llm_learning_rate": 1.8133082513178674e-06, + "loss": 0.8429, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00018133082513178674, + "step": 5865 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2507, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.8125105822539785e-06, + "loss": 0.7739, + "output_text_mask_loss": 0.032, + "predictor_learning_rate": 0.00018125105822539784, + "step": 5866 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2506, + "image_masks_loss": 0.0167, + "llm_learning_rate": 1.8117129888979556e-06, + "loss": 0.7084, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 0.00018117129888979555, + "step": 5867 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2505, + "image_masks_loss": 0.0145, + "llm_learning_rate": 1.8109154713376318e-06, + "loss": 0.7329, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00018109154713376316, + "step": 5868 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2505, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.8101180296608302e-06, + "loss": 0.7871, + "output_text_mask_loss": 0.1389, + "predictor_learning_rate": 0.00018101180296608302, + "step": 5869 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2504, + "image_masks_loss": 0.0147, + "llm_learning_rate": 1.8093206639553688e-06, + "loss": 0.7669, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.00018093206639553685, + "step": 5870 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2504, + "image_masks_loss": 0.0215, + "llm_learning_rate": 1.8085233743090542e-06, + "loss": 0.7279, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001808523374309054, + "step": 5871 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2503, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.807726160809685e-06, + "loss": 0.6467, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001807726160809685, + "step": 5872 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2502, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.8069290235450538e-06, + "loss": 0.6616, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00018069290235450536, + "step": 5873 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2502, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.8061319626029416e-06, + "loss": 0.7207, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00018061319626029416, + "step": 5874 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2501, + "image_masks_loss": 0.0237, + "llm_learning_rate": 1.8053349780711238e-06, + "loss": 0.7791, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00018053349780711237, + "step": 5875 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2501, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.8045380700373648e-06, + "loss": 0.8347, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 0.00018045380700373648, + "step": 5876 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.25, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.8037412385894238e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018037412385894235, + "step": 5877 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2499, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.8029444838150477e-06, + "loss": 0.7678, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00018029444838150478, + "step": 5878 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2499, + "image_masks_loss": 0.0197, + "llm_learning_rate": 1.8021478058019784e-06, + "loss": 0.7278, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00018021478058019783, + "step": 5879 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2498, + "image_masks_loss": 0.0265, + "llm_learning_rate": 1.8013512046379467e-06, + "loss": 0.7181, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00018013512046379464, + "step": 5880 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2498, + "image_masks_loss": 0.0303, + "llm_learning_rate": 1.8005546804106758e-06, + "loss": 0.805, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.00018005546804106756, + "step": 5881 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2497, + "image_masks_loss": 0.0187, + "llm_learning_rate": 1.7997582332078823e-06, + "loss": 0.8182, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00017997582332078822, + "step": 5882 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2497, + "image_masks_loss": 0.0139, + "llm_learning_rate": 1.7989618631172714e-06, + "loss": 0.7178, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00017989618631172714, + "step": 5883 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2496, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.7981655702265422e-06, + "loss": 0.709, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0001798165570226542, + "step": 5884 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2495, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.7973693546233827e-06, + "loss": 0.625, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00017973693546233826, + "step": 5885 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2495, + "image_masks_loss": 0.016, + "llm_learning_rate": 1.7965732163954756e-06, + "loss": 0.7805, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00017965732163954755, + "step": 5886 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2494, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.7957771556304926e-06, + "loss": 0.8065, + "output_text_mask_loss": 0.0024, + "predictor_learning_rate": 0.00017957771556304924, + "step": 5887 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2494, + "image_masks_loss": 0.0204, + "llm_learning_rate": 1.7949811724160976e-06, + "loss": 0.7189, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00017949811724160975, + "step": 5888 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2493, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.7941852668399465e-06, + "loss": 0.6855, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00017941852668399463, + "step": 5889 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2492, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.793389438989686e-06, + "loss": 0.7318, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00017933894389896859, + "step": 5890 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2492, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.7925936889529544e-06, + "loss": 0.6769, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 0.00017925936889529544, + "step": 5891 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2491, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.791798016817381e-06, + "loss": 0.7024, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.0001791798016817381, + "step": 5892 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2491, + "image_masks_loss": 0.0147, + "llm_learning_rate": 1.791002422670588e-06, + "loss": 0.8008, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.00017910024226705878, + "step": 5893 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.249, + "image_masks_loss": 0.019, + "llm_learning_rate": 1.7902069066001875e-06, + "loss": 0.7521, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00017902069066001874, + "step": 5894 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2489, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.789411468693783e-06, + "loss": 0.7198, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.0001789411468693783, + "step": 5895 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2489, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.7886161090389705e-06, + "loss": 0.7842, + "output_text_mask_loss": 0.0463, + "predictor_learning_rate": 0.00017886161090389704, + "step": 5896 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2488, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.7878208277233368e-06, + "loss": 0.7493, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0001787820827723337, + "step": 5897 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2488, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.7870256248344606e-06, + "loss": 0.7551, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 0.00017870256248344604, + "step": 5898 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2487, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.7862305004599097e-06, + "loss": 0.7775, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.00017862305004599097, + "step": 5899 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2487, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.7854354546872465e-06, + "loss": 0.6612, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017854354546872463, + "step": 5900 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2486, + "image_masks_loss": 0.0178, + "llm_learning_rate": 1.7846404876040218e-06, + "loss": 0.7536, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.0001784640487604022, + "step": 5901 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2485, + "image_masks_loss": 0.0206, + "llm_learning_rate": 1.7838455992977805e-06, + "loss": 0.7542, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.00017838455992977804, + "step": 5902 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2485, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.7830507898560573e-06, + "loss": 0.7501, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017830507898560573, + "step": 5903 + }, + { + "epoch": 0.6, + "gumbel_tau": 0.2484, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.7822560593663771e-06, + "loss": 0.761, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0001782256059366377, + "step": 5904 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2484, + "image_masks_loss": 0.0166, + "llm_learning_rate": 1.781461407916259e-06, + "loss": 0.8013, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0001781461407916259, + "step": 5905 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2483, + "image_masks_loss": 0.0023, + "llm_learning_rate": 1.7806668355932113e-06, + "loss": 0.716, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 0.00017806668355932112, + "step": 5906 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2482, + "image_masks_loss": 0.0166, + "llm_learning_rate": 1.7798723424847338e-06, + "loss": 0.741, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00017798723424847338, + "step": 5907 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2482, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.7790779286783176e-06, + "loss": 0.7459, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00017790779286783174, + "step": 5908 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2481, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.778283594261446e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.00017782835942614457, + "step": 5909 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2481, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.7774893393215925e-06, + "loss": 0.7211, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.00017774893393215924, + "step": 5910 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.248, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.7766951639462222e-06, + "loss": 0.7674, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001776695163946222, + "step": 5911 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.248, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.7759010682227918e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.00017759010682227917, + "step": 5912 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2479, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.7751070522387484e-06, + "loss": 0.7333, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.00017751070522387482, + "step": 5913 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2478, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.7743131160815314e-06, + "loss": 0.7489, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00017743131160815313, + "step": 5914 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2478, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.7735192598385697e-06, + "loss": 0.7697, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00017735192598385696, + "step": 5915 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2477, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.7727254835972868e-06, + "loss": 0.7509, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00017727254835972866, + "step": 5916 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2477, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.7719317874450922e-06, + "loss": 0.6563, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001771931787445092, + "step": 5917 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2476, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.7711381714693915e-06, + "loss": 0.6959, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00017711381714693913, + "step": 5918 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2475, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.7703446357575787e-06, + "loss": 0.7124, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 0.00017703446357575787, + "step": 5919 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2475, + "image_masks_loss": 0.0035, + "llm_learning_rate": 1.7695511803970394e-06, + "loss": 0.7224, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00017695511803970392, + "step": 5920 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2474, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.7687578054751514e-06, + "loss": 0.7693, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.00017687578054751513, + "step": 5921 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2474, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.7679645110792829e-06, + "loss": 0.6551, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00017679645110792828, + "step": 5922 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2473, + "image_masks_loss": 0.0222, + "llm_learning_rate": 1.767171297296793e-06, + "loss": 0.8347, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 0.0001767171297296793, + "step": 5923 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2472, + "image_masks_loss": 0.0136, + "llm_learning_rate": 1.7663781642150313e-06, + "loss": 0.7238, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00017663781642150313, + "step": 5924 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2472, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.765585111921341e-06, + "loss": 0.829, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00017655851119213409, + "step": 5925 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2471, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.764792140503054e-06, + "loss": 0.7671, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.0001764792140503054, + "step": 5926 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2471, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.7639992500474938e-06, + "loss": 0.8072, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00017639992500474935, + "step": 5927 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.247, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.7632064406419758e-06, + "loss": 0.7703, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00017632064406419756, + "step": 5928 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.247, + "image_masks_loss": 0.0167, + "llm_learning_rate": 1.7624137123738056e-06, + "loss": 0.7598, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.00017624137123738054, + "step": 5929 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2469, + "image_masks_loss": 0.0193, + "llm_learning_rate": 1.7616210653302807e-06, + "loss": 0.7006, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00017616210653302806, + "step": 5930 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2468, + "image_masks_loss": 0.0211, + "llm_learning_rate": 1.760828499598688e-06, + "loss": 0.7092, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00017608284995986878, + "step": 5931 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2468, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.7600360152663076e-06, + "loss": 0.6528, + "output_text_mask_loss": 0.0362, + "predictor_learning_rate": 0.00017600360152663074, + "step": 5932 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2467, + "image_masks_loss": 0.0189, + "llm_learning_rate": 1.7592436124204098e-06, + "loss": 0.7549, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 0.00017592436124204096, + "step": 5933 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2467, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.758451291148255e-06, + "loss": 0.7708, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 0.0001758451291148255, + "step": 5934 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2466, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.7576590515370967e-06, + "loss": 0.7115, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.00017576590515370967, + "step": 5935 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2466, + "image_masks_loss": 0.0227, + "llm_learning_rate": 1.7568668936741761e-06, + "loss": 0.85, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001756866893674176, + "step": 5936 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2465, + "image_masks_loss": 0.0033, + "llm_learning_rate": 1.7560748176467287e-06, + "loss": 0.7055, + "output_text_mask_loss": 0.0587, + "predictor_learning_rate": 0.00017560748176467287, + "step": 5937 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2464, + "image_masks_loss": 0.0038, + "llm_learning_rate": 1.7552828235419798e-06, + "loss": 0.76, + "output_text_mask_loss": 0.0336, + "predictor_learning_rate": 0.00017552828235419796, + "step": 5938 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2464, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.7544909114471447e-06, + "loss": 0.7139, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017544909114471446, + "step": 5939 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2463, + "image_masks_loss": 0.0216, + "llm_learning_rate": 1.7536990814494306e-06, + "loss": 0.6959, + "output_text_mask_loss": 0.0387, + "predictor_learning_rate": 0.00017536990814494304, + "step": 5940 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2463, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.7529073336360364e-06, + "loss": 0.7473, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00017529073336360363, + "step": 5941 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2462, + "image_masks_loss": 0.0176, + "llm_learning_rate": 1.7521156680941509e-06, + "loss": 0.6557, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0001752115668094151, + "step": 5942 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2461, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.7513240849109534e-06, + "loss": 0.7584, + "output_text_mask_loss": 0.0492, + "predictor_learning_rate": 0.00017513240849109532, + "step": 5943 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2461, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.7505325841736158e-06, + "loss": 0.7555, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00017505325841736156, + "step": 5944 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.246, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.7497411659692993e-06, + "loss": 0.817, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00017497411659692991, + "step": 5945 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.246, + "image_masks_loss": 0.0334, + "llm_learning_rate": 1.7489498303851565e-06, + "loss": 0.7731, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00017489498303851564, + "step": 5946 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2459, + "image_masks_loss": 0.0165, + "llm_learning_rate": 1.748158577508331e-06, + "loss": 0.7007, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00017481585775083307, + "step": 5947 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2459, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.7473674074259583e-06, + "loss": 0.8311, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00017473674074259583, + "step": 5948 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2458, + "image_masks_loss": 0.0357, + "llm_learning_rate": 1.746576320225163e-06, + "loss": 0.8125, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00017465763202251626, + "step": 5949 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2457, + "image_masks_loss": 0.0196, + "llm_learning_rate": 1.7457853159930613e-06, + "loss": 0.7635, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00017457853159930613, + "step": 5950 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2457, + "image_masks_loss": 0.0157, + "llm_learning_rate": 1.7449943948167602e-06, + "loss": 0.6888, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.000174499439481676, + "step": 5951 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2456, + "image_masks_loss": 0.0252, + "llm_learning_rate": 1.7442035567833585e-06, + "loss": 0.6748, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017442035567833582, + "step": 5952 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2456, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.7434128019799452e-06, + "loss": 0.6787, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001743412801979945, + "step": 5953 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2455, + "image_masks_loss": 0.0184, + "llm_learning_rate": 1.7426221304935986e-06, + "loss": 0.6631, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00017426221304935984, + "step": 5954 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2454, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.7418315424113902e-06, + "loss": 0.7032, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.000174183154241139, + "step": 5955 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2454, + "image_masks_loss": 0.0228, + "llm_learning_rate": 1.7410410378203807e-06, + "loss": 0.8018, + "output_text_mask_loss": 0.0018, + "predictor_learning_rate": 0.00017410410378203807, + "step": 5956 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2453, + "image_masks_loss": 0.0311, + "llm_learning_rate": 1.740250616807623e-06, + "loss": 0.8002, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001740250616807623, + "step": 5957 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2453, + "image_masks_loss": 0.0219, + "llm_learning_rate": 1.7394602794601598e-06, + "loss": 0.7472, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00017394602794601596, + "step": 5958 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2452, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.7386700258650247e-06, + "loss": 0.7789, + "output_text_mask_loss": 0.012, + "predictor_learning_rate": 0.00017386700258650246, + "step": 5959 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2452, + "image_masks_loss": 0.0203, + "llm_learning_rate": 1.7378798561092413e-06, + "loss": 0.79, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00017378798561092413, + "step": 5960 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2451, + "image_masks_loss": 0.0187, + "llm_learning_rate": 1.7370897702798265e-06, + "loss": 0.8109, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00017370897702798265, + "step": 5961 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.245, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.7362997684637853e-06, + "loss": 0.6993, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001736299768463785, + "step": 5962 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.245, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.7355098507481144e-06, + "loss": 0.7605, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.00017355098507481143, + "step": 5963 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2449, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.734720017219802e-06, + "loss": 0.7616, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.00017347200172198018, + "step": 5964 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2449, + "image_masks_loss": 0.016, + "llm_learning_rate": 1.7339302679658257e-06, + "loss": 0.7207, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00017339302679658257, + "step": 5965 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2448, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.7331406030731546e-06, + "loss": 0.84, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00017331406030731544, + "step": 5966 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2448, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.7323510226287482e-06, + "loss": 0.7478, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0001732351022628748, + "step": 5967 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2447, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.7315615267195573e-06, + "loss": 0.6786, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00017315615267195572, + "step": 5968 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2446, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.730772115432523e-06, + "loss": 0.6498, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017307721154325229, + "step": 5969 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2446, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.7299827888545762e-06, + "loss": 0.769, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00017299827888545761, + "step": 5970 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2445, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.7291935470726402e-06, + "loss": 0.786, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.000172919354707264, + "step": 5971 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2445, + "image_masks_loss": 0.0147, + "llm_learning_rate": 1.728404390173628e-06, + "loss": 0.7021, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0001728404390173628, + "step": 5972 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2444, + "image_masks_loss": 0.0114, + "llm_learning_rate": 1.727615318244443e-06, + "loss": 0.8076, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001727615318244443, + "step": 5973 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2443, + "image_masks_loss": 0.0167, + "llm_learning_rate": 1.7268263313719794e-06, + "loss": 0.7625, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017268263313719793, + "step": 5974 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2443, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.726037429643122e-06, + "loss": 0.8157, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00017260374296431218, + "step": 5975 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2442, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.7252486131447466e-06, + "loss": 0.7318, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017252486131447465, + "step": 5976 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2442, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.7244598819637204e-06, + "loss": 0.672, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017244598819637203, + "step": 5977 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2441, + "image_masks_loss": 0.0157, + "llm_learning_rate": 1.723671236186899e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0001723671236186899, + "step": 5978 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2441, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.7228826759011298e-06, + "loss": 0.8447, + "output_text_mask_loss": 0.0698, + "predictor_learning_rate": 0.00017228826759011295, + "step": 5979 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.244, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.7220942011932519e-06, + "loss": 0.72, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00017220942011932517, + "step": 5980 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2439, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.7213058121500934e-06, + "loss": 0.6916, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00017213058121500933, + "step": 5981 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2439, + "image_masks_loss": 0.0017, + "llm_learning_rate": 1.720517508858473e-06, + "loss": 0.841, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00017205175088584728, + "step": 5982 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2438, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.7197292914052005e-06, + "loss": 0.7622, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00017197292914052003, + "step": 5983 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2438, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.7189411598770773e-06, + "loss": 0.7596, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001718941159877077, + "step": 5984 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2437, + "image_masks_loss": 0.0261, + "llm_learning_rate": 1.7181531143608927e-06, + "loss": 0.6535, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 0.00017181531143608926, + "step": 5985 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2437, + "image_masks_loss": 0.0176, + "llm_learning_rate": 1.7173651549434283e-06, + "loss": 0.7412, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.00017173651549434282, + "step": 5986 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2436, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.7165772817114567e-06, + "loss": 0.6915, + "output_text_mask_loss": 0.0408, + "predictor_learning_rate": 0.00017165772817114567, + "step": 5987 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2435, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.71578949475174e-06, + "loss": 0.7709, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.000171578949475174, + "step": 5988 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2435, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.715001794151031e-06, + "loss": 0.775, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0001715001794151031, + "step": 5989 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2434, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.7142141799960732e-06, + "loss": 0.8693, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001714214179996073, + "step": 5990 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2434, + "image_masks_loss": 0.0187, + "llm_learning_rate": 1.7134266523735999e-06, + "loss": 0.7034, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00017134266523735996, + "step": 5991 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2433, + "image_masks_loss": 0.011, + "llm_learning_rate": 1.712639211370335e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0001712639211370335, + "step": 5992 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2433, + "image_masks_loss": 0.02, + "llm_learning_rate": 1.7118518570729944e-06, + "loss": 0.7776, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00017118518570729944, + "step": 5993 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2432, + "image_masks_loss": 0.017, + "llm_learning_rate": 1.711064589568283e-06, + "loss": 0.7524, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0001711064589568283, + "step": 5994 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2431, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.7102774089428959e-06, + "loss": 0.7132, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017102774089428958, + "step": 5995 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2431, + "image_masks_loss": 0.0248, + "llm_learning_rate": 1.7094903152835195e-06, + "loss": 0.7666, + "output_text_mask_loss": 0.012, + "predictor_learning_rate": 0.00017094903152835194, + "step": 5996 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.243, + "image_masks_loss": 0.02, + "llm_learning_rate": 1.7087033086768305e-06, + "loss": 0.7715, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00017087033086768305, + "step": 5997 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.243, + "image_masks_loss": 0.0184, + "llm_learning_rate": 1.7079163892094963e-06, + "loss": 0.6956, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00017079163892094962, + "step": 5998 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2429, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.7071295569681728e-06, + "loss": 0.8151, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00017071295569681729, + "step": 5999 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2429, + "image_masks_loss": 0.0116, + "llm_learning_rate": 1.7063428120395087e-06, + "loss": 0.8142, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.00017063428120395085, + "step": 6000 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2428, + "image_masks_loss": 0.0031, + "llm_learning_rate": 1.7055561545101423e-06, + "loss": 0.8089, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00017055561545101423, + "step": 6001 + }, + { + "epoch": 0.61, + "gumbel_tau": 0.2427, + "image_masks_loss": 0.0114, + "llm_learning_rate": 1.7047695844667011e-06, + "loss": 0.757, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001704769584466701, + "step": 6002 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2427, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.7039831019958053e-06, + "loss": 0.6956, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00017039831019958053, + "step": 6003 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2426, + "image_masks_loss": 0.0233, + "llm_learning_rate": 1.703196707184063e-06, + "loss": 0.8231, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.0001703196707184063, + "step": 6004 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2426, + "image_masks_loss": 0.0189, + "llm_learning_rate": 1.702410400118074e-06, + "loss": 0.8058, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001702410400118074, + "step": 6005 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2425, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.7016241808844284e-06, + "loss": 0.6151, + "output_text_mask_loss": 0.0125, + "predictor_learning_rate": 0.00017016241808844281, + "step": 6006 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2425, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.7008380495697063e-06, + "loss": 0.7242, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00017008380495697062, + "step": 6007 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2424, + "image_masks_loss": 0.0037, + "llm_learning_rate": 1.7000520062604785e-06, + "loss": 0.6729, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00017000520062604785, + "step": 6008 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2423, + "image_masks_loss": 0.0171, + "llm_learning_rate": 1.6992660510433062e-06, + "loss": 0.7857, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00016992660510433061, + "step": 6009 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2423, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.6984801840047394e-06, + "loss": 0.728, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0001698480184004739, + "step": 6010 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2422, + "image_masks_loss": 0.0227, + "llm_learning_rate": 1.6976944052313194e-06, + "loss": 0.6747, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00016976944052313192, + "step": 6011 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2422, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.6969087148095787e-06, + "loss": 0.7336, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016969087148095786, + "step": 6012 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2421, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.6961231128260398e-06, + "loss": 0.7817, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00016961231128260397, + "step": 6013 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2421, + "image_masks_loss": 0.0203, + "llm_learning_rate": 1.6953375993672138e-06, + "loss": 0.7109, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.00016953375993672138, + "step": 6014 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.242, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.6945521745196035e-06, + "loss": 0.6712, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00016945521745196032, + "step": 6015 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2419, + "image_masks_loss": 0.0213, + "llm_learning_rate": 1.6937668383697021e-06, + "loss": 0.6846, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.0001693766838369702, + "step": 6016 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2419, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.692981591003992e-06, + "loss": 0.7526, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001692981591003992, + "step": 6017 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2418, + "image_masks_loss": 0.0164, + "llm_learning_rate": 1.6921964325089468e-06, + "loss": 0.7061, + "output_text_mask_loss": 0.0713, + "predictor_learning_rate": 0.00016921964325089468, + "step": 6018 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2418, + "image_masks_loss": 0.0219, + "llm_learning_rate": 1.6914113629710298e-06, + "loss": 0.7582, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 0.00016914113629710298, + "step": 6019 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2417, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.690626382476695e-06, + "loss": 0.8017, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.0001690626382476695, + "step": 6020 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2417, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.6898414911123853e-06, + "loss": 0.7668, + "output_text_mask_loss": 0.0479, + "predictor_learning_rate": 0.00016898414911123853, + "step": 6021 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2416, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.689056688964535e-06, + "loss": 0.7282, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016890566889645348, + "step": 6022 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2415, + "image_masks_loss": 0.0186, + "llm_learning_rate": 1.6882719761195686e-06, + "loss": 0.7358, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00016882719761195686, + "step": 6023 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2415, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.6874873526639004e-06, + "loss": 0.7288, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 0.00016874873526639004, + "step": 6024 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2414, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.686702818683934e-06, + "loss": 0.7272, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.0001686702818683934, + "step": 6025 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2414, + "image_masks_loss": 0.0292, + "llm_learning_rate": 1.6859183742660657e-06, + "loss": 0.8256, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016859183742660655, + "step": 6026 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2413, + "image_masks_loss": 0.034, + "llm_learning_rate": 1.6851340194966798e-06, + "loss": 0.7194, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00016851340194966796, + "step": 6027 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2413, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.6843497544621498e-06, + "loss": 0.7961, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00016843497544621498, + "step": 6028 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2412, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.683565579248842e-06, + "loss": 0.8229, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0001683565579248842, + "step": 6029 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2411, + "image_masks_loss": 0.033, + "llm_learning_rate": 1.682781493943111e-06, + "loss": 0.7323, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016827814939431108, + "step": 6030 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2411, + "image_masks_loss": 0.0261, + "llm_learning_rate": 1.6819974986313015e-06, + "loss": 0.7771, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00016819974986313015, + "step": 6031 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.241, + "image_masks_loss": 0.0178, + "llm_learning_rate": 1.6812135933997504e-06, + "loss": 0.7599, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.000168121359339975, + "step": 6032 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.241, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.6804297783347818e-06, + "loss": 0.8165, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00016804297783347817, + "step": 6033 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2409, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.679646053522711e-06, + "loss": 0.6936, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.0001679646053522711, + "step": 6034 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2409, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.6788624190498448e-06, + "loss": 0.7267, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016788624190498447, + "step": 6035 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2408, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.678078875002478e-06, + "loss": 0.7199, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001678078875002478, + "step": 6036 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2407, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.6772954214668962e-06, + "loss": 0.7195, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.0001677295421466896, + "step": 6037 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2407, + "image_masks_loss": 0.011, + "llm_learning_rate": 1.6765120585293745e-06, + "loss": 0.7642, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00016765120585293745, + "step": 6038 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2406, + "image_masks_loss": 0.0173, + "llm_learning_rate": 1.6757287862761799e-06, + "loss": 0.8807, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.00016757287862761798, + "step": 6039 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2406, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.6749456047935668e-06, + "loss": 0.6842, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00016749456047935668, + "step": 6040 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2405, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.6741625141677815e-06, + "loss": 0.7115, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016741625141677814, + "step": 6041 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2405, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.6733795144850601e-06, + "loss": 0.7197, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000167337951448506, + "step": 6042 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2404, + "image_masks_loss": 0.0136, + "llm_learning_rate": 1.6725966058316275e-06, + "loss": 0.8428, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.00016725966058316275, + "step": 6043 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2403, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.6718137882936996e-06, + "loss": 0.6374, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00016718137882936994, + "step": 6044 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2403, + "image_masks_loss": 0.0248, + "llm_learning_rate": 1.6710310619574821e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0001671031061957482, + "step": 6045 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2402, + "image_masks_loss": 0.0232, + "llm_learning_rate": 1.6702484269091715e-06, + "loss": 0.792, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.00016702484269091715, + "step": 6046 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2402, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.669465883234951e-06, + "loss": 0.814, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00016694658832349507, + "step": 6047 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2401, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.6686834310209977e-06, + "loss": 0.7562, + "output_text_mask_loss": 0.0166, + "predictor_learning_rate": 0.00016686834310209974, + "step": 6048 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2401, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.667901070353477e-06, + "loss": 0.7272, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 0.0001667901070353477, + "step": 6049 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.24, + "image_masks_loss": 0.0353, + "llm_learning_rate": 1.667118801318543e-06, + "loss": 0.6937, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00016671188013185428, + "step": 6050 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.24, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.6663366240023424e-06, + "loss": 0.7602, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00016663366240023423, + "step": 6051 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2399, + "image_masks_loss": 0.0125, + "llm_learning_rate": 1.66555453849101e-06, + "loss": 0.6862, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00016655545384910097, + "step": 6052 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2398, + "image_masks_loss": 0.0232, + "llm_learning_rate": 1.6647725448706697e-06, + "loss": 0.8485, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.00016647725448706697, + "step": 6053 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2398, + "image_masks_loss": 0.0136, + "llm_learning_rate": 1.6639906432274372e-06, + "loss": 0.7338, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 0.0001663990643227437, + "step": 6054 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2397, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.6632088336474178e-06, + "loss": 0.7914, + "output_text_mask_loss": 0.0717, + "predictor_learning_rate": 0.00016632088336474178, + "step": 6055 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2397, + "image_masks_loss": 0.0047, + "llm_learning_rate": 1.6624271162167054e-06, + "loss": 0.5777, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00016624271162167054, + "step": 6056 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2396, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.6616454910213838e-06, + "loss": 0.78, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.00016616454910213838, + "step": 6057 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2396, + "image_masks_loss": 0.028, + "llm_learning_rate": 1.660863958147529e-06, + "loss": 0.6794, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016608639581475288, + "step": 6058 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2395, + "image_masks_loss": 0.0147, + "llm_learning_rate": 1.660082517681204e-06, + "loss": 0.7261, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001660082517681204, + "step": 6059 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2394, + "image_masks_loss": 0.0201, + "llm_learning_rate": 1.6593011697084633e-06, + "loss": 0.7874, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00016593011697084632, + "step": 6060 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2394, + "image_masks_loss": 0.0214, + "llm_learning_rate": 1.6585199143153494e-06, + "loss": 0.7501, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.0001658519914315349, + "step": 6061 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2393, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.6577387515878975e-06, + "loss": 0.7122, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00016577387515878976, + "step": 6062 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2393, + "image_masks_loss": 0.0168, + "llm_learning_rate": 1.6569576816121305e-06, + "loss": 0.7151, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00016569576816121304, + "step": 6063 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2392, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.6561767044740618e-06, + "loss": 0.6979, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00016561767044740615, + "step": 6064 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2392, + "image_masks_loss": 0.0232, + "llm_learning_rate": 1.6553958202596937e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00016553958202596936, + "step": 6065 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2391, + "image_masks_loss": 0.0115, + "llm_learning_rate": 1.6546150290550184e-06, + "loss": 0.7506, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00016546150290550182, + "step": 6066 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.239, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.6538343309460195e-06, + "loss": 0.8026, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 0.00016538343309460194, + "step": 6067 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.239, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.653053726018669e-06, + "loss": 0.7215, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.0001653053726018669, + "step": 6068 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2389, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.6522732143589288e-06, + "loss": 0.7652, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00016522732143589287, + "step": 6069 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2389, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.65149279605275e-06, + "loss": 0.7409, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00016514927960527498, + "step": 6070 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2388, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.6507124711860745e-06, + "loss": 0.8215, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00016507124711860743, + "step": 6071 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2388, + "image_masks_loss": 0.0254, + "llm_learning_rate": 1.6499322398448337e-06, + "loss": 0.759, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00016499322398448336, + "step": 6072 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2387, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.6491521021149476e-06, + "loss": 0.7336, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00016491521021149474, + "step": 6073 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2387, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.6483720580823276e-06, + "loss": 0.8095, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00016483720580823274, + "step": 6074 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2386, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.647592107832874e-06, + "loss": 0.766, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.00016475921078328736, + "step": 6075 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2385, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.6468122514524754e-06, + "loss": 0.7142, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.00016468122514524752, + "step": 6076 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2385, + "image_masks_loss": 0.0168, + "llm_learning_rate": 1.646032489027012e-06, + "loss": 0.6089, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001646032489027012, + "step": 6077 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2384, + "image_masks_loss": 0.0182, + "llm_learning_rate": 1.6452528206423535e-06, + "loss": 0.7899, + "output_text_mask_loss": 0.0656, + "predictor_learning_rate": 0.00016452528206423533, + "step": 6078 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2384, + "image_masks_loss": 0.0047, + "llm_learning_rate": 1.6444732463843582e-06, + "loss": 0.7019, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001644473246384358, + "step": 6079 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2383, + "image_masks_loss": 0.0165, + "llm_learning_rate": 1.643693766338874e-06, + "loss": 0.7131, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001643693766338874, + "step": 6080 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2383, + "image_masks_loss": 0.0192, + "llm_learning_rate": 1.6429143805917404e-06, + "loss": 0.7089, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00016429143805917403, + "step": 6081 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2382, + "image_masks_loss": 0.0204, + "llm_learning_rate": 1.642135089228784e-06, + "loss": 0.7364, + "output_text_mask_loss": 0.0196, + "predictor_learning_rate": 0.0001642135089228784, + "step": 6082 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2381, + "image_masks_loss": 0.0032, + "llm_learning_rate": 1.6413558923358233e-06, + "loss": 0.7309, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016413558923358232, + "step": 6083 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2381, + "image_masks_loss": 0.0298, + "llm_learning_rate": 1.6405767899986635e-06, + "loss": 0.7578, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00016405767899986635, + "step": 6084 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.238, + "image_masks_loss": 0.0114, + "llm_learning_rate": 1.639797782303102e-06, + "loss": 0.7808, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0001639797782303102, + "step": 6085 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.238, + "image_masks_loss": 0.0265, + "llm_learning_rate": 1.639018869334924e-06, + "loss": 0.7972, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001639018869334924, + "step": 6086 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2379, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.6382400511799068e-06, + "loss": 0.7457, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00016382400511799066, + "step": 6087 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2379, + "image_masks_loss": 0.0269, + "llm_learning_rate": 1.6374613279238144e-06, + "loss": 0.8003, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00016374613279238143, + "step": 6088 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2378, + "image_masks_loss": 0.0187, + "llm_learning_rate": 1.636682699652401e-06, + "loss": 0.7547, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00016366826996524008, + "step": 6089 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2378, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.6359041664514124e-06, + "loss": 0.6397, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016359041664514124, + "step": 6090 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2377, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.635125728406582e-06, + "loss": 0.8164, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.00016351257284065817, + "step": 6091 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2376, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.6343473856036323e-06, + "loss": 0.7378, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.00016343473856036324, + "step": 6092 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2376, + "image_masks_loss": 0.0187, + "llm_learning_rate": 1.6335691381282763e-06, + "loss": 0.7616, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.00016335691381282763, + "step": 6093 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2375, + "image_masks_loss": 0.0216, + "llm_learning_rate": 1.6327909860662178e-06, + "loss": 0.7439, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00016327909860662176, + "step": 6094 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2375, + "image_masks_loss": 0.022, + "llm_learning_rate": 1.6320129295031467e-06, + "loss": 0.8252, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 0.00016320129295031467, + "step": 6095 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2374, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.6312349685247453e-06, + "loss": 0.775, + "output_text_mask_loss": 0.0336, + "predictor_learning_rate": 0.00016312349685247452, + "step": 6096 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2374, + "image_masks_loss": 0.0185, + "llm_learning_rate": 1.6304571032166844e-06, + "loss": 0.7637, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016304571032166844, + "step": 6097 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2373, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.6296793336646243e-06, + "loss": 0.6387, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016296793336646243, + "step": 6098 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2372, + "image_masks_loss": 0.0045, + "llm_learning_rate": 1.6289016599542146e-06, + "loss": 0.7342, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00016289016599542144, + "step": 6099 + }, + { + "epoch": 0.62, + "gumbel_tau": 0.2372, + "image_masks_loss": 0.002, + "llm_learning_rate": 1.6281240821710937e-06, + "loss": 0.6888, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.00016281240821710937, + "step": 6100 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2371, + "image_masks_loss": 0.0234, + "llm_learning_rate": 1.6273466004008915e-06, + "loss": 0.6463, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00016273466004008914, + "step": 6101 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2371, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.6265692147292259e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00016265692147292258, + "step": 6102 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.237, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.6257919252417032e-06, + "loss": 0.705, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.00016257919252417032, + "step": 6103 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.237, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.6250147320239212e-06, + "loss": 0.7283, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.0001625014732023921, + "step": 6104 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2369, + "image_masks_loss": 0.0178, + "llm_learning_rate": 1.6242376351614653e-06, + "loss": 0.7532, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00016242376351614652, + "step": 6105 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2369, + "image_masks_loss": 0.0217, + "llm_learning_rate": 1.623460634739912e-06, + "loss": 0.7809, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.0001623460634739912, + "step": 6106 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2368, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.6226837308448262e-06, + "loss": 0.6372, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0001622683730844826, + "step": 6107 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2367, + "image_masks_loss": 0.0151, + "llm_learning_rate": 1.621906923561762e-06, + "loss": 0.7457, + "output_text_mask_loss": 0.0328, + "predictor_learning_rate": 0.0001621906923561762, + "step": 6108 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2367, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.621130212976263e-06, + "loss": 0.6843, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 0.00016211302129762629, + "step": 6109 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2366, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.6203535991738627e-06, + "loss": 0.7164, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.00016203535991738628, + "step": 6110 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2366, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.6195770822400837e-06, + "loss": 0.7016, + "output_text_mask_loss": 0.0546, + "predictor_learning_rate": 0.00016195770822400834, + "step": 6111 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2365, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.618800662260437e-06, + "loss": 0.7627, + "output_text_mask_loss": 0.0717, + "predictor_learning_rate": 0.00016188006622604368, + "step": 6112 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2365, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.6180243393204242e-06, + "loss": 0.6283, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0001618024339320424, + "step": 6113 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2364, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.6172481135055363e-06, + "loss": 0.7112, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.0001617248113505536, + "step": 6114 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2364, + "image_masks_loss": 0.0095, + "llm_learning_rate": 1.6164719849012517e-06, + "loss": 0.7982, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00016164719849012517, + "step": 6115 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2363, + "image_masks_loss": 0.016, + "llm_learning_rate": 1.61569595359304e-06, + "loss": 0.7613, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00016156959535930398, + "step": 6116 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2362, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.61492001966636e-06, + "loss": 0.7051, + "output_text_mask_loss": 0.0839, + "predictor_learning_rate": 0.000161492001966636, + "step": 6117 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2362, + "image_masks_loss": 0.0127, + "llm_learning_rate": 1.6141441832066589e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0001614144183206659, + "step": 6118 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2361, + "image_masks_loss": 0.003, + "llm_learning_rate": 1.6133684442993734e-06, + "loss": 0.784, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.0001613368444299373, + "step": 6119 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2361, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.6125928030299308e-06, + "loss": 0.7434, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016125928030299308, + "step": 6120 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.236, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.611817259483744e-06, + "loss": 0.7388, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00016118172594837437, + "step": 6121 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.236, + "image_masks_loss": 0.0263, + "llm_learning_rate": 1.6110418137462192e-06, + "loss": 0.7801, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0001611041813746219, + "step": 6122 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2359, + "image_masks_loss": 0.0298, + "llm_learning_rate": 1.61026646590275e-06, + "loss": 0.7308, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.000161026646590275, + "step": 6123 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2359, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.6094912160387197e-06, + "loss": 0.7844, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 0.00016094912160387196, + "step": 6124 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2358, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.6087160642394994e-06, + "loss": 0.6783, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00016087160642394994, + "step": 6125 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2357, + "image_masks_loss": 0.0203, + "llm_learning_rate": 1.6079410105904518e-06, + "loss": 0.8296, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00016079410105904518, + "step": 6126 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2357, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.6071660551769274e-06, + "loss": 0.7212, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.00016071660551769275, + "step": 6127 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2356, + "image_masks_loss": 0.0176, + "llm_learning_rate": 1.606391198084265e-06, + "loss": 0.6787, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 0.0001606391198084265, + "step": 6128 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2356, + "image_masks_loss": 0.0189, + "llm_learning_rate": 1.6056164393977952e-06, + "loss": 0.7415, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 0.00016056164393977952, + "step": 6129 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2355, + "image_masks_loss": 0.0036, + "llm_learning_rate": 1.6048417792028349e-06, + "loss": 0.8451, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 0.00016048417792028348, + "step": 6130 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2355, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.6040672175846916e-06, + "loss": 0.7814, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.00016040672175846916, + "step": 6131 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2354, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.6032927546286616e-06, + "loss": 0.7082, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.00016032927546286616, + "step": 6132 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2354, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.6025183904200315e-06, + "loss": 0.6801, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016025183904200315, + "step": 6133 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2353, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.601744125044075e-06, + "loss": 0.8298, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00016017441250440752, + "step": 6134 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2352, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.6009699585860558e-06, + "loss": 0.7919, + "output_text_mask_loss": 0.0633, + "predictor_learning_rate": 0.00016009699585860557, + "step": 6135 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2352, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.6001958911312278e-06, + "loss": 0.711, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00016001958911312277, + "step": 6136 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2351, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.5994219227648328e-06, + "loss": 0.7643, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00015994219227648328, + "step": 6137 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2351, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.598648053572101e-06, + "loss": 0.679, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001598648053572101, + "step": 6138 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.235, + "image_masks_loss": 0.0025, + "llm_learning_rate": 1.5978742836382543e-06, + "loss": 0.809, + "output_text_mask_loss": 0.1175, + "predictor_learning_rate": 0.00015978742836382542, + "step": 6139 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.235, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.5971006130485e-06, + "loss": 0.7176, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015971006130485, + "step": 6140 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2349, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.5963270418880372e-06, + "loss": 0.657, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015963270418880372, + "step": 6141 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2349, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.5955535702420536e-06, + "loss": 0.7947, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00015955535702420534, + "step": 6142 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2348, + "image_masks_loss": 0.0035, + "llm_learning_rate": 1.5947801981957255e-06, + "loss": 0.7275, + "output_text_mask_loss": 0.0584, + "predictor_learning_rate": 0.00015947801981957256, + "step": 6143 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2347, + "image_masks_loss": 0.0244, + "llm_learning_rate": 1.5940069258342177e-06, + "loss": 0.7411, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00015940069258342176, + "step": 6144 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2347, + "image_masks_loss": 0.0204, + "llm_learning_rate": 1.593233753242686e-06, + "loss": 0.716, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00015932337532426858, + "step": 6145 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2346, + "image_masks_loss": 0.02, + "llm_learning_rate": 1.5924606805062727e-06, + "loss": 0.8063, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.00015924606805062727, + "step": 6146 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2346, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.5916877077101113e-06, + "loss": 0.7333, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0001591687707710111, + "step": 6147 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2345, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.5909148349393215e-06, + "loss": 0.8012, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00015909148349393216, + "step": 6148 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2345, + "image_masks_loss": 0.028, + "llm_learning_rate": 1.5901420622790162e-06, + "loss": 0.6986, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001590142062279016, + "step": 6149 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2344, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.5893693898142932e-06, + "loss": 0.6906, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001589369389814293, + "step": 6150 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2344, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.5885968176302411e-06, + "loss": 0.6931, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001588596817630241, + "step": 6151 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2343, + "image_masks_loss": 0.0035, + "llm_learning_rate": 1.587824345811938e-06, + "loss": 0.7632, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015878243458119378, + "step": 6152 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2342, + "image_masks_loss": 0.0196, + "llm_learning_rate": 1.5870519744444495e-06, + "loss": 0.7295, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.00015870519744444494, + "step": 6153 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2342, + "image_masks_loss": 0.0178, + "llm_learning_rate": 1.586279703612831e-06, + "loss": 0.6888, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0001586279703612831, + "step": 6154 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2341, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.5855075334021266e-06, + "loss": 0.746, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00015855075334021264, + "step": 6155 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2341, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.58473546389737e-06, + "loss": 0.813, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.000158473546389737, + "step": 6156 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.234, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.583963495183583e-06, + "loss": 0.7064, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00015839634951835829, + "step": 6157 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.234, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.5831916273457762e-06, + "loss": 0.7534, + "output_text_mask_loss": 0.0607, + "predictor_learning_rate": 0.0001583191627345776, + "step": 6158 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2339, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.5824198604689495e-06, + "loss": 0.7486, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.00015824198604689492, + "step": 6159 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2339, + "image_masks_loss": 0.0206, + "llm_learning_rate": 1.5816481946380908e-06, + "loss": 0.844, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00015816481946380906, + "step": 6160 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2338, + "image_masks_loss": 0.0037, + "llm_learning_rate": 1.580876629938179e-06, + "loss": 0.6809, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0001580876629938179, + "step": 6161 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2337, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.58010516645418e-06, + "loss": 0.7847, + "output_text_mask_loss": 0.1106, + "predictor_learning_rate": 0.000158010516645418, + "step": 6162 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2337, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.5793338042710494e-06, + "loss": 0.7449, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00015793338042710493, + "step": 6163 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2336, + "image_masks_loss": 0.0047, + "llm_learning_rate": 1.57856254347373e-06, + "loss": 0.6319, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.000157856254347373, + "step": 6164 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2336, + "image_masks_loss": 0.0022, + "llm_learning_rate": 1.5777913841471566e-06, + "loss": 0.7359, + "output_text_mask_loss": 0.0018, + "predictor_learning_rate": 0.00015777913841471565, + "step": 6165 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2335, + "image_masks_loss": 0.0202, + "llm_learning_rate": 1.5770203263762499e-06, + "loss": 0.7452, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 0.00015770203263762499, + "step": 6166 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2335, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.5762493702459203e-06, + "loss": 0.7293, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.000157624937024592, + "step": 6167 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2334, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.575478515841068e-06, + "loss": 0.7581, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.0001575478515841068, + "step": 6168 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2334, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.5747077632465806e-06, + "loss": 0.8151, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015747077632465806, + "step": 6169 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2333, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.5739371125473358e-06, + "loss": 0.8298, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00015739371125473355, + "step": 6170 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2333, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.5731665638281985e-06, + "loss": 0.7877, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 0.00015731665638281982, + "step": 6171 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2332, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.5723961171740237e-06, + "loss": 0.7609, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015723961171740236, + "step": 6172 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2331, + "image_masks_loss": 0.0147, + "llm_learning_rate": 1.571625772669655e-06, + "loss": 0.6766, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00015716257726696548, + "step": 6173 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2331, + "image_masks_loss": 0.0225, + "llm_learning_rate": 1.5708555303999231e-06, + "loss": 0.665, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.0001570855530399923, + "step": 6174 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.233, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.5700853904496506e-06, + "loss": 0.7727, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.00015700853904496503, + "step": 6175 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.233, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.5693153529036467e-06, + "loss": 0.6724, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00015693153529036465, + "step": 6176 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2329, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.5685454178467085e-06, + "loss": 0.7626, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00015685454178467084, + "step": 6177 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2329, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.567775585363624e-06, + "loss": 0.779, + "output_text_mask_loss": 0.0458, + "predictor_learning_rate": 0.00015677755853636238, + "step": 6178 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2328, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.5670058555391682e-06, + "loss": 0.7949, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 0.0001567005855539168, + "step": 6179 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2328, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.5662362284581051e-06, + "loss": 0.7882, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00015662362284581048, + "step": 6180 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2327, + "image_masks_loss": 0.0155, + "llm_learning_rate": 1.565466704205189e-06, + "loss": 0.6871, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001565466704205189, + "step": 6181 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2326, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.5646972828651613e-06, + "loss": 0.7526, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015646972828651612, + "step": 6182 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2326, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.5639279645227517e-06, + "loss": 0.6608, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00015639279645227516, + "step": 6183 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2325, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.5631587492626796e-06, + "loss": 0.7357, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.00015631587492626796, + "step": 6184 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2325, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.562389637169653e-06, + "loss": 0.7438, + "output_text_mask_loss": 0.0064, + "predictor_learning_rate": 0.00015623896371696528, + "step": 6185 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2324, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.561620628328368e-06, + "loss": 0.7336, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00015616206283283678, + "step": 6186 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2324, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.5608517228235093e-06, + "loss": 0.7371, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001560851722823509, + "step": 6187 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2323, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.560082920739751e-06, + "loss": 0.8015, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 0.00015600829207397507, + "step": 6188 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2323, + "image_masks_loss": 0.0292, + "llm_learning_rate": 1.5593142221617553e-06, + "loss": 0.8048, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.0001559314222161755, + "step": 6189 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2322, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.5585456271741722e-06, + "loss": 0.6454, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001558545627174172, + "step": 6190 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2322, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.5577771358616416e-06, + "loss": 0.7368, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00015577771358616415, + "step": 6191 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2321, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.557008748308792e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 0.0001557008748308792, + "step": 6192 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.232, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.5562404646002394e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.00015562404646002394, + "step": 6193 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.232, + "image_masks_loss": 0.0168, + "llm_learning_rate": 1.5554722848205894e-06, + "loss": 0.6917, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00015554722848205892, + "step": 6194 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2319, + "image_masks_loss": 0.0172, + "llm_learning_rate": 1.5547042090544348e-06, + "loss": 0.7173, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.00015547042090544347, + "step": 6195 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2319, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.5539362373863579e-06, + "loss": 0.7764, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00015539362373863578, + "step": 6196 + }, + { + "epoch": 0.63, + "gumbel_tau": 0.2318, + "image_masks_loss": 0.0145, + "llm_learning_rate": 1.55316836990093e-06, + "loss": 0.7945, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 0.000155316836990093, + "step": 6197 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2318, + "image_masks_loss": 0.0066, + "llm_learning_rate": 1.5524006066827105e-06, + "loss": 0.7138, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00015524006066827104, + "step": 6198 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2317, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.5516329478162468e-06, + "loss": 0.7378, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00015516329478162467, + "step": 6199 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2317, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.5508653933860748e-06, + "loss": 0.8142, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00015508653933860747, + "step": 6200 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2316, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.5500979434767201e-06, + "loss": 0.7344, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.000155009794347672, + "step": 6201 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2316, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.5493305981726958e-06, + "loss": 0.74, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00015493305981726958, + "step": 6202 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2315, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.5485633575585035e-06, + "loss": 0.7424, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00015485633575585035, + "step": 6203 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2314, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.5477962217186335e-06, + "loss": 0.7933, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00015477962217186334, + "step": 6204 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2314, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.5470291907375645e-06, + "loss": 0.7551, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00015470291907375644, + "step": 6205 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2313, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.5462622646997638e-06, + "loss": 0.7564, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00015462622646997636, + "step": 6206 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2313, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.5454954436896864e-06, + "loss": 0.7982, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00015454954436896863, + "step": 6207 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2312, + "image_masks_loss": 0.0089, + "llm_learning_rate": 1.5447287277917775e-06, + "loss": 0.7864, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 0.00015447287277917774, + "step": 6208 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2312, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.5439621170904687e-06, + "loss": 0.7573, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00015439621170904686, + "step": 6209 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2311, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.5431956116701805e-06, + "loss": 0.7464, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.00015431956116701803, + "step": 6210 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2311, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.5424292116153233e-06, + "loss": 0.7596, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00015424292116153232, + "step": 6211 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.231, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.5416629170102943e-06, + "loss": 0.8197, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00015416629170102942, + "step": 6212 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.231, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.5408967279394798e-06, + "loss": 0.8285, + "output_text_mask_loss": 0.0311, + "predictor_learning_rate": 0.000154089672793948, + "step": 6213 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2309, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.5401306444872534e-06, + "loss": 0.7199, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 0.00015401306444872535, + "step": 6214 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2308, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.5393646667379785e-06, + "loss": 0.7928, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00015393646667379786, + "step": 6215 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2308, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.538598794776006e-06, + "loss": 0.753, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.00015385987947760061, + "step": 6216 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2307, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.537833028685676e-06, + "loss": 0.7347, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001537833028685676, + "step": 6217 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2307, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.537067368551316e-06, + "loss": 0.7294, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00015370673685513158, + "step": 6218 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2306, + "image_masks_loss": 0.0176, + "llm_learning_rate": 1.5363018144572418e-06, + "loss": 0.7829, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00015363018144572417, + "step": 6219 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2306, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.5355363664877588e-06, + "loss": 0.7559, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015355363664877586, + "step": 6220 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2305, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.5347710247271592e-06, + "loss": 0.7837, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001534771024727159, + "step": 6221 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2305, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.5340057892597246e-06, + "loss": 0.6879, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00015340057892597244, + "step": 6222 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2304, + "image_masks_loss": 0.0011, + "llm_learning_rate": 1.5332406601697237e-06, + "loss": 0.771, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00015332406601697236, + "step": 6223 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2304, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.5324756375414145e-06, + "loss": 0.7389, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.00015324756375414145, + "step": 6224 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2303, + "image_masks_loss": 0.0095, + "llm_learning_rate": 1.5317107214590437e-06, + "loss": 0.7718, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.00015317107214590436, + "step": 6225 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2302, + "image_masks_loss": 0.0025, + "llm_learning_rate": 1.5309459120068448e-06, + "loss": 0.6998, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.00015309459120068449, + "step": 6226 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2302, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.5301812092690406e-06, + "loss": 0.7733, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00015301812092690406, + "step": 6227 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2301, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.5294166133298422e-06, + "loss": 0.6844, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 0.0001529416613329842, + "step": 6228 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2301, + "image_masks_loss": 0.0162, + "llm_learning_rate": 1.5286521242734481e-06, + "loss": 0.7501, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.0001528652124273448, + "step": 6229 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.23, + "image_masks_loss": 0.0161, + "llm_learning_rate": 1.5278877421840448e-06, + "loss": 0.8111, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015278877421840448, + "step": 6230 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.23, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.5271234671458106e-06, + "loss": 0.736, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015271234671458106, + "step": 6231 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2299, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.5263592992429055e-06, + "loss": 0.7626, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00015263592992429054, + "step": 6232 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2299, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.5255952385594836e-06, + "loss": 0.75, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 0.00015255952385594834, + "step": 6233 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2298, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.5248312851796844e-06, + "loss": 0.7454, + "output_text_mask_loss": 0.0362, + "predictor_learning_rate": 0.00015248312851796843, + "step": 6234 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2298, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.5240674391876357e-06, + "loss": 0.752, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.00015240674391876356, + "step": 6235 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2297, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.5233037006674547e-06, + "loss": 0.7497, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00015233037006674548, + "step": 6236 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2296, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.5225400697032462e-06, + "loss": 0.5912, + "output_text_mask_loss": 0.0033, + "predictor_learning_rate": 0.0001522540069703246, + "step": 6237 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2296, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.5217765463791017e-06, + "loss": 0.8032, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.00015217765463791018, + "step": 6238 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2295, + "image_masks_loss": 0.0038, + "llm_learning_rate": 1.521013130779103e-06, + "loss": 0.6764, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.00015210131307791028, + "step": 6239 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2295, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.5202498229873191e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00015202498229873192, + "step": 6240 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2294, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.519486623087807e-06, + "loss": 0.7418, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015194866230878068, + "step": 6241 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2294, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.5187235311646114e-06, + "loss": 0.805, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00015187235311646115, + "step": 6242 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2293, + "image_masks_loss": 0.0144, + "llm_learning_rate": 1.517960547301767e-06, + "loss": 0.764, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.00015179605473017668, + "step": 6243 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2293, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.5171976715832946e-06, + "loss": 0.7805, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00015171976715832943, + "step": 6244 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2292, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.5164349040932037e-06, + "loss": 0.6916, + "output_text_mask_loss": 0.0119, + "predictor_learning_rate": 0.00015164349040932036, + "step": 6245 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2292, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.5156722449154908e-06, + "loss": 0.7125, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0001515672244915491, + "step": 6246 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2291, + "image_masks_loss": 0.0037, + "llm_learning_rate": 1.514909694134144e-06, + "loss": 0.7313, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.0001514909694134144, + "step": 6247 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2291, + "image_masks_loss": 0.0186, + "llm_learning_rate": 1.5141472518331357e-06, + "loss": 0.8661, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00015141472518331357, + "step": 6248 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.229, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.5133849180964274e-06, + "loss": 0.7314, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00015133849180964273, + "step": 6249 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2289, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.5126226930079712e-06, + "loss": 0.7633, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015126226930079712, + "step": 6250 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2289, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.5118605766517014e-06, + "loss": 0.7735, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00015118605766517012, + "step": 6251 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2288, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.5110985691115468e-06, + "loss": 0.772, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 0.00015110985691115466, + "step": 6252 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2288, + "image_masks_loss": 0.0139, + "llm_learning_rate": 1.5103366704714204e-06, + "loss": 0.7989, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 0.00015103366704714204, + "step": 6253 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2287, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.5095748808152242e-06, + "loss": 0.7517, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0001509574880815224, + "step": 6254 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2287, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.5088132002268476e-06, + "loss": 0.7284, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015088132002268477, + "step": 6255 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2286, + "image_masks_loss": 0.0164, + "llm_learning_rate": 1.5080516287901698e-06, + "loss": 0.6577, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00015080516287901697, + "step": 6256 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2286, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.507290166589056e-06, + "loss": 0.6912, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0001507290166589056, + "step": 6257 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2285, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.5065288137073597e-06, + "loss": 0.7168, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00015065288137073595, + "step": 6258 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2285, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.5057675702289244e-06, + "loss": 0.7251, + "output_text_mask_loss": 0.1465, + "predictor_learning_rate": 0.00015057675702289243, + "step": 6259 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2284, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.505006436237578e-06, + "loss": 0.6911, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.00015050064362375778, + "step": 6260 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2284, + "image_masks_loss": 0.0095, + "llm_learning_rate": 1.5042454118171395e-06, + "loss": 0.7434, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 0.00015042454118171394, + "step": 6261 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2283, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.5034844970514135e-06, + "loss": 0.7419, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00015034844970514133, + "step": 6262 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2282, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.502723692024195e-06, + "loss": 0.6661, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 0.0001502723692024195, + "step": 6263 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2282, + "image_masks_loss": 0.0217, + "llm_learning_rate": 1.501962996819265e-06, + "loss": 0.8018, + "output_text_mask_loss": 0.0423, + "predictor_learning_rate": 0.0001501962996819265, + "step": 6264 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2281, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.5012024115203921e-06, + "loss": 0.6969, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.0001501202411520392, + "step": 6265 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2281, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.5004419362113349e-06, + "loss": 0.7485, + "output_text_mask_loss": 0.0135, + "predictor_learning_rate": 0.00015004419362113348, + "step": 6266 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.228, + "image_masks_loss": 0.0164, + "llm_learning_rate": 1.4996815709758384e-06, + "loss": 0.8665, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00014996815709758383, + "step": 6267 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.228, + "image_masks_loss": 0.0136, + "llm_learning_rate": 1.4989213158976356e-06, + "loss": 0.6856, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00014989213158976355, + "step": 6268 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2279, + "image_masks_loss": 0.0218, + "llm_learning_rate": 1.4981611710604472e-06, + "loss": 0.7525, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.0001498161171060447, + "step": 6269 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2279, + "image_masks_loss": 0.0172, + "llm_learning_rate": 1.4974011365479818e-06, + "loss": 0.7623, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00014974011365479818, + "step": 6270 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2278, + "image_masks_loss": 0.0161, + "llm_learning_rate": 1.4966412124439359e-06, + "loss": 0.7279, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.00014966412124439358, + "step": 6271 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2278, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.4958813988319954e-06, + "loss": 0.7827, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00014958813988319954, + "step": 6272 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2277, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.4951216957958314e-06, + "loss": 0.771, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.00014951216957958312, + "step": 6273 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2277, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.4943621034191042e-06, + "loss": 0.7137, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00014943621034191042, + "step": 6274 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2276, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.4936026217854626e-06, + "loss": 0.7312, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00014936026217854625, + "step": 6275 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2275, + "image_masks_loss": 0.0031, + "llm_learning_rate": 1.4928432509785423e-06, + "loss": 0.7895, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00014928432509785422, + "step": 6276 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2275, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.4920839910819662e-06, + "loss": 0.8529, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.00014920839910819662, + "step": 6277 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2274, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.4913248421793458e-06, + "loss": 0.7409, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00014913248421793458, + "step": 6278 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2274, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.4905658043542809e-06, + "loss": 0.8394, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00014905658043542808, + "step": 6279 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2273, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.4898068776903579e-06, + "loss": 0.7196, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001489806877690358, + "step": 6280 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2273, + "image_masks_loss": 0.0305, + "llm_learning_rate": 1.4890480622711515e-06, + "loss": 0.7558, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00014890480622711515, + "step": 6281 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2272, + "image_masks_loss": 0.0216, + "llm_learning_rate": 1.4882893581802244e-06, + "loss": 0.7697, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00014882893581802243, + "step": 6282 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2272, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.487530765501127e-06, + "loss": 0.7188, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001487530765501127, + "step": 6283 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2271, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.4867722843173965e-06, + "loss": 0.7988, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00014867722843173964, + "step": 6284 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2271, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.4860139147125592e-06, + "loss": 0.7252, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001486013914712559, + "step": 6285 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.227, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.485255656770128e-06, + "loss": 0.7955, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0001485255656770128, + "step": 6286 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.227, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.4844975105736048e-06, + "loss": 0.762, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 0.00014844975105736047, + "step": 6287 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2269, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.4837394762064771e-06, + "loss": 0.7509, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.0001483739476206477, + "step": 6288 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2268, + "image_masks_loss": 0.0201, + "llm_learning_rate": 1.482981553752222e-06, + "loss": 0.7711, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 0.00014829815537522218, + "step": 6289 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2268, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.4822237432943029e-06, + "loss": 0.775, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0001482223743294303, + "step": 6290 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2267, + "image_masks_loss": 0.0195, + "llm_learning_rate": 1.4814660449161728e-06, + "loss": 0.7573, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00014814660449161726, + "step": 6291 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2267, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.4807084587012705e-06, + "loss": 0.7798, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.00014807084587012704, + "step": 6292 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2266, + "image_masks_loss": 0.0038, + "llm_learning_rate": 1.4799509847330228e-06, + "loss": 0.7301, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0001479950984733023, + "step": 6293 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2266, + "image_masks_loss": 0.0034, + "llm_learning_rate": 1.479193623094845e-06, + "loss": 0.7124, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.0001479193623094845, + "step": 6294 + }, + { + "epoch": 0.64, + "gumbel_tau": 0.2265, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.478436373870139e-06, + "loss": 0.7304, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001478436373870139, + "step": 6295 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2265, + "image_masks_loss": 0.0205, + "llm_learning_rate": 1.4776792371422954e-06, + "loss": 0.7566, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00014776792371422955, + "step": 6296 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2264, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.4769222129946906e-06, + "loss": 0.7318, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00014769222129946906, + "step": 6297 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2264, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.4761653015106907e-06, + "loss": 0.8585, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00014761653015106909, + "step": 6298 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2263, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.475408502773649e-06, + "loss": 0.7314, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0001475408502773649, + "step": 6299 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2263, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.4746518168669047e-06, + "loss": 0.753, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00014746518168669046, + "step": 6300 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2262, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.4738952438737858e-06, + "loss": 0.819, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00014738952438737857, + "step": 6301 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2262, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.4731387838776088e-06, + "loss": 0.7432, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00014731387838776087, + "step": 6302 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2261, + "image_masks_loss": 0.0161, + "llm_learning_rate": 1.472382436961676e-06, + "loss": 0.7456, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.00014723824369616758, + "step": 6303 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.226, + "image_masks_loss": 0.0181, + "llm_learning_rate": 1.4716262032092776e-06, + "loss": 0.7189, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.00014716262032092776, + "step": 6304 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.226, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.4708700827036943e-06, + "loss": 0.7054, + "output_text_mask_loss": 0.0136, + "predictor_learning_rate": 0.00014708700827036943, + "step": 6305 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2259, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.4701140755281878e-06, + "loss": 0.642, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00014701140755281878, + "step": 6306 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2259, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.4693581817660141e-06, + "loss": 0.6712, + "output_text_mask_loss": 0.0561, + "predictor_learning_rate": 0.0001469358181766014, + "step": 6307 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2258, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.468602401500413e-06, + "loss": 0.7843, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.00014686024015004128, + "step": 6308 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2258, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.4678467348146124e-06, + "loss": 0.8105, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 0.00014678467348146123, + "step": 6309 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2257, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.4670911817918282e-06, + "loss": 0.793, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.0001467091181791828, + "step": 6310 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2257, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.4663357425152644e-06, + "loss": 0.7137, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00014663357425152642, + "step": 6311 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2256, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.465580417068111e-06, + "loss": 0.7046, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.00014655804170681108, + "step": 6312 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2256, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.4648252055335454e-06, + "loss": 0.729, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00014648252055335454, + "step": 6313 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2255, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.4640701079947347e-06, + "loss": 0.6905, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00014640701079947347, + "step": 6314 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2255, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.4633151245348308e-06, + "loss": 0.6818, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00014633151245348308, + "step": 6315 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2254, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.462560255236975e-06, + "loss": 0.7147, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00014625602552369748, + "step": 6316 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2254, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.461805500184294e-06, + "loss": 0.7297, + "output_text_mask_loss": 0.0138, + "predictor_learning_rate": 0.0001461805500184294, + "step": 6317 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2253, + "image_masks_loss": 0.0147, + "llm_learning_rate": 1.4610508594599044e-06, + "loss": 0.8082, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00014610508594599042, + "step": 6318 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2252, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.4602963331469083e-06, + "loss": 0.6881, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00014602963331469083, + "step": 6319 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2252, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.4595419213283952e-06, + "loss": 0.7288, + "output_text_mask_loss": 0.0237, + "predictor_learning_rate": 0.0001459541921328395, + "step": 6320 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2251, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.4587876240874438e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0425, + "predictor_learning_rate": 0.00014587876240874437, + "step": 6321 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2251, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.4580334415071185e-06, + "loss": 0.7435, + "output_text_mask_loss": 0.0128, + "predictor_learning_rate": 0.00014580334415071185, + "step": 6322 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.225, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.4572793736704713e-06, + "loss": 0.7545, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 0.0001457279373670471, + "step": 6323 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.225, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.4565254206605422e-06, + "loss": 0.7445, + "output_text_mask_loss": 0.0108, + "predictor_learning_rate": 0.00014565254206605423, + "step": 6324 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2249, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.455771582560358e-06, + "loss": 0.7143, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001455771582560358, + "step": 6325 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2249, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.455017859452933e-06, + "loss": 0.7022, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00014550178594529328, + "step": 6326 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2248, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.454264251421269e-06, + "loss": 0.6557, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.0001454264251421269, + "step": 6327 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2248, + "image_masks_loss": 0.0197, + "llm_learning_rate": 1.453510758548355e-06, + "loss": 0.7487, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00014535107585483549, + "step": 6328 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2247, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.4527573809171671e-06, + "loss": 0.74, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0001452757380917167, + "step": 6329 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2247, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.4520041186106692e-06, + "loss": 0.6595, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0001452004118610669, + "step": 6330 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2246, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.4512509717118108e-06, + "loss": 0.5991, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 0.00014512509717118109, + "step": 6331 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2246, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.4504979403035326e-06, + "loss": 0.7794, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 0.00014504979403035325, + "step": 6332 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2245, + "image_masks_loss": 0.0169, + "llm_learning_rate": 1.449745024468759e-06, + "loss": 0.7558, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00014497450244687588, + "step": 6333 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2245, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.4489922242904025e-06, + "loss": 0.8142, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 0.00014489922242904026, + "step": 6334 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2244, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.4482395398513633e-06, + "loss": 0.809, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001448239539851363, + "step": 6335 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2243, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.4474869712345286e-06, + "loss": 0.746, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00014474869712345284, + "step": 6336 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2243, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.4467345185227732e-06, + "loss": 0.6909, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001446734518522773, + "step": 6337 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2242, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.4459821817989575e-06, + "loss": 0.7637, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 0.00014459821817989574, + "step": 6338 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2242, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.4452299611459329e-06, + "loss": 0.6818, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00014452299611459327, + "step": 6339 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2241, + "image_masks_loss": 0.004, + "llm_learning_rate": 1.444477856646534e-06, + "loss": 0.6277, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0001444477856646534, + "step": 6340 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2241, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.4437258683835847e-06, + "loss": 0.6241, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00014437258683835846, + "step": 6341 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.224, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.442973996439897e-06, + "loss": 0.7158, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00014429739964398969, + "step": 6342 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.224, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.4422222408982659e-06, + "loss": 0.6676, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00014422222408982659, + "step": 6343 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2239, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.4414706018414765e-06, + "loss": 0.6623, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 0.00014414706018414764, + "step": 6344 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2239, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.4407190793523035e-06, + "loss": 0.7899, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 0.00014407190793523035, + "step": 6345 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2238, + "image_masks_loss": 0.0116, + "llm_learning_rate": 1.439967673513505e-06, + "loss": 0.8244, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.00014399676735135048, + "step": 6346 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2238, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.4392163844078272e-06, + "loss": 0.7671, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.0001439216384407827, + "step": 6347 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2237, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.4384652121180043e-06, + "loss": 0.6782, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0001438465212118004, + "step": 6348 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2237, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.4377141567267567e-06, + "loss": 0.7246, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00014377141567267564, + "step": 6349 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2236, + "image_masks_loss": 0.0144, + "llm_learning_rate": 1.436963218316792e-06, + "loss": 0.6947, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 0.0001436963218316792, + "step": 6350 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2236, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.4362123969708043e-06, + "loss": 0.6906, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00014362123969708043, + "step": 6351 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2235, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.4354616927714785e-06, + "loss": 0.7711, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00014354616927714784, + "step": 6352 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2234, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.4347111058014818e-06, + "loss": 0.7331, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00014347111058014817, + "step": 6353 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2234, + "image_masks_loss": 0.0034, + "llm_learning_rate": 1.4339606361434714e-06, + "loss": 0.6787, + "output_text_mask_loss": 0.0212, + "predictor_learning_rate": 0.00014339606361434711, + "step": 6354 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2233, + "image_masks_loss": 0.0165, + "llm_learning_rate": 1.43321028388009e-06, + "loss": 0.7355, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00014332102838800898, + "step": 6355 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2233, + "image_masks_loss": 0.0183, + "llm_learning_rate": 1.4324600490939682e-06, + "loss": 0.751, + "output_text_mask_loss": 0.0146, + "predictor_learning_rate": 0.0001432460049093968, + "step": 6356 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2232, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.4317099318677242e-06, + "loss": 0.769, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.0001431709931867724, + "step": 6357 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2232, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.4309599322839605e-06, + "loss": 0.854, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00014309599322839602, + "step": 6358 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2231, + "image_masks_loss": 0.0131, + "llm_learning_rate": 1.4302100504252714e-06, + "loss": 0.6654, + "output_text_mask_loss": 0.0097, + "predictor_learning_rate": 0.00014302100504252712, + "step": 6359 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2231, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.4294602863742344e-06, + "loss": 0.7208, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00014294602863742343, + "step": 6360 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.223, + "image_masks_loss": 0.0235, + "llm_learning_rate": 1.428710640213416e-06, + "loss": 0.779, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 0.0001428710640213416, + "step": 6361 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.223, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.4279611120253673e-06, + "loss": 0.7153, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00014279611120253673, + "step": 6362 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2229, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.427211701892628e-06, + "loss": 0.7197, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00014272117018926278, + "step": 6363 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2229, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.4264624098977265e-06, + "loss": 0.6566, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00014264624098977264, + "step": 6364 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2228, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.4257132361231754e-06, + "loss": 0.7794, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00014257132361231753, + "step": 6365 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2228, + "image_masks_loss": 0.003, + "llm_learning_rate": 1.4249641806514757e-06, + "loss": 0.7761, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00014249641806514756, + "step": 6366 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2227, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.4242152435651152e-06, + "loss": 0.7509, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00014242152435651152, + "step": 6367 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2227, + "image_masks_loss": 0.0163, + "llm_learning_rate": 1.4234664249465682e-06, + "loss": 0.6837, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.00014234664249465682, + "step": 6368 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2226, + "image_masks_loss": 0.0089, + "llm_learning_rate": 1.4227177248782964e-06, + "loss": 0.7569, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 0.00014227177248782964, + "step": 6369 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2226, + "image_masks_loss": 0.0187, + "llm_learning_rate": 1.4219691434427472e-06, + "loss": 0.7128, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001421969143442747, + "step": 6370 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2225, + "image_masks_loss": 0.0234, + "llm_learning_rate": 1.4212206807223583e-06, + "loss": 0.6735, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00014212206807223582, + "step": 6371 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2225, + "image_masks_loss": 0.0012, + "llm_learning_rate": 1.4204723367995504e-06, + "loss": 0.7021, + "output_text_mask_loss": 0.0463, + "predictor_learning_rate": 0.00014204723367995504, + "step": 6372 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2224, + "image_masks_loss": 0.0257, + "llm_learning_rate": 1.419724111756734e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0001419724111756734, + "step": 6373 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2223, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.4189760056763041e-06, + "loss": 0.7257, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.0001418976005676304, + "step": 6374 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2223, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.418228018640645e-06, + "loss": 0.7379, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00014182280186406449, + "step": 6375 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2222, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.4174801507321253e-06, + "loss": 0.7774, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00014174801507321253, + "step": 6376 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2222, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.416732402033102e-06, + "loss": 0.8038, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.0001416732402033102, + "step": 6377 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2221, + "image_masks_loss": 0.0236, + "llm_learning_rate": 1.4159847726259204e-06, + "loss": 0.749, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00014159847726259205, + "step": 6378 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2221, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.415237262592911e-06, + "loss": 0.7776, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.0001415237262592911, + "step": 6379 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.222, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.41448987201639e-06, + "loss": 0.8086, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00014144898720163898, + "step": 6380 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.222, + "image_masks_loss": 0.0116, + "llm_learning_rate": 1.413742600978662e-06, + "loss": 0.7956, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0001413742600978662, + "step": 6381 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2219, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.412995449562018e-06, + "loss": 0.7954, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00014129954495620178, + "step": 6382 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2219, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.4122484178487355e-06, + "loss": 0.6904, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.00014122484178487354, + "step": 6383 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2218, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.4115015059210812e-06, + "loss": 0.8266, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.0001411501505921081, + "step": 6384 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2218, + "image_masks_loss": 0.0066, + "llm_learning_rate": 1.4107547138613053e-06, + "loss": 0.7272, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00014107547138613051, + "step": 6385 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2217, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.4100080417516466e-06, + "loss": 0.6831, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 0.00014100080417516465, + "step": 6386 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2217, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.4092614896743303e-06, + "loss": 0.6896, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 0.00014092614896743302, + "step": 6387 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2216, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.4085150577115683e-06, + "loss": 0.8368, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00014085150577115683, + "step": 6388 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2216, + "image_masks_loss": 0.0155, + "llm_learning_rate": 1.4077687459455593e-06, + "loss": 0.7279, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00014077687459455594, + "step": 6389 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2215, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.407022554458488e-06, + "loss": 0.6688, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0001407022554458488, + "step": 6390 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2215, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.406276483332528e-06, + "loss": 0.7043, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001406276483332528, + "step": 6391 + }, + { + "epoch": 0.65, + "gumbel_tau": 0.2214, + "image_masks_loss": 0.0183, + "llm_learning_rate": 1.4055305326498385e-06, + "loss": 0.8212, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00014055305326498386, + "step": 6392 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2214, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.4047847024925643e-06, + "loss": 0.6078, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00014047847024925642, + "step": 6393 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2213, + "image_masks_loss": 0.0131, + "llm_learning_rate": 1.404038992942838e-06, + "loss": 0.8128, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.0001404038992942838, + "step": 6394 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2212, + "image_masks_loss": 0.0205, + "llm_learning_rate": 1.4032934040827792e-06, + "loss": 0.7585, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.00014032934040827792, + "step": 6395 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2212, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.4025479359944932e-06, + "loss": 0.6367, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.00014025479359944931, + "step": 6396 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2211, + "image_masks_loss": 0.0189, + "llm_learning_rate": 1.4018025887600722e-06, + "loss": 0.764, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.00014018025887600722, + "step": 6397 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2211, + "image_masks_loss": 0.0223, + "llm_learning_rate": 1.4010573624615986e-06, + "loss": 0.7151, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00014010573624615986, + "step": 6398 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.221, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.4003122571811345e-06, + "loss": 0.7609, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00014003122571811343, + "step": 6399 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.221, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.3995672730007343e-06, + "loss": 0.7781, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001399567273000734, + "step": 6400 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2209, + "image_masks_loss": 0.0177, + "llm_learning_rate": 1.3988224100024372e-06, + "loss": 0.7003, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.00013988224100024371, + "step": 6401 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2209, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.3980776682682681e-06, + "loss": 0.7989, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0001398077668268268, + "step": 6402 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2208, + "image_masks_loss": 0.0182, + "llm_learning_rate": 1.3973330478802416e-06, + "loss": 0.7182, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00013973330478802415, + "step": 6403 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2208, + "image_masks_loss": 0.0205, + "llm_learning_rate": 1.3965885489203556e-06, + "loss": 0.8022, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00013965885489203555, + "step": 6404 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2207, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.3958441714705968e-06, + "loss": 0.7205, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00013958441714705965, + "step": 6405 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2207, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.3950999156129372e-06, + "loss": 0.7228, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013950999156129372, + "step": 6406 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2206, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.3943557814293357e-06, + "loss": 0.7386, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00013943557814293357, + "step": 6407 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2206, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.3936117690017386e-06, + "loss": 0.7741, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00013936117690017385, + "step": 6408 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2205, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.3928678784120766e-06, + "loss": 0.7605, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013928678784120766, + "step": 6409 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2205, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.3921241097422711e-06, + "loss": 0.7118, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.0001392124109742271, + "step": 6410 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2204, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.3913804630742262e-06, + "loss": 0.7768, + "output_text_mask_loss": 0.0584, + "predictor_learning_rate": 0.0001391380463074226, + "step": 6411 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2204, + "image_masks_loss": 0.0114, + "llm_learning_rate": 1.3906369384898339e-06, + "loss": 0.7372, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00013906369384898338, + "step": 6412 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2203, + "image_masks_loss": 0.0151, + "llm_learning_rate": 1.3898935360709732e-06, + "loss": 0.6929, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.0001389893536070973, + "step": 6413 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2203, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.3891502558995088e-06, + "loss": 0.7796, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.00013891502558995087, + "step": 6414 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2202, + "image_masks_loss": 0.0145, + "llm_learning_rate": 1.3884070980572927e-06, + "loss": 0.7953, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00013884070980572927, + "step": 6415 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2202, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.387664062626163e-06, + "loss": 0.7179, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.0001387664062626163, + "step": 6416 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2201, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.386921149687944e-06, + "loss": 0.7051, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00013869211496879438, + "step": 6417 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.22, + "image_masks_loss": 0.0161, + "llm_learning_rate": 1.3861783593244476e-06, + "loss": 0.7424, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.00013861783593244477, + "step": 6418 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.22, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.385435691617471e-06, + "loss": 0.7223, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001385435691617471, + "step": 6419 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2199, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.384693146648799e-06, + "loss": 0.7298, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00013846931466487988, + "step": 6420 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2199, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.3839507245002015e-06, + "loss": 0.723, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00013839507245002014, + "step": 6421 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2198, + "image_masks_loss": 0.0176, + "llm_learning_rate": 1.3832084252534353e-06, + "loss": 0.789, + "output_text_mask_loss": 0.0698, + "predictor_learning_rate": 0.0001383208425253435, + "step": 6422 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2198, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.3824662489902457e-06, + "loss": 0.8051, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 0.00013824662489902455, + "step": 6423 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2197, + "image_masks_loss": 0.0202, + "llm_learning_rate": 1.381724195792362e-06, + "loss": 0.7042, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001381724195792362, + "step": 6424 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2197, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.3809822657415013e-06, + "loss": 0.8782, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.0001380982265741501, + "step": 6425 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2196, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.3802404589193653e-06, + "loss": 0.6902, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.0001380240458919365, + "step": 6426 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2196, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.3794987754076443e-06, + "loss": 0.6885, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013794987754076442, + "step": 6427 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2195, + "image_masks_loss": 0.0225, + "llm_learning_rate": 1.378757215288014e-06, + "loss": 0.6565, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0001378757215288014, + "step": 6428 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2195, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.3780157786421361e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001378015778642136, + "step": 6429 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2194, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.37727446555166e-06, + "loss": 0.7616, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.000137727446555166, + "step": 6430 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2194, + "image_masks_loss": 0.0207, + "llm_learning_rate": 1.3765332760982212e-06, + "loss": 0.7705, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00013765332760982212, + "step": 6431 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2193, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.3757922103634402e-06, + "loss": 0.7183, + "output_text_mask_loss": 0.0016, + "predictor_learning_rate": 0.00013757922103634402, + "step": 6432 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2193, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.3750512684289246e-06, + "loss": 0.7129, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013750512684289245, + "step": 6433 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2192, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.3743104503762697e-06, + "loss": 0.7161, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 0.00013743104503762695, + "step": 6434 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2192, + "image_masks_loss": 0.0212, + "llm_learning_rate": 1.3735697562870553e-06, + "loss": 0.743, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00013735697562870553, + "step": 6435 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2191, + "image_masks_loss": 0.0317, + "llm_learning_rate": 1.372829186242848e-06, + "loss": 0.7055, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00013728291862428479, + "step": 6436 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2191, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.3720887403252018e-06, + "loss": 0.7321, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00013720887403252015, + "step": 6437 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.219, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.3713484186156555e-06, + "loss": 0.7325, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00013713484186156556, + "step": 6438 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.219, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.3706082211957355e-06, + "loss": 0.6838, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00013706082211957355, + "step": 6439 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2189, + "image_masks_loss": 0.0023, + "llm_learning_rate": 1.369868148146954e-06, + "loss": 0.737, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.0001369868148146954, + "step": 6440 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2189, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.3691281995508076e-06, + "loss": 0.7395, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.00013691281995508077, + "step": 6441 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2188, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.3683883754887844e-06, + "loss": 0.7902, + "output_text_mask_loss": 0.0664, + "predictor_learning_rate": 0.00013683883754887844, + "step": 6442 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2188, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.3676486760423538e-06, + "loss": 0.6959, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013676486760423538, + "step": 6443 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2187, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.366909101292973e-06, + "loss": 0.91, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001366909101292973, + "step": 6444 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2187, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.366169651322086e-06, + "loss": 0.756, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00013661696513220858, + "step": 6445 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2186, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.3654303262111223e-06, + "loss": 0.7389, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00013654303262111223, + "step": 6446 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2185, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.3646911260414985e-06, + "loss": 0.7955, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00013646911260414984, + "step": 6447 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2185, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.3639520508946156e-06, + "loss": 0.6881, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00013639520508946154, + "step": 6448 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2184, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.3632131008518642e-06, + "loss": 0.7218, + "output_text_mask_loss": 0.0441, + "predictor_learning_rate": 0.0001363213100851864, + "step": 6449 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2184, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.3624742759946183e-06, + "loss": 0.7375, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.00013624742759946182, + "step": 6450 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2183, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.361735576404239e-06, + "loss": 0.6021, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0001361735576404239, + "step": 6451 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2183, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.3609970021620733e-06, + "loss": 0.756, + "output_text_mask_loss": 0.0576, + "predictor_learning_rate": 0.00013609970021620732, + "step": 6452 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2182, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.3602585533494545e-06, + "loss": 0.7036, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00013602585533494544, + "step": 6453 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2182, + "image_masks_loss": 0.0184, + "llm_learning_rate": 1.3595202300477028e-06, + "loss": 0.7247, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00013595202300477026, + "step": 6454 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2181, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.3587820323381235e-06, + "loss": 0.7353, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 0.00013587820323381232, + "step": 6455 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2181, + "image_masks_loss": 0.0158, + "llm_learning_rate": 1.358043960302009e-06, + "loss": 0.8018, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 0.0001358043960302009, + "step": 6456 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.218, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.357306014020637e-06, + "loss": 0.8062, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0001357306014020637, + "step": 6457 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.218, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.3565681935752722e-06, + "loss": 0.7493, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 0.00013565681935752722, + "step": 6458 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2179, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.3558304990471643e-06, + "loss": 0.6921, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00013558304990471642, + "step": 6459 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2179, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.3550929305175508e-06, + "loss": 0.7374, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00013550929305175508, + "step": 6460 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2178, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.354355488067653e-06, + "loss": 0.814, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.0001354355488067653, + "step": 6461 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2178, + "image_masks_loss": 0.0029, + "llm_learning_rate": 1.3536181717786816e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.00013536181717786817, + "step": 6462 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2177, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.3528809817318307e-06, + "loss": 0.7401, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013528809817318305, + "step": 6463 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2177, + "image_masks_loss": 0.0217, + "llm_learning_rate": 1.3521439180082812e-06, + "loss": 0.7192, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.00013521439180082812, + "step": 6464 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2176, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.3514069806892e-06, + "loss": 0.8068, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 0.00013514069806891998, + "step": 6465 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2176, + "image_masks_loss": 0.0231, + "llm_learning_rate": 1.3506701698557406e-06, + "loss": 0.6862, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.00013506701698557405, + "step": 6466 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2175, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.349933485589042e-06, + "loss": 0.7908, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00013499334855890418, + "step": 6467 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2175, + "image_masks_loss": 0.0027, + "llm_learning_rate": 1.349196927970229e-06, + "loss": 0.7219, + "output_text_mask_loss": 0.0267, + "predictor_learning_rate": 0.00013491969279702288, + "step": 6468 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2174, + "image_masks_loss": 0.0252, + "llm_learning_rate": 1.3484604970804142e-06, + "loss": 0.7778, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 0.0001348460497080414, + "step": 6469 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2174, + "image_masks_loss": 0.0212, + "llm_learning_rate": 1.3477241930006946e-06, + "loss": 0.7423, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013477241930006945, + "step": 6470 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2173, + "image_masks_loss": 0.0213, + "llm_learning_rate": 1.3469880158121532e-06, + "loss": 0.7317, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 0.00013469880158121532, + "step": 6471 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2173, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.346251965595861e-06, + "loss": 0.7337, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001346251965595861, + "step": 6472 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2172, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.34551604243287e-06, + "loss": 0.8038, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.000134551604243287, + "step": 6473 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2172, + "image_masks_loss": 0.0123, + "llm_learning_rate": 1.3447802464042252e-06, + "loss": 0.7515, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00013447802464042254, + "step": 6474 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2171, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.3440445775909526e-06, + "loss": 0.5924, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013440445775909527, + "step": 6475 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2171, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.3433090360740661e-06, + "loss": 0.7022, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00013433090360740661, + "step": 6476 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.217, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.3425736219345647e-06, + "loss": 0.7639, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00013425736219345647, + "step": 6477 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.217, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.3418383352534344e-06, + "loss": 0.6546, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013418383352534343, + "step": 6478 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2169, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.3411031761116462e-06, + "loss": 0.7754, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0001341103176111646, + "step": 6479 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2169, + "image_masks_loss": 0.0167, + "llm_learning_rate": 1.3403681445901568e-06, + "loss": 0.7955, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00013403681445901567, + "step": 6480 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2168, + "image_masks_loss": 0.0225, + "llm_learning_rate": 1.339633240769911e-06, + "loss": 0.711, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.0001339633240769911, + "step": 6481 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2168, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.3388984647318377e-06, + "loss": 0.6875, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00013388984647318375, + "step": 6482 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2167, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.338163816556851e-06, + "loss": 0.7485, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00013381638165568512, + "step": 6483 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2166, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.3374292963258534e-06, + "loss": 0.7416, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00013374292963258533, + "step": 6484 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2166, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.3366949041197308e-06, + "loss": 0.6838, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 0.00013366949041197307, + "step": 6485 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2165, + "image_masks_loss": 0.0043, + "llm_learning_rate": 1.335960640019357e-06, + "loss": 0.7828, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00013359606400193568, + "step": 6486 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2165, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.335226504105589e-06, + "loss": 0.7701, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.0001335226504105589, + "step": 6487 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2164, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.3344924964592737e-06, + "loss": 0.802, + "output_text_mask_loss": 0.061, + "predictor_learning_rate": 0.00013344924964592736, + "step": 6488 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2164, + "image_masks_loss": 0.0167, + "llm_learning_rate": 1.333758617161241e-06, + "loss": 0.6913, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001333758617161241, + "step": 6489 + }, + { + "epoch": 0.66, + "gumbel_tau": 0.2163, + "image_masks_loss": 0.003, + "llm_learning_rate": 1.3330248662923068e-06, + "loss": 0.7528, + "output_text_mask_loss": 0.0713, + "predictor_learning_rate": 0.00013330248662923066, + "step": 6490 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2163, + "image_masks_loss": 0.0015, + "llm_learning_rate": 1.332291243933275e-06, + "loss": 0.6875, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.0001332291243933275, + "step": 6491 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2162, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.3315577501649312e-06, + "loss": 0.8075, + "output_text_mask_loss": 0.008, + "predictor_learning_rate": 0.00013315577501649313, + "step": 6492 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2162, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.3308243850680496e-06, + "loss": 0.6928, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013308243850680495, + "step": 6493 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2161, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.330091148723392e-06, + "loss": 0.7303, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0001330091148723392, + "step": 6494 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2161, + "image_masks_loss": 0.0292, + "llm_learning_rate": 1.3293580412117026e-06, + "loss": 0.7411, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 0.00013293580412117027, + "step": 6495 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.216, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.3286250626137136e-06, + "loss": 0.748, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013286250626137137, + "step": 6496 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.216, + "image_masks_loss": 0.0221, + "llm_learning_rate": 1.3278922130101418e-06, + "loss": 0.6656, + "output_text_mask_loss": 0.0748, + "predictor_learning_rate": 0.00013278922130101418, + "step": 6497 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2159, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.32715949248169e-06, + "loss": 0.6855, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000132715949248169, + "step": 6498 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2159, + "image_masks_loss": 0.0213, + "llm_learning_rate": 1.3264269011090474e-06, + "loss": 0.7381, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013264269011090474, + "step": 6499 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2158, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.3256944389728877e-06, + "loss": 0.7499, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013256944389728874, + "step": 6500 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2158, + "image_masks_loss": 0.0045, + "llm_learning_rate": 1.324962106153873e-06, + "loss": 0.8088, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.0001324962106153873, + "step": 6501 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2157, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.3242299027326484e-06, + "loss": 0.6464, + "output_text_mask_loss": 0.0626, + "predictor_learning_rate": 0.00013242299027326482, + "step": 6502 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2157, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.3234978287898458e-06, + "loss": 0.7436, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00013234978287898457, + "step": 6503 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2156, + "image_masks_loss": 0.0212, + "llm_learning_rate": 1.3227658844060828e-06, + "loss": 0.7862, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00013227658844060827, + "step": 6504 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2156, + "image_masks_loss": 0.0183, + "llm_learning_rate": 1.322034069661963e-06, + "loss": 0.7466, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.0001322034069661963, + "step": 6505 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2155, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.3213023846380751e-06, + "loss": 0.7772, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 0.0001321302384638075, + "step": 6506 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2155, + "image_masks_loss": 0.0125, + "llm_learning_rate": 1.320570829414993e-06, + "loss": 0.6848, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0001320570829414993, + "step": 6507 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2154, + "image_masks_loss": 0.0239, + "llm_learning_rate": 1.3198394040732792e-06, + "loss": 0.7603, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00013198394040732792, + "step": 6508 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2154, + "image_masks_loss": 0.0139, + "llm_learning_rate": 1.3191081086934805e-06, + "loss": 0.7638, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 0.00013191081086934804, + "step": 6509 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2153, + "image_masks_loss": 0.0035, + "llm_learning_rate": 1.3183769433561255e-06, + "loss": 0.7362, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 0.00013183769433561254, + "step": 6510 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2153, + "image_masks_loss": 0.0095, + "llm_learning_rate": 1.3176459081417336e-06, + "loss": 0.712, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013176459081417336, + "step": 6511 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2152, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.316915003130807e-06, + "loss": 0.7942, + "output_text_mask_loss": 0.0315, + "predictor_learning_rate": 0.0001316915003130807, + "step": 6512 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2152, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.3161842284038365e-06, + "loss": 0.6324, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.00013161842284038365, + "step": 6513 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2151, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.3154535840412957e-06, + "loss": 0.6815, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00013154535840412956, + "step": 6514 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2151, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.3147230701236447e-06, + "loss": 0.7577, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00013147230701236447, + "step": 6515 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.215, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.313992686731329e-06, + "loss": 0.7386, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 0.0001313992686731329, + "step": 6516 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.215, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.3132624339447805e-06, + "loss": 0.7764, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 0.00013132624339447804, + "step": 6517 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2149, + "image_masks_loss": 0.0123, + "llm_learning_rate": 1.3125323118444162e-06, + "loss": 0.7221, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001312532311844416, + "step": 6518 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2149, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.3118023205106379e-06, + "loss": 0.6855, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.00013118023205106378, + "step": 6519 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2148, + "image_masks_loss": 0.0357, + "llm_learning_rate": 1.3110724600238354e-06, + "loss": 0.7334, + "output_text_mask_loss": 0.0372, + "predictor_learning_rate": 0.00013110724600238352, + "step": 6520 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2148, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.3103427304643818e-06, + "loss": 0.7226, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.00013103427304643817, + "step": 6521 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2147, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.3096131319126365e-06, + "loss": 0.649, + "output_text_mask_loss": 0.0362, + "predictor_learning_rate": 0.00013096131319126364, + "step": 6522 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2147, + "image_masks_loss": 0.0201, + "llm_learning_rate": 1.308883664448945e-06, + "loss": 0.7306, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00013088836644489449, + "step": 6523 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2146, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.3081543281536373e-06, + "loss": 0.8076, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00013081543281536374, + "step": 6524 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2146, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.3074251231070295e-06, + "loss": 0.7589, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00013074251231070294, + "step": 6525 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2145, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.3066960493894226e-06, + "loss": 0.8041, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00013066960493894226, + "step": 6526 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2145, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.3059671070811058e-06, + "loss": 0.632, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00013059671070811058, + "step": 6527 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2144, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.305238296262352e-06, + "loss": 0.6378, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00013052382962623518, + "step": 6528 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2144, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.304509617013417e-06, + "loss": 0.7895, + "output_text_mask_loss": 0.0418, + "predictor_learning_rate": 0.0001304509617013417, + "step": 6529 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2143, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.303781069414546e-06, + "loss": 0.8243, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 0.00013037810694145458, + "step": 6530 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2143, + "image_masks_loss": 0.0243, + "llm_learning_rate": 1.3030526535459675e-06, + "loss": 0.7395, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00013030526535459674, + "step": 6531 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2142, + "image_masks_loss": 0.0193, + "llm_learning_rate": 1.3023243694878967e-06, + "loss": 0.6592, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00013023243694878966, + "step": 6532 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2142, + "image_masks_loss": 0.0151, + "llm_learning_rate": 1.3015962173205345e-06, + "loss": 0.7976, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00013015962173205345, + "step": 6533 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2141, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.3008681971240664e-06, + "loss": 0.7195, + "output_text_mask_loss": 0.0296, + "predictor_learning_rate": 0.00013008681971240664, + "step": 6534 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2141, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.3001403089786635e-06, + "loss": 0.7543, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00013001403089786634, + "step": 6535 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.214, + "image_masks_loss": 0.0033, + "llm_learning_rate": 1.2994125529644821e-06, + "loss": 0.7413, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00012994125529644822, + "step": 6536 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.214, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.298684929161665e-06, + "loss": 0.7784, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 0.00012986849291616649, + "step": 6537 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2139, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.2979574376503387e-06, + "loss": 0.7057, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.00012979574376503387, + "step": 6538 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2139, + "image_masks_loss": 0.011, + "llm_learning_rate": 1.2972300785106161e-06, + "loss": 0.7454, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00012972300785106162, + "step": 6539 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2138, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.2965028518225976e-06, + "loss": 0.7212, + "output_text_mask_loss": 0.025, + "predictor_learning_rate": 0.00012965028518225973, + "step": 6540 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2138, + "image_masks_loss": 0.0026, + "llm_learning_rate": 1.2957757576663655e-06, + "loss": 0.8, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00012957757576663654, + "step": 6541 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2137, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.2950487961219891e-06, + "loss": 0.695, + "output_text_mask_loss": 0.0444, + "predictor_learning_rate": 0.0001295048796121989, + "step": 6542 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2137, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.2943219672695233e-06, + "loss": 0.7578, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 0.00012943219672695235, + "step": 6543 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2136, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.293595271189008e-06, + "loss": 0.8084, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001293595271189008, + "step": 6544 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2136, + "image_masks_loss": 0.0034, + "llm_learning_rate": 1.2928687079604687e-06, + "loss": 0.7096, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00012928687079604686, + "step": 6545 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2135, + "image_masks_loss": 0.0034, + "llm_learning_rate": 1.2921422776639159e-06, + "loss": 0.785, + "output_text_mask_loss": 0.0443, + "predictor_learning_rate": 0.00012921422776639158, + "step": 6546 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2135, + "image_masks_loss": 0.0131, + "llm_learning_rate": 1.291415980379346e-06, + "loss": 0.7678, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.0001291415980379346, + "step": 6547 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2134, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.2906898161867404e-06, + "loss": 0.7056, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 0.00012906898161867402, + "step": 6548 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2134, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.289963785166066e-06, + "loss": 0.5749, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00012899637851660658, + "step": 6549 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2133, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.2892378873972743e-06, + "loss": 0.7206, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 0.00012892378873972742, + "step": 6550 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2133, + "image_masks_loss": 0.0221, + "llm_learning_rate": 1.288512122960304e-06, + "loss": 0.6654, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001288512122960304, + "step": 6551 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2132, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.2877864919350757e-06, + "loss": 0.7858, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.00012877864919350756, + "step": 6552 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2132, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.2870609944015003e-06, + "loss": 0.7631, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012870609944015, + "step": 6553 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2131, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.2863356304394696e-06, + "loss": 0.7208, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00012863356304394696, + "step": 6554 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2131, + "image_masks_loss": 0.0157, + "llm_learning_rate": 1.285610400128863e-06, + "loss": 0.745, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001285610400128863, + "step": 6555 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.213, + "image_masks_loss": 0.0155, + "llm_learning_rate": 1.284885303549544e-06, + "loss": 0.7087, + "output_text_mask_loss": 0.069, + "predictor_learning_rate": 0.0001284885303549544, + "step": 6556 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.213, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.284160340781362e-06, + "loss": 0.7685, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 0.00012841603407813619, + "step": 6557 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2129, + "image_masks_loss": 0.0089, + "llm_learning_rate": 1.2834355119041507e-06, + "loss": 0.7566, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.0001283435511904151, + "step": 6558 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2129, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.2827108169977304e-06, + "loss": 0.7551, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012827108169977304, + "step": 6559 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2128, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.2819862561419071e-06, + "loss": 0.7677, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 0.0001281986256141907, + "step": 6560 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2127, + "image_masks_loss": 0.0027, + "llm_learning_rate": 1.2812618294164703e-06, + "loss": 0.6709, + "output_text_mask_loss": 0.0127, + "predictor_learning_rate": 0.00012812618294164702, + "step": 6561 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2127, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.2805375369011957e-06, + "loss": 0.7069, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00012805375369011958, + "step": 6562 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2126, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.2798133786758434e-06, + "loss": 0.7053, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00012798133786758435, + "step": 6563 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2126, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.2790893548201599e-06, + "loss": 0.754, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.000127908935482016, + "step": 6564 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2125, + "image_masks_loss": 0.0066, + "llm_learning_rate": 1.278365465413876e-06, + "loss": 0.6745, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001278365465413876, + "step": 6565 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2125, + "image_masks_loss": 0.0158, + "llm_learning_rate": 1.2776417105367084e-06, + "loss": 0.6609, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012776417105367084, + "step": 6566 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2124, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.276918090268358e-06, + "loss": 0.7758, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001276918090268358, + "step": 6567 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2124, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.276194604688512e-06, + "loss": 0.751, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001276194604688512, + "step": 6568 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2123, + "image_masks_loss": 0.005, + "llm_learning_rate": 1.275471253876842e-06, + "loss": 0.727, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0001275471253876842, + "step": 6569 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2123, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.274748037913005e-06, + "loss": 0.7988, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 0.0001274748037913005, + "step": 6570 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2122, + "image_masks_loss": 0.0203, + "llm_learning_rate": 1.2740249568766424e-06, + "loss": 0.7243, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00012740249568766424, + "step": 6571 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2122, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.2733020108473832e-06, + "loss": 0.7804, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.0001273302010847383, + "step": 6572 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2121, + "image_masks_loss": 0.0172, + "llm_learning_rate": 1.272579199904839e-06, + "loss": 0.7224, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001272579199904839, + "step": 6573 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2121, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.271856524128607e-06, + "loss": 0.7342, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.0001271856524128607, + "step": 6574 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.212, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.2711339835982705e-06, + "loss": 0.6219, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00012711339835982705, + "step": 6575 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.212, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.2704115783933968e-06, + "loss": 0.763, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 0.00012704115783933968, + "step": 6576 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2119, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.2696893085935385e-06, + "loss": 0.5764, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00012696893085935385, + "step": 6577 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2119, + "image_masks_loss": 0.0026, + "llm_learning_rate": 1.2689671742782336e-06, + "loss": 0.6692, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 0.00012689671742782335, + "step": 6578 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2118, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.2682451755270062e-06, + "loss": 0.7205, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 0.0001268245175527006, + "step": 6579 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2118, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.2675233124193637e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00012675233124193635, + "step": 6580 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2117, + "image_masks_loss": 0.0257, + "llm_learning_rate": 1.2668015850347993e-06, + "loss": 0.9015, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.00012668015850347993, + "step": 6581 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2117, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.266079993452791e-06, + "loss": 0.6741, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00012660799934527907, + "step": 6582 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2116, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.2653585377528033e-06, + "loss": 0.7418, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0001265358537752803, + "step": 6583 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2116, + "image_masks_loss": 0.0178, + "llm_learning_rate": 1.2646372180142819e-06, + "loss": 0.6972, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00012646372180142817, + "step": 6584 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2115, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.2639160343166622e-06, + "loss": 0.7601, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 0.00012639160343166624, + "step": 6585 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2115, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.2631949867393622e-06, + "loss": 0.8408, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 0.00012631949867393622, + "step": 6586 + }, + { + "epoch": 0.67, + "gumbel_tau": 0.2114, + "image_masks_loss": 0.021, + "llm_learning_rate": 1.2624740753617854e-06, + "loss": 0.6681, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00012624740753617854, + "step": 6587 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2114, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.2617533002633198e-06, + "loss": 0.6427, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00012617533002633196, + "step": 6588 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2113, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.2610326615233388e-06, + "loss": 0.7484, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012610326615233386, + "step": 6589 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2113, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.2603121592212005e-06, + "loss": 0.7132, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.00012603121592212005, + "step": 6590 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2112, + "image_masks_loss": 0.0038, + "llm_learning_rate": 1.2595917934362478e-06, + "loss": 0.7908, + "output_text_mask_loss": 0.0622, + "predictor_learning_rate": 0.00012595917934362478, + "step": 6591 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2112, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.2588715642478108e-06, + "loss": 0.617, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00012588715642478107, + "step": 6592 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2111, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.2581514717352016e-06, + "loss": 0.773, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00012581514717352016, + "step": 6593 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2111, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.2574315159777184e-06, + "loss": 0.7128, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00012574315159777184, + "step": 6594 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.211, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.2567116970546444e-06, + "loss": 0.7692, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 0.00012567116970546444, + "step": 6595 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.211, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.2559920150452475e-06, + "loss": 0.7011, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00012559920150452476, + "step": 6596 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.211, + "image_masks_loss": 0.0207, + "llm_learning_rate": 1.255272470028781e-06, + "loss": 0.6725, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.0001255272470028781, + "step": 6597 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2109, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.2545530620844814e-06, + "loss": 0.6833, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00012545530620844815, + "step": 6598 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2109, + "image_masks_loss": 0.019, + "llm_learning_rate": 1.253833791291574e-06, + "loss": 0.6246, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.0001253833791291574, + "step": 6599 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2108, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.2531146577292651e-06, + "loss": 0.7028, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 0.00012531146577292651, + "step": 6600 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2108, + "image_masks_loss": 0.003, + "llm_learning_rate": 1.2523956614767473e-06, + "loss": 0.8003, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.00012523956614767475, + "step": 6601 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2107, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.2516768026131993e-06, + "loss": 0.7607, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.0001251676802613199, + "step": 6602 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2107, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.2509580812177805e-06, + "loss": 0.7576, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00012509580812177803, + "step": 6603 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2106, + "image_masks_loss": 0.0157, + "llm_learning_rate": 1.2502394973696409e-06, + "loss": 0.7287, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00012502394973696408, + "step": 6604 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2106, + "image_masks_loss": 0.0189, + "llm_learning_rate": 1.2495210511479113e-06, + "loss": 0.7456, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 0.00012495210511479114, + "step": 6605 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2105, + "image_masks_loss": 0.0286, + "llm_learning_rate": 1.248802742631709e-06, + "loss": 0.8199, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.0001248802742631709, + "step": 6606 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2105, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.2480845719001358e-06, + "loss": 0.7359, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00012480845719001356, + "step": 6607 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2104, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.2473665390322779e-06, + "loss": 0.702, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012473665390322778, + "step": 6608 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2104, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.2466486441072068e-06, + "loss": 0.7576, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00012466486441072068, + "step": 6609 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2103, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.245930887203978e-06, + "loss": 0.7743, + "output_text_mask_loss": 0.038, + "predictor_learning_rate": 0.0001245930887203978, + "step": 6610 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2103, + "image_masks_loss": 0.0226, + "llm_learning_rate": 1.2452132684016338e-06, + "loss": 0.8156, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00012452132684016338, + "step": 6611 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2102, + "image_masks_loss": 0.0131, + "llm_learning_rate": 1.2444957877791996e-06, + "loss": 0.7598, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 0.00012444957877791996, + "step": 6612 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2102, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.243778445415686e-06, + "loss": 0.7902, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 0.00012437784454156857, + "step": 6613 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2101, + "image_masks_loss": 0.0298, + "llm_learning_rate": 1.243061241390088e-06, + "loss": 0.709, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.0001243061241390088, + "step": 6614 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2101, + "image_masks_loss": 0.0183, + "llm_learning_rate": 1.242344175781386e-06, + "loss": 0.7613, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00012423441757813857, + "step": 6615 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.21, + "image_masks_loss": 0.022, + "llm_learning_rate": 1.2416272486685443e-06, + "loss": 0.7303, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00012416272486685443, + "step": 6616 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.21, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.240910460130512e-06, + "loss": 0.7503, + "output_text_mask_loss": 0.0626, + "predictor_learning_rate": 0.00012409104601305117, + "step": 6617 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2099, + "image_masks_loss": 0.0334, + "llm_learning_rate": 1.2401938102462255e-06, + "loss": 0.8186, + "output_text_mask_loss": 0.0706, + "predictor_learning_rate": 0.00012401938102462253, + "step": 6618 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2099, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.2394772990946024e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00012394772990946025, + "step": 6619 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2098, + "image_masks_loss": 0.016, + "llm_learning_rate": 1.238760926754548e-06, + "loss": 0.7934, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 0.0001238760926754548, + "step": 6620 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2098, + "image_masks_loss": 0.0158, + "llm_learning_rate": 1.2380446933049487e-06, + "loss": 0.7837, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012380446933049486, + "step": 6621 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2097, + "image_masks_loss": 0.0201, + "llm_learning_rate": 1.2373285988246787e-06, + "loss": 0.723, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00012373285988246784, + "step": 6622 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2097, + "image_masks_loss": 0.0229, + "llm_learning_rate": 1.2366126433925949e-06, + "loss": 0.7266, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00012366126433925947, + "step": 6623 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2096, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.2358968270875418e-06, + "loss": 0.7685, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012358968270875418, + "step": 6624 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2096, + "image_masks_loss": 0.0338, + "llm_learning_rate": 1.235181149988346e-06, + "loss": 0.7588, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.0001235181149988346, + "step": 6625 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2095, + "image_masks_loss": 0.0254, + "llm_learning_rate": 1.2344656121738188e-06, + "loss": 0.7047, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00012344656121738188, + "step": 6626 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2095, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.2337502137227574e-06, + "loss": 0.6881, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00012337502137227575, + "step": 6627 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2094, + "image_masks_loss": 0.0209, + "llm_learning_rate": 1.2330349547139434e-06, + "loss": 0.6544, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012330349547139434, + "step": 6628 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2094, + "image_masks_loss": 0.0286, + "llm_learning_rate": 1.2323198352261418e-06, + "loss": 0.6896, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00012323198352261417, + "step": 6629 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2093, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.2316048553381027e-06, + "loss": 0.7366, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00012316048553381027, + "step": 6630 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2093, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.230890015128563e-06, + "loss": 0.7842, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.0001230890015128563, + "step": 6631 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2092, + "image_masks_loss": 0.0271, + "llm_learning_rate": 1.2301753146762418e-06, + "loss": 0.7631, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00012301753146762417, + "step": 6632 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2092, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.2294607540598435e-06, + "loss": 0.7602, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 0.00012294607540598435, + "step": 6633 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2091, + "image_masks_loss": 0.0353, + "llm_learning_rate": 1.2287463333580567e-06, + "loss": 0.5617, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00012287463333580568, + "step": 6634 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2091, + "image_masks_loss": 0.0299, + "llm_learning_rate": 1.2280320526495556e-06, + "loss": 0.763, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 0.00012280320526495556, + "step": 6635 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.209, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.2273179120129978e-06, + "loss": 0.7395, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.00012273179120129976, + "step": 6636 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.209, + "image_masks_loss": 0.0168, + "llm_learning_rate": 1.2266039115270253e-06, + "loss": 0.7406, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012266039115270253, + "step": 6637 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2089, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.2258900512702677e-06, + "loss": 0.6576, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 0.00012258900512702675, + "step": 6638 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2089, + "image_masks_loss": 0.0029, + "llm_learning_rate": 1.2251763313213363e-06, + "loss": 0.7254, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.00012251763313213363, + "step": 6639 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2088, + "image_masks_loss": 0.0213, + "llm_learning_rate": 1.224462751758826e-06, + "loss": 0.7317, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.0001224462751758826, + "step": 6640 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2088, + "image_masks_loss": 0.0136, + "llm_learning_rate": 1.2237493126613186e-06, + "loss": 0.7327, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012237493126613185, + "step": 6641 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2087, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.2230360141073788e-06, + "loss": 0.6913, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00012230360141073787, + "step": 6642 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2087, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.2223228561755584e-06, + "loss": 0.6861, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 0.00012223228561755584, + "step": 6643 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2086, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.221609838944391e-06, + "loss": 0.7044, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012216098389443908, + "step": 6644 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2086, + "image_masks_loss": 0.0208, + "llm_learning_rate": 1.2208969624923955e-06, + "loss": 0.7298, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012208969624923955, + "step": 6645 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2085, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.2201842268980757e-06, + "loss": 0.7433, + "output_text_mask_loss": 0.0085, + "predictor_learning_rate": 0.00012201842268980756, + "step": 6646 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2085, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.2194716322399195e-06, + "loss": 0.8132, + "output_text_mask_loss": 0.053, + "predictor_learning_rate": 0.00012194716322399193, + "step": 6647 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2084, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.2187591785963993e-06, + "loss": 0.7777, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012187591785963992, + "step": 6648 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2084, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.2180468660459712e-06, + "loss": 0.7872, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00012180468660459713, + "step": 6649 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2083, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.217334694667079e-06, + "loss": 0.8098, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00012173346946670788, + "step": 6650 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2083, + "image_masks_loss": 0.0171, + "llm_learning_rate": 1.216622664538147e-06, + "loss": 0.8028, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00012166226645381468, + "step": 6651 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2082, + "image_masks_loss": 0.0244, + "llm_learning_rate": 1.2159107757375858e-06, + "loss": 0.7137, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00012159107757375857, + "step": 6652 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2082, + "image_masks_loss": 0.0021, + "llm_learning_rate": 1.21519902834379e-06, + "loss": 0.6507, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.000121519902834379, + "step": 6653 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2081, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.2144874224351391e-06, + "loss": 0.7387, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 0.0001214487422435139, + "step": 6654 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2081, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.2137759580899966e-06, + "loss": 0.6612, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 0.00012137759580899966, + "step": 6655 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.208, + "image_masks_loss": 0.0131, + "llm_learning_rate": 1.2130646353867095e-06, + "loss": 0.5977, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00012130646353867094, + "step": 6656 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.208, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.2123534544036133e-06, + "loss": 0.7919, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00012123534544036133, + "step": 6657 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2079, + "image_masks_loss": 0.0186, + "llm_learning_rate": 1.211642415219022e-06, + "loss": 0.6599, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001211642415219022, + "step": 6658 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2079, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.2109315179112374e-06, + "loss": 0.7657, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00012109315179112374, + "step": 6659 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2078, + "image_masks_loss": 0.0222, + "llm_learning_rate": 1.2102207625585454e-06, + "loss": 0.7195, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012102207625585454, + "step": 6660 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2078, + "image_masks_loss": 0.0225, + "llm_learning_rate": 1.2095101492392162e-06, + "loss": 0.7317, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.00012095101492392161, + "step": 6661 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2077, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.2087996780315025e-06, + "loss": 0.6674, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00012087996780315025, + "step": 6662 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2077, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.2080893490136456e-06, + "loss": 0.6904, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00012080893490136455, + "step": 6663 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2076, + "image_masks_loss": 0.0151, + "llm_learning_rate": 1.207379162263867e-06, + "loss": 0.7066, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.00012073791622638669, + "step": 6664 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2076, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.206669117860374e-06, + "loss": 0.7523, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 0.00012066691178603739, + "step": 6665 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2075, + "image_masks_loss": 0.0234, + "llm_learning_rate": 1.2059592158813587e-06, + "loss": 0.7471, + "output_text_mask_loss": 0.0357, + "predictor_learning_rate": 0.00012059592158813588, + "step": 6666 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2075, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.2052494564049971e-06, + "loss": 0.7577, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.0001205249456404997, + "step": 6667 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2074, + "image_masks_loss": 0.011, + "llm_learning_rate": 1.204539839509449e-06, + "loss": 0.7039, + "output_text_mask_loss": 0.0595, + "predictor_learning_rate": 0.0001204539839509449, + "step": 6668 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2074, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.2038303652728587e-06, + "loss": 0.7095, + "output_text_mask_loss": 0.0385, + "predictor_learning_rate": 0.00012038303652728585, + "step": 6669 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2073, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.2031210337733565e-06, + "loss": 0.7234, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00012031210337733565, + "step": 6670 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2073, + "image_masks_loss": 0.0174, + "llm_learning_rate": 1.2024118450890545e-06, + "loss": 0.7289, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 0.00012024118450890544, + "step": 6671 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2073, + "image_masks_loss": 0.015, + "llm_learning_rate": 1.2017027992980507e-06, + "loss": 0.7953, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.00012017027992980505, + "step": 6672 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2072, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.2009938964784261e-06, + "loss": 0.7669, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 0.00012009938964784261, + "step": 6673 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2072, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.2002851367082475e-06, + "loss": 0.6914, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00012002851367082473, + "step": 6674 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2071, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.1995765200655647e-06, + "loss": 0.6679, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 0.00011995765200655646, + "step": 6675 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2071, + "image_masks_loss": 0.0214, + "llm_learning_rate": 1.1988680466284116e-06, + "loss": 0.6328, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.00011988680466284116, + "step": 6676 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.207, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.1981597164748077e-06, + "loss": 0.7424, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00011981597164748075, + "step": 6677 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.207, + "image_masks_loss": 0.005, + "llm_learning_rate": 1.1974515296827554e-06, + "loss": 0.7248, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.00011974515296827554, + "step": 6678 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2069, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.196743486330242e-06, + "loss": 0.7071, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00011967434863302418, + "step": 6679 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2069, + "image_masks_loss": 0.0236, + "llm_learning_rate": 1.196035586495239e-06, + "loss": 0.7158, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.00011960355864952388, + "step": 6680 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2068, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.1953278302557004e-06, + "loss": 0.6482, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011953278302557002, + "step": 6681 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2068, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.1946202176895682e-06, + "loss": 0.6714, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011946202176895682, + "step": 6682 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2067, + "image_masks_loss": 0.0027, + "llm_learning_rate": 1.193912748874765e-06, + "loss": 0.7239, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001193912748874765, + "step": 6683 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2067, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.1932054238891997e-06, + "loss": 0.6032, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00011932054238891995, + "step": 6684 + }, + { + "epoch": 0.68, + "gumbel_tau": 0.2066, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.1924982428107634e-06, + "loss": 0.6256, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00011924982428107633, + "step": 6685 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2066, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.1917912057173333e-06, + "loss": 0.7702, + "output_text_mask_loss": 0.0313, + "predictor_learning_rate": 0.00011917912057173333, + "step": 6686 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2065, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.1910843126867694e-06, + "loss": 0.7443, + "output_text_mask_loss": 0.0406, + "predictor_learning_rate": 0.00011910843126867694, + "step": 6687 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2065, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.1903775637969154e-06, + "loss": 0.8355, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00011903775637969152, + "step": 6688 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2064, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.1896709591256023e-06, + "loss": 0.745, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00011896709591256022, + "step": 6689 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2064, + "image_masks_loss": 0.0115, + "llm_learning_rate": 1.1889644987506416e-06, + "loss": 0.7529, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00011889644987506414, + "step": 6690 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2063, + "image_masks_loss": 0.0036, + "llm_learning_rate": 1.1882581827498307e-06, + "loss": 0.8074, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.00011882581827498307, + "step": 6691 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2063, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.1875520112009507e-06, + "loss": 0.7964, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00011875520112009505, + "step": 6692 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2062, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.1868459841817662e-06, + "loss": 0.7744, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011868459841817662, + "step": 6693 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2062, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.186140101770027e-06, + "loss": 0.7281, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00011861401017700269, + "step": 6694 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2061, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.1854343640434662e-06, + "loss": 0.7379, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00011854343640434661, + "step": 6695 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2061, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.1847287710798014e-06, + "loss": 0.7485, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 0.00011847287710798013, + "step": 6696 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.206, + "image_masks_loss": 0.016, + "llm_learning_rate": 1.1840233229567342e-06, + "loss": 0.6613, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00011840233229567341, + "step": 6697 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.206, + "image_masks_loss": 0.029, + "llm_learning_rate": 1.1833180197519495e-06, + "loss": 0.7014, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011833180197519494, + "step": 6698 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2059, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.1826128615431174e-06, + "loss": 0.7359, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00011826128615431172, + "step": 6699 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2059, + "image_masks_loss": 0.0066, + "llm_learning_rate": 1.1819078484078914e-06, + "loss": 0.8007, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 0.00011819078484078912, + "step": 6700 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2058, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.181202980423908e-06, + "loss": 0.6525, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0001181202980423908, + "step": 6701 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2058, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.1804982576687913e-06, + "loss": 0.7063, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00011804982576687912, + "step": 6702 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2057, + "image_masks_loss": 0.017, + "llm_learning_rate": 1.1797936802201452e-06, + "loss": 0.7614, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.0001179793680220145, + "step": 6703 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2057, + "image_masks_loss": 0.0144, + "llm_learning_rate": 1.1790892481555595e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011790892481555596, + "step": 6704 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2056, + "image_masks_loss": 0.0224, + "llm_learning_rate": 1.1783849615526083e-06, + "loss": 0.7484, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.00011783849615526082, + "step": 6705 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2056, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.1776808204888488e-06, + "loss": 0.7131, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 0.00011776808204888487, + "step": 6706 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2055, + "image_masks_loss": 0.019, + "llm_learning_rate": 1.1769768250418227e-06, + "loss": 0.7071, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 0.00011769768250418225, + "step": 6707 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2055, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.1762729752890544e-06, + "loss": 0.7482, + "output_text_mask_loss": 0.0441, + "predictor_learning_rate": 0.00011762729752890545, + "step": 6708 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2054, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.1755692713080558e-06, + "loss": 0.7044, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011755692713080557, + "step": 6709 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2054, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.1748657131763192e-06, + "loss": 0.6719, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011748657131763191, + "step": 6710 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2054, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.1741623009713216e-06, + "loss": 0.7219, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00011741623009713215, + "step": 6711 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2053, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.1734590347705246e-06, + "loss": 0.734, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 0.00011734590347705246, + "step": 6712 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2053, + "image_masks_loss": 0.0209, + "llm_learning_rate": 1.1727559146513736e-06, + "loss": 0.6287, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011727559146513736, + "step": 6713 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2052, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.1720529406912976e-06, + "loss": 0.7544, + "output_text_mask_loss": 0.0097, + "predictor_learning_rate": 0.00011720529406912975, + "step": 6714 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2052, + "image_masks_loss": 0.0169, + "llm_learning_rate": 1.1713501129677095e-06, + "loss": 0.7593, + "output_text_mask_loss": 0.0587, + "predictor_learning_rate": 0.00011713501129677096, + "step": 6715 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2051, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.1706474315580065e-06, + "loss": 0.6727, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00011706474315580065, + "step": 6716 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2051, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.1699448965395693e-06, + "loss": 0.653, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011699448965395692, + "step": 6717 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.205, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.169242507989763e-06, + "loss": 0.7081, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00011692425079897628, + "step": 6718 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.205, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.1685402659859355e-06, + "loss": 0.7695, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 0.00011685402659859354, + "step": 6719 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2049, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.1678381706054187e-06, + "loss": 0.7035, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00011678381706054186, + "step": 6720 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2049, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.1671362219255309e-06, + "loss": 0.798, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00011671362219255307, + "step": 6721 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2048, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.1664344200235711e-06, + "loss": 0.8615, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 0.0001166434420023571, + "step": 6722 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2048, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.1657327649768233e-06, + "loss": 0.7308, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 0.00011657327649768232, + "step": 6723 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2047, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.1650312568625554e-06, + "loss": 0.7396, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011650312568625554, + "step": 6724 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2047, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.1643298957580192e-06, + "loss": 0.6818, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00011643298957580192, + "step": 6725 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2046, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.1636286817404499e-06, + "loss": 0.7258, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011636286817404498, + "step": 6726 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2046, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.1629276148870658e-06, + "loss": 0.7104, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00011629276148870657, + "step": 6727 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2045, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.1622266952750719e-06, + "loss": 0.7791, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00011622266952750718, + "step": 6728 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2045, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.1615259229816543e-06, + "loss": 0.7597, + "output_text_mask_loss": 0.028, + "predictor_learning_rate": 0.00011615259229816541, + "step": 6729 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2044, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.1608252980839831e-06, + "loss": 0.6859, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00011608252980839832, + "step": 6730 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2044, + "image_masks_loss": 0.0029, + "llm_learning_rate": 1.1601248206592141e-06, + "loss": 0.7152, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 0.00011601248206592141, + "step": 6731 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2043, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.1594244907844831e-06, + "loss": 0.7067, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.0001159424490784483, + "step": 6732 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2043, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.1587243085369127e-06, + "loss": 0.8397, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011587243085369126, + "step": 6733 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2042, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.1580242739936095e-06, + "loss": 0.7068, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00011580242739936095, + "step": 6734 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2042, + "image_masks_loss": 0.0226, + "llm_learning_rate": 1.1573243872316628e-06, + "loss": 0.8322, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.00011573243872316627, + "step": 6735 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2041, + "image_masks_loss": 0.0257, + "llm_learning_rate": 1.1566246483281451e-06, + "loss": 0.7768, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00011566246483281451, + "step": 6736 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2041, + "image_masks_loss": 0.0157, + "llm_learning_rate": 1.1559250573601133e-06, + "loss": 0.7628, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 0.00011559250573601132, + "step": 6737 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.204, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.155225614404608e-06, + "loss": 0.7464, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 0.00011552256144046081, + "step": 6738 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.204, + "image_masks_loss": 0.004, + "llm_learning_rate": 1.1545263195386537e-06, + "loss": 0.6791, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011545263195386536, + "step": 6739 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.204, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.1538271728392567e-06, + "loss": 0.7259, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 0.00011538271728392566, + "step": 6740 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2039, + "image_masks_loss": 0.004, + "llm_learning_rate": 1.1531281743834109e-06, + "loss": 0.7251, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.00011531281743834107, + "step": 6741 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2039, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.1524293242480905e-06, + "loss": 0.7756, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00011524293242480904, + "step": 6742 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2038, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.1517306225102548e-06, + "loss": 0.7642, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 0.00011517306225102548, + "step": 6743 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2038, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.151032069246846e-06, + "loss": 0.769, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 0.0001151032069246846, + "step": 6744 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2037, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.1503336645347903e-06, + "loss": 0.6832, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00011503336645347903, + "step": 6745 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2037, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.1496354084509977e-06, + "loss": 0.672, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00011496354084509977, + "step": 6746 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2036, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.1489373010723612e-06, + "loss": 0.6621, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00011489373010723612, + "step": 6747 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2036, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.148239342475759e-06, + "loss": 0.7138, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 0.00011482393424757589, + "step": 6748 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2035, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.1475415327380515e-06, + "loss": 0.7186, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00011475415327380514, + "step": 6749 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2035, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.1468438719360838e-06, + "loss": 0.7106, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00011468438719360838, + "step": 6750 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2034, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.1461463601466821e-06, + "loss": 0.6483, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.0001146146360146682, + "step": 6751 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2034, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.1454489974466579e-06, + "loss": 0.7473, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 0.00011454489974466578, + "step": 6752 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2033, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.144751783912808e-06, + "loss": 0.7679, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.0001144751783912808, + "step": 6753 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2033, + "image_masks_loss": 0.0197, + "llm_learning_rate": 1.1440547196219104e-06, + "loss": 0.7514, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00011440547196219104, + "step": 6754 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2032, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.1433578046507273e-06, + "loss": 0.6714, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00011433578046507273, + "step": 6755 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2032, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.1426610390760048e-06, + "loss": 0.7886, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.00011426610390760048, + "step": 6756 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2031, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.1419644229744717e-06, + "loss": 0.6099, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.00011419644229744716, + "step": 6757 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2031, + "image_masks_loss": 0.0031, + "llm_learning_rate": 1.1412679564228416e-06, + "loss": 0.766, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011412679564228415, + "step": 6758 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.203, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.1405716394978102e-06, + "loss": 0.7126, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00011405716394978102, + "step": 6759 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.203, + "image_masks_loss": 0.0144, + "llm_learning_rate": 1.1398754722760572e-06, + "loss": 0.6908, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011398754722760571, + "step": 6760 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2029, + "image_masks_loss": 0.0135, + "llm_learning_rate": 1.1391794548342477e-06, + "loss": 0.7248, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00011391794548342476, + "step": 6761 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2029, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.1384835872490277e-06, + "loss": 0.735, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.00011384835872490277, + "step": 6762 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2028, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.1377878695970277e-06, + "loss": 0.7681, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 0.00011377878695970276, + "step": 6763 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2028, + "image_masks_loss": 0.0115, + "llm_learning_rate": 1.1370923019548615e-06, + "loss": 0.651, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00011370923019548615, + "step": 6764 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2028, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.1363968843991272e-06, + "loss": 0.7033, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 0.0001136396884399127, + "step": 6765 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2027, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.135701617006405e-06, + "loss": 0.8066, + "output_text_mask_loss": 0.029, + "predictor_learning_rate": 0.00011357016170064049, + "step": 6766 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2027, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.1350064998532587e-06, + "loss": 0.7305, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 0.00011350064998532586, + "step": 6767 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2026, + "image_masks_loss": 0.0209, + "llm_learning_rate": 1.1343115330162388e-06, + "loss": 0.7772, + "output_text_mask_loss": 0.0553, + "predictor_learning_rate": 0.00011343115330162387, + "step": 6768 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2026, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.1336167165718742e-06, + "loss": 0.7349, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 0.00011336167165718742, + "step": 6769 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2025, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.13292205059668e-06, + "loss": 0.6431, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.000113292205059668, + "step": 6770 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2025, + "image_masks_loss": 0.0016, + "llm_learning_rate": 1.1322275351671547e-06, + "loss": 0.7053, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00011322275351671546, + "step": 6771 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2024, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.131533170359779e-06, + "loss": 0.5931, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001131533170359779, + "step": 6772 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2024, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.1308389562510196e-06, + "loss": 0.7406, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 0.00011308389562510196, + "step": 6773 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2023, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.130144892917324e-06, + "loss": 0.7079, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 0.00011301448929173239, + "step": 6774 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2023, + "image_masks_loss": 0.0216, + "llm_learning_rate": 1.129450980435124e-06, + "loss": 0.7838, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 0.00011294509804351238, + "step": 6775 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2022, + "image_masks_loss": 0.0034, + "llm_learning_rate": 1.128757218880835e-06, + "loss": 0.6741, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00011287572188808348, + "step": 6776 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2022, + "image_masks_loss": 0.019, + "llm_learning_rate": 1.1280636083308553e-06, + "loss": 0.6905, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00011280636083308551, + "step": 6777 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2021, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.1273701488615665e-06, + "loss": 0.734, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 0.00011273701488615665, + "step": 6778 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2021, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.126676840549334e-06, + "loss": 0.6959, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 0.00011266768405493338, + "step": 6779 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.202, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.1259836834705074e-06, + "loss": 0.8082, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 0.00011259836834705073, + "step": 6780 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.202, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.125290677701418e-06, + "loss": 0.6447, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.0001125290677701418, + "step": 6781 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2019, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.1245978233183813e-06, + "loss": 0.796, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00011245978233183812, + "step": 6782 + }, + { + "epoch": 0.69, + "gumbel_tau": 0.2019, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.123905120397696e-06, + "loss": 0.6987, + "output_text_mask_loss": 0.0227, + "predictor_learning_rate": 0.0001123905120397696, + "step": 6783 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2018, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.123212569015644e-06, + "loss": 0.7851, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 0.00011232125690156439, + "step": 6784 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2018, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.1225201692484903e-06, + "loss": 0.732, + "output_text_mask_loss": 0.0206, + "predictor_learning_rate": 0.00011225201692484901, + "step": 6785 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2018, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.121827921172483e-06, + "loss": 0.7012, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 0.0001121827921172483, + "step": 6786 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2017, + "image_masks_loss": 0.0212, + "llm_learning_rate": 1.1211358248638568e-06, + "loss": 0.7482, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 0.00011211358248638568, + "step": 6787 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2017, + "image_masks_loss": 0.0168, + "llm_learning_rate": 1.1204438803988238e-06, + "loss": 0.6933, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011204438803988237, + "step": 6788 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2016, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.1197520878535833e-06, + "loss": 0.7885, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00011197520878535833, + "step": 6789 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2016, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.1190604473043175e-06, + "loss": 0.7294, + "output_text_mask_loss": 0.025, + "predictor_learning_rate": 0.00011190604473043176, + "step": 6790 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2015, + "image_masks_loss": 0.0029, + "llm_learning_rate": 1.1183689588271912e-06, + "loss": 0.7269, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 0.00011183689588271912, + "step": 6791 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2015, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.1176776224983519e-06, + "loss": 0.7017, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 0.00011176776224983518, + "step": 6792 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2014, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.1169864383939321e-06, + "loss": 0.7772, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001116986438393932, + "step": 6793 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2014, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.1162954065900468e-06, + "loss": 0.6989, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00011162954065900468, + "step": 6794 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2013, + "image_masks_loss": 0.004, + "llm_learning_rate": 1.1156045271627933e-06, + "loss": 0.7587, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00011156045271627932, + "step": 6795 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2013, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.1149138001882529e-06, + "loss": 0.7982, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00011149138001882528, + "step": 6796 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2012, + "image_masks_loss": 0.0334, + "llm_learning_rate": 1.1142232257424897e-06, + "loss": 0.7982, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 0.00011142232257424897, + "step": 6797 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2012, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.113532803901552e-06, + "loss": 0.6457, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011135328039015519, + "step": 6798 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2011, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.112842534741469e-06, + "loss": 0.7282, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 0.00011128425347414689, + "step": 6799 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2011, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.1121524183382568e-06, + "loss": 0.7948, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 0.00011121524183382567, + "step": 6800 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.201, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.1114624547679114e-06, + "loss": 0.742, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00011114624547679114, + "step": 6801 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.201, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.1107726441064137e-06, + "loss": 0.7761, + "output_text_mask_loss": 0.0456, + "predictor_learning_rate": 0.00011107726441064137, + "step": 6802 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2009, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.1100829864297266e-06, + "loss": 0.6315, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00011100829864297266, + "step": 6803 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2009, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.109393481813797e-06, + "loss": 0.7507, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00011093934818137969, + "step": 6804 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2008, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.1087041303345547e-06, + "loss": 0.7854, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00011087041303345546, + "step": 6805 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2008, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.1080149320679123e-06, + "loss": 0.6438, + "output_text_mask_loss": 0.0401, + "predictor_learning_rate": 0.00011080149320679123, + "step": 6806 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2008, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.1073258870897664e-06, + "loss": 0.6301, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.00011073258870897662, + "step": 6807 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2007, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.1066369954759956e-06, + "loss": 0.7504, + "output_text_mask_loss": 0.0452, + "predictor_learning_rate": 0.00011066369954759956, + "step": 6808 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2007, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.1059482573024627e-06, + "loss": 0.5836, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011059482573024626, + "step": 6809 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2006, + "image_masks_loss": 0.02, + "llm_learning_rate": 1.105259672645013e-06, + "loss": 0.7619, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 0.00011052596726450129, + "step": 6810 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2006, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.1045712415794734e-06, + "loss": 0.7536, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 0.00011045712415794734, + "step": 6811 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2005, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.103882964181658e-06, + "loss": 0.678, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.00011038829641816581, + "step": 6812 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2005, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.1031948405273604e-06, + "loss": 0.7463, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00011031948405273603, + "step": 6813 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2004, + "image_masks_loss": 0.0095, + "llm_learning_rate": 1.1025068706923583e-06, + "loss": 0.7842, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 0.00011025068706923583, + "step": 6814 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2004, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.1018190547524122e-06, + "loss": 0.7601, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00011018190547524121, + "step": 6815 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2003, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.1011313927832663e-06, + "loss": 0.711, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 0.00011011313927832662, + "step": 6816 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2003, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.1004438848606471e-06, + "loss": 0.7155, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00011004438848606471, + "step": 6817 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2002, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.0997565310602637e-06, + "loss": 0.8041, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.00010997565310602636, + "step": 6818 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2002, + "image_masks_loss": 0.02, + "llm_learning_rate": 1.0990693314578114e-06, + "loss": 0.7802, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010990693314578112, + "step": 6819 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2001, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.0983822861289642e-06, + "loss": 0.8082, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00010983822861289641, + "step": 6820 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2001, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.097695395149382e-06, + "loss": 0.7406, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00010976953951493818, + "step": 6821 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.0970086585947062e-06, + "loss": 0.7919, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00010970086585947061, + "step": 6822 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.0963220765405623e-06, + "loss": 0.6658, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.0001096322076540562, + "step": 6823 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.2, + "image_masks_loss": 0.0157, + "llm_learning_rate": 1.0956356490625574e-06, + "loss": 0.7493, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 0.00010956356490625575, + "step": 6824 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1999, + "image_masks_loss": 0.0045, + "llm_learning_rate": 1.094949376236283e-06, + "loss": 0.7466, + "output_text_mask_loss": 0.0226, + "predictor_learning_rate": 0.00010949493762362831, + "step": 6825 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1999, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.0942632581373133e-06, + "loss": 0.7706, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00010942632581373132, + "step": 6826 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1998, + "image_masks_loss": 0.0125, + "llm_learning_rate": 1.0935772948412045e-06, + "loss": 0.6436, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00010935772948412045, + "step": 6827 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1998, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.0928914864234965e-06, + "loss": 0.7506, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00010928914864234965, + "step": 6828 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1997, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.0922058329597128e-06, + "loss": 0.6151, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00010922058329597126, + "step": 6829 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1997, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.0915203345253581e-06, + "loss": 0.7709, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 0.0001091520334525358, + "step": 6830 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1996, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.0908349911959204e-06, + "loss": 0.6835, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010908349911959204, + "step": 6831 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1996, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.0901498030468736e-06, + "loss": 0.7476, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 0.00010901498030468734, + "step": 6832 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1995, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.0894647701536704e-06, + "loss": 0.7377, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010894647701536704, + "step": 6833 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1995, + "image_masks_loss": 0.011, + "llm_learning_rate": 1.0887798925917486e-06, + "loss": 0.6929, + "output_text_mask_loss": 0.008, + "predictor_learning_rate": 0.00010887798925917486, + "step": 6834 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1994, + "image_masks_loss": 0.003, + "llm_learning_rate": 1.0880951704365287e-06, + "loss": 0.6947, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 0.00010880951704365285, + "step": 6835 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1994, + "image_masks_loss": 0.0155, + "llm_learning_rate": 1.0874106037634132e-06, + "loss": 0.7301, + "output_text_mask_loss": 0.0885, + "predictor_learning_rate": 0.00010874106037634132, + "step": 6836 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1993, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.0867261926477888e-06, + "loss": 0.7609, + "output_text_mask_loss": 0.0135, + "predictor_learning_rate": 0.00010867261926477887, + "step": 6837 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1993, + "image_masks_loss": 0.0144, + "llm_learning_rate": 1.0860419371650228e-06, + "loss": 0.7177, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010860419371650226, + "step": 6838 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1992, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.0853578373904691e-06, + "loss": 0.7035, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.0001085357837390469, + "step": 6839 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1992, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.0846738933994614e-06, + "loss": 0.7563, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010846738933994612, + "step": 6840 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1992, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.083990105267317e-06, + "loss": 0.7663, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 0.00010839901052673168, + "step": 6841 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1991, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.0833064730693364e-06, + "loss": 0.764, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00010833064730693363, + "step": 6842 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1991, + "image_masks_loss": 0.0127, + "llm_learning_rate": 1.0826229968808022e-06, + "loss": 0.7183, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00010826229968808021, + "step": 6843 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.199, + "image_masks_loss": 0.0174, + "llm_learning_rate": 1.081939676776981e-06, + "loss": 0.7255, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 0.0001081939676776981, + "step": 6844 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.199, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.081256512833121e-06, + "loss": 0.6795, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 0.00010812565128331208, + "step": 6845 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1989, + "image_masks_loss": 0.0234, + "llm_learning_rate": 1.080573505124454e-06, + "loss": 0.79, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00010805735051244539, + "step": 6846 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1989, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.0798906537261945e-06, + "loss": 0.7625, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00010798906537261943, + "step": 6847 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1988, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.079207958713539e-06, + "loss": 0.752, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.0001079207958713539, + "step": 6848 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1988, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.078525420161668e-06, + "loss": 0.7437, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00010785254201616678, + "step": 6849 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1987, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.077843038145743e-06, + "loss": 0.737, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 0.0001077843038145743, + "step": 6850 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1987, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.0771608127409113e-06, + "loss": 0.7995, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 0.00010771608127409113, + "step": 6851 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1986, + "image_masks_loss": 0.0157, + "llm_learning_rate": 1.0764787440223002e-06, + "loss": 0.6511, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 0.00010764787440223, + "step": 6852 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1986, + "image_masks_loss": 0.0045, + "llm_learning_rate": 1.0757968320650202e-06, + "loss": 0.7362, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 0.00010757968320650203, + "step": 6853 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1985, + "image_masks_loss": 0.0123, + "llm_learning_rate": 1.075115076944166e-06, + "loss": 0.7588, + "output_text_mask_loss": 0.0824, + "predictor_learning_rate": 0.00010751150769441659, + "step": 6854 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1985, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.0744334787348129e-06, + "loss": 0.7431, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00010744334787348129, + "step": 6855 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1984, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.0737520375120206e-06, + "loss": 0.7335, + "output_text_mask_loss": 0.0169, + "predictor_learning_rate": 0.00010737520375120205, + "step": 6856 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1984, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.0730707533508297e-06, + "loss": 0.6879, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.00010730707533508296, + "step": 6857 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1984, + "image_masks_loss": 0.017, + "llm_learning_rate": 1.0723896263262667e-06, + "loss": 0.7916, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010723896263262667, + "step": 6858 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1983, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.071708656513338e-06, + "loss": 0.674, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.0001071708656513338, + "step": 6859 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1983, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.0710278439870336e-06, + "loss": 0.7686, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 0.00010710278439870336, + "step": 6860 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1982, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.0703471888223265e-06, + "loss": 0.7885, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 0.00010703471888223265, + "step": 6861 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1982, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.0696666910941709e-06, + "loss": 0.7577, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 0.00010696666910941708, + "step": 6862 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1981, + "image_masks_loss": 0.0248, + "llm_learning_rate": 1.068986350877504e-06, + "loss": 0.7377, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 0.00010689863508775038, + "step": 6863 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1981, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.0683061682472486e-06, + "loss": 0.689, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00010683061682472486, + "step": 6864 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.198, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.0676261432783069e-06, + "loss": 0.6622, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00010676261432783069, + "step": 6865 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.198, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.066946276045565e-06, + "loss": 0.6808, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00010669462760455648, + "step": 6866 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1979, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.066266566623891e-06, + "loss": 0.7759, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.00010662665666238908, + "step": 6867 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1979, + "image_masks_loss": 0.0167, + "llm_learning_rate": 1.065587015088136e-06, + "loss": 0.7394, + "output_text_mask_loss": 0.0334, + "predictor_learning_rate": 0.00010655870150881358, + "step": 6868 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1978, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.064907621513134e-06, + "loss": 0.7846, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00010649076215131339, + "step": 6869 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1978, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.0642283859737006e-06, + "loss": 0.6537, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 0.00010642283859737006, + "step": 6870 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1977, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.0635493085446363e-06, + "loss": 0.7151, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 0.00010635493085446362, + "step": 6871 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1977, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.0628703893007213e-06, + "loss": 0.8257, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.00010628703893007213, + "step": 6872 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1977, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.0621916283167205e-06, + "loss": 0.7327, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00010621916283167205, + "step": 6873 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1976, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.0615130256673803e-06, + "loss": 0.7365, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 0.00010615130256673803, + "step": 6874 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1976, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.0608345814274296e-06, + "loss": 0.7289, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 0.00010608345814274295, + "step": 6875 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1975, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.0601562956715802e-06, + "loss": 0.794, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.00010601562956715802, + "step": 6876 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1975, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.0594781684745263e-06, + "loss": 0.7419, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010594781684745261, + "step": 6877 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1974, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.0588001999109453e-06, + "loss": 0.7991, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 0.00010588001999109454, + "step": 6878 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1974, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.0581223900554972e-06, + "loss": 0.7225, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00010581223900554971, + "step": 6879 + }, + { + "epoch": 0.7, + "gumbel_tau": 0.1973, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.0574447389828235e-06, + "loss": 0.7561, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 0.00010574447389828235, + "step": 6880 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1973, + "image_masks_loss": 0.0066, + "llm_learning_rate": 1.0567672467675473e-06, + "loss": 0.7499, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00010567672467675474, + "step": 6881 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1972, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.0560899134842761e-06, + "loss": 0.7336, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001056089913484276, + "step": 6882 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1972, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.0554127392076008e-06, + "loss": 0.7113, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 0.00010554127392076007, + "step": 6883 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1971, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.054735724012092e-06, + "loss": 0.7052, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 0.00010547357240120919, + "step": 6884 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1971, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.0540588679723046e-06, + "loss": 0.6995, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00010540588679723045, + "step": 6885 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.197, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.0533821711627753e-06, + "loss": 0.7792, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00010533821711627753, + "step": 6886 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.197, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.0527056336580236e-06, + "loss": 0.7375, + "output_text_mask_loss": 0.0047, + "predictor_learning_rate": 0.00010527056336580235, + "step": 6887 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.197, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.0520292555325512e-06, + "loss": 0.8576, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00010520292555325512, + "step": 6888 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1969, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.0513530368608416e-06, + "loss": 0.6823, + "output_text_mask_loss": 0.0431, + "predictor_learning_rate": 0.00010513530368608415, + "step": 6889 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1969, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.0506769777173632e-06, + "loss": 0.668, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 0.00010506769777173631, + "step": 6890 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1968, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.050001078176564e-06, + "loss": 0.7066, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 0.0001050001078176564, + "step": 6891 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1968, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.0493253383128763e-06, + "loss": 0.7297, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 0.00010493253383128762, + "step": 6892 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1967, + "image_masks_loss": 0.0244, + "llm_learning_rate": 1.0486497582007132e-06, + "loss": 0.7663, + "output_text_mask_loss": 0.053, + "predictor_learning_rate": 0.00010486497582007132, + "step": 6893 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1967, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.0479743379144715e-06, + "loss": 0.7569, + "output_text_mask_loss": 0.008, + "predictor_learning_rate": 0.00010479743379144716, + "step": 6894 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1966, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.0472990775285302e-06, + "loss": 0.7755, + "output_text_mask_loss": 0.0633, + "predictor_learning_rate": 0.00010472990775285302, + "step": 6895 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1966, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.046623977117249e-06, + "loss": 0.7045, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 0.00010466239771172489, + "step": 6896 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1965, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.045949036754974e-06, + "loss": 0.7458, + "output_text_mask_loss": 0.0549, + "predictor_learning_rate": 0.0001045949036754974, + "step": 6897 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1965, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.0452742565160304e-06, + "loss": 0.6596, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010452742565160303, + "step": 6898 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1964, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.0445996364747254e-06, + "loss": 0.7214, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00010445996364747254, + "step": 6899 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1964, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.0439251767053499e-06, + "loss": 0.7531, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 0.00010439251767053498, + "step": 6900 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1964, + "image_masks_loss": 0.0036, + "llm_learning_rate": 1.0432508772821772e-06, + "loss": 0.7031, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00010432508772821772, + "step": 6901 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1963, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.042576738279462e-06, + "loss": 0.7502, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00010425767382794621, + "step": 6902 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1963, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.0419027597714434e-06, + "loss": 0.785, + "output_text_mask_loss": 0.0045, + "predictor_learning_rate": 0.00010419027597714434, + "step": 6903 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1962, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.0412289418323409e-06, + "loss": 0.7358, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 0.00010412289418323409, + "step": 6904 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1962, + "image_masks_loss": 0.003, + "llm_learning_rate": 1.0405552845363565e-06, + "loss": 0.7284, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010405552845363563, + "step": 6905 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1961, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.039881787957675e-06, + "loss": 0.7612, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 0.00010398817879576749, + "step": 6906 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1961, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.0392084521704632e-06, + "loss": 0.788, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010392084521704632, + "step": 6907 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.196, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.0385352772488704e-06, + "loss": 0.7467, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 0.00010385352772488705, + "step": 6908 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.196, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.0378622632670274e-06, + "loss": 0.7161, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00010378622632670273, + "step": 6909 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1959, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.0371894102990496e-06, + "loss": 0.7245, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00010371894102990494, + "step": 6910 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1959, + "image_masks_loss": 0.0147, + "llm_learning_rate": 1.0365167184190319e-06, + "loss": 0.7805, + "output_text_mask_loss": 0.0128, + "predictor_learning_rate": 0.00010365167184190318, + "step": 6911 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1958, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.0358441877010534e-06, + "loss": 0.7577, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 0.00010358441877010532, + "step": 6912 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1958, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.0351718182191741e-06, + "loss": 0.7233, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 0.00010351718182191742, + "step": 6913 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1957, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.0344996100474369e-06, + "loss": 0.7392, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00010344996100474368, + "step": 6914 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1957, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.0338275632598668e-06, + "loss": 0.713, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00010338275632598668, + "step": 6915 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1957, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.0331556779304705e-06, + "loss": 0.7386, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 0.00010331556779304705, + "step": 6916 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1956, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.0324839541332399e-06, + "loss": 0.7227, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 0.00010324839541332398, + "step": 6917 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1956, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.0318123919421443e-06, + "loss": 0.7411, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 0.00010318123919421443, + "step": 6918 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1955, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.0311409914311382e-06, + "loss": 0.7516, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 0.00010311409914311382, + "step": 6919 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1955, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.030469752674158e-06, + "loss": 0.6703, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.0001030469752674158, + "step": 6920 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1954, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.029798675745121e-06, + "loss": 0.7914, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00010297986757451209, + "step": 6921 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1954, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.0291277607179292e-06, + "loss": 0.7442, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 0.00010291277607179292, + "step": 6922 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1953, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.028457007666465e-06, + "loss": 0.726, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 0.00010284570076664649, + "step": 6923 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1953, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.0277864166645924e-06, + "loss": 0.7148, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00010277864166645922, + "step": 6924 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1952, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.0271159877861592e-06, + "loss": 0.6542, + "output_text_mask_loss": 0.0165, + "predictor_learning_rate": 0.00010271159877861591, + "step": 6925 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1952, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.026445721104994e-06, + "loss": 0.8141, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 0.00010264457211049938, + "step": 6926 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1952, + "image_masks_loss": 0.0139, + "llm_learning_rate": 1.0257756166949082e-06, + "loss": 0.7549, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 0.00010257756166949081, + "step": 6927 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1951, + "image_masks_loss": 0.0022, + "llm_learning_rate": 1.0251056746296944e-06, + "loss": 0.7594, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 0.00010251056746296944, + "step": 6928 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1951, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.0244358949831296e-06, + "loss": 0.8474, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00010244358949831295, + "step": 6929 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.195, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.0237662778289708e-06, + "loss": 0.7181, + "output_text_mask_loss": 0.0234, + "predictor_learning_rate": 0.00010237662778289707, + "step": 6930 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.195, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.023096823240958e-06, + "loss": 0.753, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 0.0001023096823240958, + "step": 6931 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1949, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.0224275312928123e-06, + "loss": 0.7265, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00010224275312928122, + "step": 6932 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1949, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.0217584020582384e-06, + "loss": 0.7886, + "output_text_mask_loss": 0.0148, + "predictor_learning_rate": 0.00010217584020582382, + "step": 6933 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1948, + "image_masks_loss": 0.0017, + "llm_learning_rate": 1.0210894356109214e-06, + "loss": 0.7735, + "output_text_mask_loss": 0.0854, + "predictor_learning_rate": 0.00010210894356109213, + "step": 6934 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1948, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.0204206320245302e-06, + "loss": 0.7729, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 0.00010204206320245302, + "step": 6935 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1947, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.0197519913727146e-06, + "loss": 0.7176, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 0.00010197519913727146, + "step": 6936 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1947, + "image_masks_loss": 0.0036, + "llm_learning_rate": 1.0190835137291068e-06, + "loss": 0.6922, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010190835137291068, + "step": 6937 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1946, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.018415199167321e-06, + "loss": 0.6905, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.0001018415199167321, + "step": 6938 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1946, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.0177470477609538e-06, + "loss": 0.7652, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00010177470477609537, + "step": 6939 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1946, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.0170790595835834e-06, + "loss": 0.7603, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00010170790595835832, + "step": 6940 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1945, + "image_masks_loss": 0.005, + "llm_learning_rate": 1.0164112347087688e-06, + "loss": 0.6358, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010164112347087687, + "step": 6941 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1945, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.0157435732100545e-06, + "loss": 0.7631, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 0.00010157435732100545, + "step": 6942 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1944, + "image_masks_loss": 0.0047, + "llm_learning_rate": 1.0150760751609642e-06, + "loss": 0.6827, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 0.00010150760751609642, + "step": 6943 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1944, + "image_masks_loss": 0.0025, + "llm_learning_rate": 1.014408740635004e-06, + "loss": 0.6673, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 0.00010144087406350039, + "step": 6944 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1943, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.0137415697056621e-06, + "loss": 0.7479, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 0.0001013741569705662, + "step": 6945 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1943, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.0130745624464092e-06, + "loss": 0.6658, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 0.00010130745624464091, + "step": 6946 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1942, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.0124077189306974e-06, + "loss": 0.7734, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 0.00010124077189306974, + "step": 6947 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1942, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.01174103923196e-06, + "loss": 0.7153, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010117410392319601, + "step": 6948 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1941, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.0110745234236157e-06, + "loss": 0.7407, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 0.00010110745234236157, + "step": 6949 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1941, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.0104081715790611e-06, + "loss": 0.6562, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00010104081715790611, + "step": 6950 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.194, + "image_masks_loss": 0.017, + "llm_learning_rate": 1.0097419837716766e-06, + "loss": 0.7135, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 0.00010097419837716765, + "step": 6951 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.194, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.0090759600748242e-06, + "loss": 0.79, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00010090759600748242, + "step": 6952 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.194, + "image_masks_loss": 0.0037, + "llm_learning_rate": 1.008410100561848e-06, + "loss": 0.6284, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 0.00010084101005618479, + "step": 6953 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1939, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.007744405306074e-06, + "loss": 0.7447, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 0.00010077444053060739, + "step": 6954 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1939, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.00707887438081e-06, + "loss": 0.6786, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00010070788743808098, + "step": 6955 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1938, + "image_masks_loss": 0.02, + "llm_learning_rate": 1.0064135078593454e-06, + "loss": 0.6645, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 0.00010064135078593453, + "step": 6956 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1938, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.0057483058149526e-06, + "loss": 0.7082, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 0.00010057483058149524, + "step": 6957 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1937, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.0050832683208842e-06, + "loss": 0.8014, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.0001005083268320884, + "step": 6958 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1937, + "image_masks_loss": 0.0114, + "llm_learning_rate": 1.0044183954503763e-06, + "loss": 0.7156, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 0.00010044183954503764, + "step": 6959 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1936, + "image_masks_loss": 0.018, + "llm_learning_rate": 1.003753687276645e-06, + "loss": 0.8016, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 0.0001003753687276645, + "step": 6960 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1936, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.0030891438728915e-06, + "loss": 0.7363, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.00010030891438728914, + "step": 6961 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1935, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.0024247653122956e-06, + "loss": 0.7339, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 0.00010024247653122956, + "step": 6962 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1935, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.0017605516680204e-06, + "loss": 0.7393, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 0.00010017605516680203, + "step": 6963 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1935, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.0010965030132103e-06, + "loss": 0.6882, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 0.00010010965030132103, + "step": 6964 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1934, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.0004326194209921e-06, + "loss": 0.7515, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 0.0001000432619420992, + "step": 6965 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1934, + "image_masks_loss": 0.012, + "llm_learning_rate": 9.997689009644736e-07, + "loss": 0.7332, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 9.997689009644736e-05, + "step": 6966 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1933, + "image_masks_loss": 0.0092, + "llm_learning_rate": 9.99105347716745e-07, + "loss": 0.7558, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 9.99105347716745e-05, + "step": 6967 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1933, + "image_masks_loss": 0.0124, + "llm_learning_rate": 9.984419597508794e-07, + "loss": 0.7037, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 9.984419597508793e-05, + "step": 6968 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1932, + "image_masks_loss": 0.0051, + "llm_learning_rate": 9.977787371399295e-07, + "loss": 0.7677, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 9.977787371399294e-05, + "step": 6969 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1932, + "image_masks_loss": 0.0075, + "llm_learning_rate": 9.971156799569312e-07, + "loss": 0.7601, + "output_text_mask_loss": 0.0192, + "predictor_learning_rate": 9.971156799569312e-05, + "step": 6970 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1931, + "image_masks_loss": 0.0149, + "llm_learning_rate": 9.96452788274902e-07, + "loss": 0.8193, + "output_text_mask_loss": 0.042, + "predictor_learning_rate": 9.964527882749019e-05, + "step": 6971 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1931, + "image_masks_loss": 0.006, + "llm_learning_rate": 9.957900621668416e-07, + "loss": 0.6926, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 9.957900621668414e-05, + "step": 6972 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.193, + "image_masks_loss": 0.0216, + "llm_learning_rate": 9.951275017057279e-07, + "loss": 0.8123, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 9.951275017057279e-05, + "step": 6973 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.193, + "image_masks_loss": 0.0023, + "llm_learning_rate": 9.94465106964527e-07, + "loss": 0.7393, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 9.944651069645269e-05, + "step": 6974 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.193, + "image_masks_loss": 0.0079, + "llm_learning_rate": 9.93802878016182e-07, + "loss": 0.8219, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 9.938028780161818e-05, + "step": 6975 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1929, + "image_masks_loss": 0.0036, + "llm_learning_rate": 9.931408149336187e-07, + "loss": 0.7164, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 9.931408149336185e-05, + "step": 6976 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1929, + "image_masks_loss": 0.0034, + "llm_learning_rate": 9.924789177897455e-07, + "loss": 0.7142, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 9.924789177897453e-05, + "step": 6977 + }, + { + "epoch": 0.71, + "gumbel_tau": 0.1928, + "image_masks_loss": 0.0108, + "llm_learning_rate": 9.918171866574514e-07, + "loss": 0.7573, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 9.918171866574513e-05, + "step": 6978 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1928, + "image_masks_loss": 0.0052, + "llm_learning_rate": 9.911556216096083e-07, + "loss": 0.5911, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 9.911556216096082e-05, + "step": 6979 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1927, + "image_masks_loss": 0.0116, + "llm_learning_rate": 9.904942227190682e-07, + "loss": 0.755, + "output_text_mask_loss": 0.0212, + "predictor_learning_rate": 9.90494222719068e-05, + "step": 6980 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1927, + "image_masks_loss": 0.0097, + "llm_learning_rate": 9.898329900586673e-07, + "loss": 0.7942, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 9.898329900586672e-05, + "step": 6981 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1926, + "image_masks_loss": 0.0188, + "llm_learning_rate": 9.891719237012212e-07, + "loss": 0.7375, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 9.891719237012211e-05, + "step": 6982 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1926, + "image_masks_loss": 0.0061, + "llm_learning_rate": 9.885110237195282e-07, + "loss": 0.6863, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 9.885110237195283e-05, + "step": 6983 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1925, + "image_masks_loss": 0.0128, + "llm_learning_rate": 9.878502901863682e-07, + "loss": 0.7248, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 9.878502901863681e-05, + "step": 6984 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1925, + "image_masks_loss": 0.009, + "llm_learning_rate": 9.871897231745023e-07, + "loss": 0.6487, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 9.871897231745023e-05, + "step": 6985 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1925, + "image_masks_loss": 0.0147, + "llm_learning_rate": 9.865293227566737e-07, + "loss": 0.7551, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 9.865293227566735e-05, + "step": 6986 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1924, + "image_masks_loss": 0.0033, + "llm_learning_rate": 9.85869089005606e-07, + "loss": 0.7185, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.85869089005606e-05, + "step": 6987 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1924, + "image_masks_loss": 0.0154, + "llm_learning_rate": 9.852090219940081e-07, + "loss": 0.7397, + "output_text_mask_loss": 0.0169, + "predictor_learning_rate": 9.85209021994008e-05, + "step": 6988 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1923, + "image_masks_loss": 0.0095, + "llm_learning_rate": 9.845491217945665e-07, + "loss": 0.7913, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 9.845491217945665e-05, + "step": 6989 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1923, + "image_masks_loss": 0.0154, + "llm_learning_rate": 9.838893884799514e-07, + "loss": 0.7206, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 9.838893884799514e-05, + "step": 6990 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1922, + "image_masks_loss": 0.0079, + "llm_learning_rate": 9.832298221228148e-07, + "loss": 0.7474, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.832298221228148e-05, + "step": 6991 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1922, + "image_masks_loss": 0.0127, + "llm_learning_rate": 9.825704227957874e-07, + "loss": 0.7076, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 9.825704227957874e-05, + "step": 6992 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1921, + "image_masks_loss": 0.0069, + "llm_learning_rate": 9.819111905714845e-07, + "loss": 0.7661, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 9.819111905714845e-05, + "step": 6993 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1921, + "image_masks_loss": 0.0422, + "llm_learning_rate": 9.812521255225032e-07, + "loss": 0.7964, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 9.812521255225032e-05, + "step": 6994 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.192, + "image_masks_loss": 0.0046, + "llm_learning_rate": 9.805932277214212e-07, + "loss": 0.7522, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 9.805932277214211e-05, + "step": 6995 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.192, + "image_masks_loss": 0.0129, + "llm_learning_rate": 9.799344972407967e-07, + "loss": 0.7787, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 9.799344972407967e-05, + "step": 6996 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.192, + "image_masks_loss": 0.0074, + "llm_learning_rate": 9.792759341531715e-07, + "loss": 0.7039, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 9.792759341531716e-05, + "step": 6997 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1919, + "image_masks_loss": 0.013, + "llm_learning_rate": 9.786175385310678e-07, + "loss": 0.6416, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 9.786175385310677e-05, + "step": 6998 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1919, + "image_masks_loss": 0.0119, + "llm_learning_rate": 9.779593104469894e-07, + "loss": 0.7443, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 9.779593104469895e-05, + "step": 6999 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1918, + "image_masks_loss": 0.009, + "llm_learning_rate": 9.77301249973421e-07, + "loss": 0.6644, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 9.77301249973421e-05, + "step": 7000 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1918, + "image_masks_loss": 0.0079, + "llm_learning_rate": 9.766433571828318e-07, + "loss": 0.6986, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 9.766433571828318e-05, + "step": 7001 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1917, + "image_masks_loss": 0.0171, + "llm_learning_rate": 9.75985632147669e-07, + "loss": 0.7078, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 9.75985632147669e-05, + "step": 7002 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1917, + "image_masks_loss": 0.0058, + "llm_learning_rate": 9.753280749403628e-07, + "loss": 0.7401, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 9.753280749403629e-05, + "step": 7003 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1916, + "image_masks_loss": 0.0109, + "llm_learning_rate": 9.746706856333255e-07, + "loss": 0.7292, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.746706856333254e-05, + "step": 7004 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1916, + "image_masks_loss": 0.0077, + "llm_learning_rate": 9.740134642989495e-07, + "loss": 0.7086, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 9.740134642989495e-05, + "step": 7005 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1915, + "image_masks_loss": 0.0104, + "llm_learning_rate": 9.7335641100961e-07, + "loss": 0.7932, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 9.733564110096099e-05, + "step": 7006 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1915, + "image_masks_loss": 0.0087, + "llm_learning_rate": 9.726995258376621e-07, + "loss": 0.7538, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.72699525837662e-05, + "step": 7007 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1915, + "image_masks_loss": 0.013, + "llm_learning_rate": 9.72042808855445e-07, + "loss": 0.7783, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 9.72042808855445e-05, + "step": 7008 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1914, + "image_masks_loss": 0.0309, + "llm_learning_rate": 9.713862601352784e-07, + "loss": 0.766, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 9.713862601352783e-05, + "step": 7009 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1914, + "image_masks_loss": 0.0218, + "llm_learning_rate": 9.707298797494604e-07, + "loss": 0.701, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 9.707298797494604e-05, + "step": 7010 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1913, + "image_masks_loss": 0.0125, + "llm_learning_rate": 9.70073667770275e-07, + "loss": 0.7261, + "output_text_mask_loss": 0.0057, + "predictor_learning_rate": 9.700736677702748e-05, + "step": 7011 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1913, + "image_masks_loss": 0.0082, + "llm_learning_rate": 9.694176242699835e-07, + "loss": 0.764, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 9.694176242699835e-05, + "step": 7012 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1912, + "image_masks_loss": 0.0081, + "llm_learning_rate": 9.687617493208336e-07, + "loss": 0.6802, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 9.687617493208337e-05, + "step": 7013 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1912, + "image_masks_loss": 0.0081, + "llm_learning_rate": 9.681060429950507e-07, + "loss": 0.8074, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 9.681060429950505e-05, + "step": 7014 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1911, + "image_masks_loss": 0.0173, + "llm_learning_rate": 9.674505053648425e-07, + "loss": 0.6906, + "output_text_mask_loss": 0.1312, + "predictor_learning_rate": 9.674505053648424e-05, + "step": 7015 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1911, + "image_masks_loss": 0.0165, + "llm_learning_rate": 9.66795136502398e-07, + "loss": 0.8012, + "output_text_mask_loss": 0.0439, + "predictor_learning_rate": 9.66795136502398e-05, + "step": 7016 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1911, + "image_masks_loss": 0.0072, + "llm_learning_rate": 9.661399364798884e-07, + "loss": 0.7307, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 9.661399364798885e-05, + "step": 7017 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.191, + "image_masks_loss": 0.011, + "llm_learning_rate": 9.654849053694653e-07, + "loss": 0.764, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 9.654849053694654e-05, + "step": 7018 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.191, + "image_masks_loss": 0.0077, + "llm_learning_rate": 9.64830043243262e-07, + "loss": 0.6198, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.64830043243262e-05, + "step": 7019 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1909, + "image_masks_loss": 0.0091, + "llm_learning_rate": 9.641753501733942e-07, + "loss": 0.7341, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 9.641753501733941e-05, + "step": 7020 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1909, + "image_masks_loss": 0.0061, + "llm_learning_rate": 9.635208262319578e-07, + "loss": 0.6797, + "output_text_mask_loss": 0.0542, + "predictor_learning_rate": 9.635208262319578e-05, + "step": 7021 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1908, + "image_masks_loss": 0.0044, + "llm_learning_rate": 9.628664714910307e-07, + "loss": 0.8387, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 9.628664714910307e-05, + "step": 7022 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1908, + "image_masks_loss": 0.0201, + "llm_learning_rate": 9.622122860226714e-07, + "loss": 0.8133, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 9.622122860226715e-05, + "step": 7023 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1907, + "image_masks_loss": 0.0046, + "llm_learning_rate": 9.615582698989204e-07, + "loss": 0.8304, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 9.615582698989203e-05, + "step": 7024 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1907, + "image_masks_loss": 0.0115, + "llm_learning_rate": 9.609044231917996e-07, + "loss": 0.7609, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 9.609044231917996e-05, + "step": 7025 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1906, + "image_masks_loss": 0.0047, + "llm_learning_rate": 9.60250745973311e-07, + "loss": 0.7853, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 9.60250745973311e-05, + "step": 7026 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1906, + "image_masks_loss": 0.0075, + "llm_learning_rate": 9.595972383154406e-07, + "loss": 0.6943, + "output_text_mask_loss": 0.0317, + "predictor_learning_rate": 9.595972383154405e-05, + "step": 7027 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1906, + "image_masks_loss": 0.0026, + "llm_learning_rate": 9.589439002901546e-07, + "loss": 0.7074, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 9.589439002901546e-05, + "step": 7028 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1905, + "image_masks_loss": 0.0132, + "llm_learning_rate": 9.582907319693976e-07, + "loss": 0.6668, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 9.582907319693975e-05, + "step": 7029 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1905, + "image_masks_loss": 0.0148, + "llm_learning_rate": 9.576377334250997e-07, + "loss": 0.6979, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 9.576377334250997e-05, + "step": 7030 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1904, + "image_masks_loss": 0.0053, + "llm_learning_rate": 9.569849047291698e-07, + "loss": 0.6934, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.569849047291698e-05, + "step": 7031 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1904, + "image_masks_loss": 0.0109, + "llm_learning_rate": 9.563322459534982e-07, + "loss": 0.8026, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.563322459534982e-05, + "step": 7032 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1903, + "image_masks_loss": 0.0215, + "llm_learning_rate": 9.556797571699592e-07, + "loss": 0.7599, + "output_text_mask_loss": 0.074, + "predictor_learning_rate": 9.55679757169959e-05, + "step": 7033 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1903, + "image_masks_loss": 0.0226, + "llm_learning_rate": 9.550274384504049e-07, + "loss": 0.7273, + "output_text_mask_loss": 0.0402, + "predictor_learning_rate": 9.550274384504049e-05, + "step": 7034 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1902, + "image_masks_loss": 0.0148, + "llm_learning_rate": 9.543752898666705e-07, + "loss": 0.7284, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 9.543752898666705e-05, + "step": 7035 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1902, + "image_masks_loss": 0.0051, + "llm_learning_rate": 9.537233114905719e-07, + "loss": 0.7631, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.537233114905719e-05, + "step": 7036 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1902, + "image_masks_loss": 0.0106, + "llm_learning_rate": 9.530715033939064e-07, + "loss": 0.7529, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.530715033939064e-05, + "step": 7037 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1901, + "image_masks_loss": 0.0118, + "llm_learning_rate": 9.524198656484526e-07, + "loss": 0.7835, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 9.524198656484526e-05, + "step": 7038 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1901, + "image_masks_loss": 0.0068, + "llm_learning_rate": 9.517683983259696e-07, + "loss": 0.7145, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 9.517683983259695e-05, + "step": 7039 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.19, + "image_masks_loss": 0.0105, + "llm_learning_rate": 9.511171014981996e-07, + "loss": 0.7536, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 9.511171014981995e-05, + "step": 7040 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.19, + "image_masks_loss": 0.0091, + "llm_learning_rate": 9.504659752368647e-07, + "loss": 0.8121, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 9.504659752368647e-05, + "step": 7041 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1899, + "image_masks_loss": 0.0081, + "llm_learning_rate": 9.498150196136679e-07, + "loss": 0.6759, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 9.498150196136679e-05, + "step": 7042 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1899, + "image_masks_loss": 0.0123, + "llm_learning_rate": 9.491642347002941e-07, + "loss": 0.7387, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.491642347002941e-05, + "step": 7043 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1898, + "image_masks_loss": 0.018, + "llm_learning_rate": 9.485136205684093e-07, + "loss": 0.7987, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 9.485136205684092e-05, + "step": 7044 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1898, + "image_masks_loss": 0.0301, + "llm_learning_rate": 9.478631772896602e-07, + "loss": 0.7732, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 9.4786317728966e-05, + "step": 7045 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1897, + "image_masks_loss": 0.0237, + "llm_learning_rate": 9.472129049356754e-07, + "loss": 0.717, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 9.472129049356753e-05, + "step": 7046 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1897, + "image_masks_loss": 0.012, + "llm_learning_rate": 9.465628035780641e-07, + "loss": 0.7493, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 9.465628035780641e-05, + "step": 7047 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1897, + "image_masks_loss": 0.0157, + "llm_learning_rate": 9.459128732884171e-07, + "loss": 0.7113, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.45912873288417e-05, + "step": 7048 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1896, + "image_masks_loss": 0.0171, + "llm_learning_rate": 9.452631141383059e-07, + "loss": 0.7804, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 9.452631141383059e-05, + "step": 7049 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1896, + "image_masks_loss": 0.0024, + "llm_learning_rate": 9.44613526199284e-07, + "loss": 0.7225, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 9.446135261992839e-05, + "step": 7050 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1895, + "image_masks_loss": 0.0093, + "llm_learning_rate": 9.439641095428837e-07, + "loss": 0.7421, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.439641095428836e-05, + "step": 7051 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1895, + "image_masks_loss": 0.0194, + "llm_learning_rate": 9.433148642406229e-07, + "loss": 0.7927, + "output_text_mask_loss": 0.0112, + "predictor_learning_rate": 9.433148642406229e-05, + "step": 7052 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1894, + "image_masks_loss": 0.0058, + "llm_learning_rate": 9.426657903639963e-07, + "loss": 0.6609, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 9.426657903639962e-05, + "step": 7053 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1894, + "image_masks_loss": 0.0115, + "llm_learning_rate": 9.420168879844818e-07, + "loss": 0.7014, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 9.420168879844817e-05, + "step": 7054 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1893, + "image_masks_loss": 0.0045, + "llm_learning_rate": 9.413681571735381e-07, + "loss": 0.6341, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 9.41368157173538e-05, + "step": 7055 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1893, + "image_masks_loss": 0.0091, + "llm_learning_rate": 9.407195980026043e-07, + "loss": 0.7257, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 9.407195980026043e-05, + "step": 7056 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1893, + "image_masks_loss": 0.025, + "llm_learning_rate": 9.400712105431017e-07, + "loss": 0.7367, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 9.400712105431017e-05, + "step": 7057 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1892, + "image_masks_loss": 0.0083, + "llm_learning_rate": 9.39422994866431e-07, + "loss": 0.6779, + "output_text_mask_loss": 0.0483, + "predictor_learning_rate": 9.39422994866431e-05, + "step": 7058 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1892, + "image_masks_loss": 0.0144, + "llm_learning_rate": 9.387749510439772e-07, + "loss": 0.6785, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.387749510439772e-05, + "step": 7059 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1891, + "image_masks_loss": 0.0146, + "llm_learning_rate": 9.381270791471034e-07, + "loss": 0.8093, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 9.381270791471033e-05, + "step": 7060 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1891, + "image_masks_loss": 0.0051, + "llm_learning_rate": 9.374793792471542e-07, + "loss": 0.7338, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 9.374793792471541e-05, + "step": 7061 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.189, + "image_masks_loss": 0.0098, + "llm_learning_rate": 9.368318514154564e-07, + "loss": 0.6589, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 9.368318514154564e-05, + "step": 7062 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.189, + "image_masks_loss": 0.0144, + "llm_learning_rate": 9.361844957233168e-07, + "loss": 0.7646, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 9.361844957233167e-05, + "step": 7063 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1889, + "image_masks_loss": 0.0043, + "llm_learning_rate": 9.355373122420239e-07, + "loss": 0.6697, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 9.355373122420238e-05, + "step": 7064 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1889, + "image_masks_loss": 0.0013, + "llm_learning_rate": 9.34890301042847e-07, + "loss": 0.7755, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 9.348903010428469e-05, + "step": 7065 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1889, + "image_masks_loss": 0.0113, + "llm_learning_rate": 9.342434621970362e-07, + "loss": 0.7592, + "output_text_mask_loss": 0.0393, + "predictor_learning_rate": 9.342434621970361e-05, + "step": 7066 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1888, + "image_masks_loss": 0.0111, + "llm_learning_rate": 9.335967957758226e-07, + "loss": 0.7738, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.335967957758226e-05, + "step": 7067 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1888, + "image_masks_loss": 0.0115, + "llm_learning_rate": 9.329503018504193e-07, + "loss": 0.7539, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 9.329503018504193e-05, + "step": 7068 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1887, + "image_masks_loss": 0.0132, + "llm_learning_rate": 9.323039804920192e-07, + "loss": 0.819, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 9.323039804920192e-05, + "step": 7069 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1887, + "image_masks_loss": 0.0157, + "llm_learning_rate": 9.316578317717964e-07, + "loss": 0.6384, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.316578317717964e-05, + "step": 7070 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1886, + "image_masks_loss": 0.0137, + "llm_learning_rate": 9.310118557609058e-07, + "loss": 0.7969, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 9.310118557609057e-05, + "step": 7071 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1886, + "image_masks_loss": 0.0147, + "llm_learning_rate": 9.303660525304853e-07, + "loss": 0.7637, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 9.303660525304852e-05, + "step": 7072 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1885, + "image_masks_loss": 0.0153, + "llm_learning_rate": 9.297204221516512e-07, + "loss": 0.6957, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 9.297204221516511e-05, + "step": 7073 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1885, + "image_masks_loss": 0.0121, + "llm_learning_rate": 9.290749646955022e-07, + "loss": 0.8129, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 9.290749646955021e-05, + "step": 7074 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1885, + "image_masks_loss": 0.0087, + "llm_learning_rate": 9.284296802331169e-07, + "loss": 0.7536, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 9.284296802331169e-05, + "step": 7075 + }, + { + "epoch": 0.72, + "gumbel_tau": 0.1884, + "image_masks_loss": 0.0101, + "llm_learning_rate": 9.277845688355558e-07, + "loss": 0.7831, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 9.277845688355558e-05, + "step": 7076 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1884, + "image_masks_loss": 0.0094, + "llm_learning_rate": 9.2713963057386e-07, + "loss": 0.7664, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 9.2713963057386e-05, + "step": 7077 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1883, + "image_masks_loss": 0.0136, + "llm_learning_rate": 9.264948655190506e-07, + "loss": 0.7177, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 9.264948655190505e-05, + "step": 7078 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1883, + "image_masks_loss": 0.0032, + "llm_learning_rate": 9.258502737421324e-07, + "loss": 0.7602, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 9.258502737421323e-05, + "step": 7079 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1882, + "image_masks_loss": 0.0052, + "llm_learning_rate": 9.25205855314088e-07, + "loss": 0.6802, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.25205855314088e-05, + "step": 7080 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1882, + "image_masks_loss": 0.0042, + "llm_learning_rate": 9.24561610305883e-07, + "loss": 0.754, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.245616103058829e-05, + "step": 7081 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1881, + "image_masks_loss": 0.0106, + "llm_learning_rate": 9.239175387884625e-07, + "loss": 0.6833, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 9.239175387884623e-05, + "step": 7082 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1881, + "image_masks_loss": 0.0063, + "llm_learning_rate": 9.23273640832753e-07, + "loss": 0.717, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 9.23273640832753e-05, + "step": 7083 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1881, + "image_masks_loss": 0.0058, + "llm_learning_rate": 9.226299165096622e-07, + "loss": 0.7438, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 9.226299165096622e-05, + "step": 7084 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.188, + "image_masks_loss": 0.0069, + "llm_learning_rate": 9.219863658900785e-07, + "loss": 0.7374, + "output_text_mask_loss": 0.0441, + "predictor_learning_rate": 9.219863658900784e-05, + "step": 7085 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.188, + "image_masks_loss": 0.0146, + "llm_learning_rate": 9.213429890448713e-07, + "loss": 0.6844, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 9.213429890448714e-05, + "step": 7086 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1879, + "image_masks_loss": 0.0068, + "llm_learning_rate": 9.206997860448902e-07, + "loss": 0.7921, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 9.206997860448902e-05, + "step": 7087 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1879, + "image_masks_loss": 0.0066, + "llm_learning_rate": 9.200567569609666e-07, + "loss": 0.6872, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 9.200567569609666e-05, + "step": 7088 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1878, + "image_masks_loss": 0.0124, + "llm_learning_rate": 9.194139018639119e-07, + "loss": 0.8089, + "output_text_mask_loss": 0.0083, + "predictor_learning_rate": 9.194139018639119e-05, + "step": 7089 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1878, + "image_masks_loss": 0.0132, + "llm_learning_rate": 9.187712208245184e-07, + "loss": 0.6589, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.187712208245183e-05, + "step": 7090 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1877, + "image_masks_loss": 0.0123, + "llm_learning_rate": 9.181287139135605e-07, + "loss": 0.6646, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.181287139135605e-05, + "step": 7091 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1877, + "image_masks_loss": 0.0037, + "llm_learning_rate": 9.174863812017925e-07, + "loss": 0.7327, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.174863812017925e-05, + "step": 7092 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1877, + "image_masks_loss": 0.0098, + "llm_learning_rate": 9.168442227599491e-07, + "loss": 0.6774, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.16844222759949e-05, + "step": 7093 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1876, + "image_masks_loss": 0.012, + "llm_learning_rate": 9.162022386587461e-07, + "loss": 0.6254, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 9.16202238658746e-05, + "step": 7094 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1876, + "image_masks_loss": 0.0125, + "llm_learning_rate": 9.155604289688804e-07, + "loss": 0.7973, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 9.155604289688804e-05, + "step": 7095 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1875, + "image_masks_loss": 0.0064, + "llm_learning_rate": 9.149187937610294e-07, + "loss": 0.6636, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.149187937610292e-05, + "step": 7096 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1875, + "image_masks_loss": 0.0045, + "llm_learning_rate": 9.142773331058508e-07, + "loss": 0.7446, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 9.142773331058507e-05, + "step": 7097 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1874, + "image_masks_loss": 0.0036, + "llm_learning_rate": 9.13636047073985e-07, + "loss": 0.6442, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 9.13636047073985e-05, + "step": 7098 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1874, + "image_masks_loss": 0.0037, + "llm_learning_rate": 9.129949357360512e-07, + "loss": 0.7841, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 9.129949357360512e-05, + "step": 7099 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1873, + "image_masks_loss": 0.0046, + "llm_learning_rate": 9.123539991626501e-07, + "loss": 0.6937, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.123539991626501e-05, + "step": 7100 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1873, + "image_masks_loss": 0.0124, + "llm_learning_rate": 9.117132374243628e-07, + "loss": 0.7478, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 9.117132374243629e-05, + "step": 7101 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1873, + "image_masks_loss": 0.0227, + "llm_learning_rate": 9.110726505917525e-07, + "loss": 0.6624, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 9.110726505917524e-05, + "step": 7102 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1872, + "image_masks_loss": 0.0018, + "llm_learning_rate": 9.104322387353592e-07, + "loss": 0.7878, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 9.104322387353592e-05, + "step": 7103 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1872, + "image_masks_loss": 0.007, + "llm_learning_rate": 9.097920019257094e-07, + "loss": 0.7857, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 9.097920019257095e-05, + "step": 7104 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1871, + "image_masks_loss": 0.0082, + "llm_learning_rate": 9.091519402333066e-07, + "loss": 0.7529, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 9.091519402333065e-05, + "step": 7105 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1871, + "image_masks_loss": 0.0133, + "llm_learning_rate": 9.085120537286355e-07, + "loss": 0.7489, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 9.085120537286353e-05, + "step": 7106 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.187, + "image_masks_loss": 0.0034, + "llm_learning_rate": 9.07872342482162e-07, + "loss": 0.7196, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 9.07872342482162e-05, + "step": 7107 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.187, + "image_masks_loss": 0.0072, + "llm_learning_rate": 9.072328065643324e-07, + "loss": 0.7299, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 9.072328065643324e-05, + "step": 7108 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1869, + "image_masks_loss": 0.0117, + "llm_learning_rate": 9.065934460455736e-07, + "loss": 0.6617, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 9.065934460455735e-05, + "step": 7109 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1869, + "image_masks_loss": 0.0102, + "llm_learning_rate": 9.059542609962932e-07, + "loss": 0.7606, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 9.059542609962932e-05, + "step": 7110 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1869, + "image_masks_loss": 0.0037, + "llm_learning_rate": 9.05315251486881e-07, + "loss": 0.7176, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.053152514868809e-05, + "step": 7111 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1868, + "image_masks_loss": 0.0068, + "llm_learning_rate": 9.046764175877051e-07, + "loss": 0.7508, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 9.046764175877051e-05, + "step": 7112 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1868, + "image_masks_loss": 0.0073, + "llm_learning_rate": 9.040377593691158e-07, + "loss": 0.686, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 9.040377593691157e-05, + "step": 7113 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1867, + "image_masks_loss": 0.0135, + "llm_learning_rate": 9.033992769014433e-07, + "loss": 0.7169, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 9.033992769014433e-05, + "step": 7114 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1867, + "image_masks_loss": 0.0098, + "llm_learning_rate": 9.027609702549986e-07, + "loss": 0.6944, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 9.027609702549986e-05, + "step": 7115 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1866, + "image_masks_loss": 0.0042, + "llm_learning_rate": 9.021228395000736e-07, + "loss": 0.699, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 9.021228395000736e-05, + "step": 7116 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1866, + "image_masks_loss": 0.0094, + "llm_learning_rate": 9.014848847069396e-07, + "loss": 0.7253, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.014848847069395e-05, + "step": 7117 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1866, + "image_masks_loss": 0.0106, + "llm_learning_rate": 9.008471059458518e-07, + "loss": 0.6958, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.008471059458517e-05, + "step": 7118 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1865, + "image_masks_loss": 0.0065, + "llm_learning_rate": 9.002095032870426e-07, + "loss": 0.7207, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 9.002095032870425e-05, + "step": 7119 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1865, + "image_masks_loss": 0.0078, + "llm_learning_rate": 8.995720768007271e-07, + "loss": 0.6605, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 8.99572076800727e-05, + "step": 7120 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1864, + "image_masks_loss": 0.0082, + "llm_learning_rate": 8.989348265570985e-07, + "loss": 0.6775, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 8.989348265570985e-05, + "step": 7121 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1864, + "image_masks_loss": 0.0139, + "llm_learning_rate": 8.98297752626332e-07, + "loss": 0.7908, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 8.98297752626332e-05, + "step": 7122 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1863, + "image_masks_loss": 0.0073, + "llm_learning_rate": 8.976608550785856e-07, + "loss": 0.748, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 8.976608550785856e-05, + "step": 7123 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1863, + "image_masks_loss": 0.0037, + "llm_learning_rate": 8.970241339839949e-07, + "loss": 0.6674, + "output_text_mask_loss": 0.0085, + "predictor_learning_rate": 8.970241339839949e-05, + "step": 7124 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1862, + "image_masks_loss": 0.0166, + "llm_learning_rate": 8.963875894126772e-07, + "loss": 0.7252, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 8.963875894126771e-05, + "step": 7125 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1862, + "image_masks_loss": 0.0188, + "llm_learning_rate": 8.9575122143473e-07, + "loss": 0.6501, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 8.9575122143473e-05, + "step": 7126 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1862, + "image_masks_loss": 0.0116, + "llm_learning_rate": 8.951150301202316e-07, + "loss": 0.7228, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 8.951150301202315e-05, + "step": 7127 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1861, + "image_masks_loss": 0.0082, + "llm_learning_rate": 8.944790155392407e-07, + "loss": 0.724, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 8.944790155392407e-05, + "step": 7128 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1861, + "image_masks_loss": 0.0156, + "llm_learning_rate": 8.938431777617962e-07, + "loss": 0.7021, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 8.938431777617961e-05, + "step": 7129 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.186, + "image_masks_loss": 0.0048, + "llm_learning_rate": 8.93207516857919e-07, + "loss": 0.7788, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 8.93207516857919e-05, + "step": 7130 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.186, + "image_masks_loss": 0.015, + "llm_learning_rate": 8.925720328976092e-07, + "loss": 0.7876, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.925720328976092e-05, + "step": 7131 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1859, + "image_masks_loss": 0.012, + "llm_learning_rate": 8.919367259508476e-07, + "loss": 0.7547, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 8.919367259508477e-05, + "step": 7132 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1859, + "image_masks_loss": 0.0033, + "llm_learning_rate": 8.913015960875956e-07, + "loss": 0.7694, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 8.913015960875955e-05, + "step": 7133 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1858, + "image_masks_loss": 0.0115, + "llm_learning_rate": 8.906666433777949e-07, + "loss": 0.7211, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 8.90666643377795e-05, + "step": 7134 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1858, + "image_masks_loss": 0.0057, + "llm_learning_rate": 8.900318678913683e-07, + "loss": 0.7466, + "output_text_mask_loss": 0.0176, + "predictor_learning_rate": 8.900318678913683e-05, + "step": 7135 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1858, + "image_masks_loss": 0.005, + "llm_learning_rate": 8.893972696982176e-07, + "loss": 0.6513, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.893972696982175e-05, + "step": 7136 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1857, + "image_masks_loss": 0.0111, + "llm_learning_rate": 8.887628488682276e-07, + "loss": 0.6757, + "output_text_mask_loss": 0.0595, + "predictor_learning_rate": 8.887628488682276e-05, + "step": 7137 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1857, + "image_masks_loss": 0.0223, + "llm_learning_rate": 8.881286054712618e-07, + "loss": 0.6132, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 8.881286054712617e-05, + "step": 7138 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1856, + "image_masks_loss": 0.0029, + "llm_learning_rate": 8.874945395771654e-07, + "loss": 0.7762, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 8.874945395771652e-05, + "step": 7139 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1856, + "image_masks_loss": 0.0122, + "llm_learning_rate": 8.868606512557607e-07, + "loss": 0.7628, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 8.868606512557606e-05, + "step": 7140 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1855, + "image_masks_loss": 0.0034, + "llm_learning_rate": 8.862269405768542e-07, + "loss": 0.7223, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 8.862269405768542e-05, + "step": 7141 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1855, + "image_masks_loss": 0.0069, + "llm_learning_rate": 8.85593407610231e-07, + "loss": 0.6918, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 8.85593407610231e-05, + "step": 7142 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1855, + "image_masks_loss": 0.0103, + "llm_learning_rate": 8.849600524256585e-07, + "loss": 0.7373, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 8.849600524256585e-05, + "step": 7143 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1854, + "image_masks_loss": 0.0044, + "llm_learning_rate": 8.843268750928822e-07, + "loss": 0.769, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 8.843268750928821e-05, + "step": 7144 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1854, + "image_masks_loss": 0.0122, + "llm_learning_rate": 8.836938756816294e-07, + "loss": 0.7324, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 8.836938756816293e-05, + "step": 7145 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1853, + "image_masks_loss": 0.0165, + "llm_learning_rate": 8.830610542616069e-07, + "loss": 0.7425, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 8.830610542616069e-05, + "step": 7146 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1853, + "image_masks_loss": 0.0084, + "llm_learning_rate": 8.824284109025031e-07, + "loss": 0.724, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 8.82428410902503e-05, + "step": 7147 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1852, + "image_masks_loss": 0.0102, + "llm_learning_rate": 8.817959456739853e-07, + "loss": 0.7662, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 8.817959456739851e-05, + "step": 7148 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1852, + "image_masks_loss": 0.0059, + "llm_learning_rate": 8.811636586457017e-07, + "loss": 0.7382, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.811636586457017e-05, + "step": 7149 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1851, + "image_masks_loss": 0.0044, + "llm_learning_rate": 8.805315498872826e-07, + "loss": 0.776, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 8.805315498872826e-05, + "step": 7150 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1851, + "image_masks_loss": 0.0118, + "llm_learning_rate": 8.798996194683365e-07, + "loss": 0.7119, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 8.798996194683364e-05, + "step": 7151 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1851, + "image_masks_loss": 0.0041, + "llm_learning_rate": 8.792678674584532e-07, + "loss": 0.6182, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 8.792678674584531e-05, + "step": 7152 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.185, + "image_masks_loss": 0.0102, + "llm_learning_rate": 8.78636293927202e-07, + "loss": 0.7322, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 8.786362939272019e-05, + "step": 7153 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.185, + "image_masks_loss": 0.0122, + "llm_learning_rate": 8.780048989441336e-07, + "loss": 0.7237, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 8.780048989441336e-05, + "step": 7154 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1849, + "image_masks_loss": 0.0062, + "llm_learning_rate": 8.773736825787788e-07, + "loss": 0.6822, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.773736825787786e-05, + "step": 7155 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1849, + "image_masks_loss": 0.0028, + "llm_learning_rate": 8.767426449006472e-07, + "loss": 0.758, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 8.767426449006471e-05, + "step": 7156 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1848, + "image_masks_loss": 0.0144, + "llm_learning_rate": 8.761117859792323e-07, + "loss": 0.6689, + "output_text_mask_loss": 0.0227, + "predictor_learning_rate": 8.761117859792322e-05, + "step": 7157 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1848, + "image_masks_loss": 0.0156, + "llm_learning_rate": 8.754811058840054e-07, + "loss": 0.7256, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 8.754811058840053e-05, + "step": 7158 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1848, + "image_masks_loss": 0.0104, + "llm_learning_rate": 8.748506046844165e-07, + "loss": 0.8287, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 8.748506046844165e-05, + "step": 7159 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1847, + "image_masks_loss": 0.0095, + "llm_learning_rate": 8.742202824498991e-07, + "loss": 0.6592, + "output_text_mask_loss": 0.0067, + "predictor_learning_rate": 8.74220282449899e-05, + "step": 7160 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1847, + "image_masks_loss": 0.0072, + "llm_learning_rate": 8.735901392498647e-07, + "loss": 0.7013, + "output_text_mask_loss": 0.0214, + "predictor_learning_rate": 8.735901392498646e-05, + "step": 7161 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1846, + "image_masks_loss": 0.0093, + "llm_learning_rate": 8.729601751537076e-07, + "loss": 0.7549, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 8.729601751537076e-05, + "step": 7162 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1846, + "image_masks_loss": 0.0096, + "llm_learning_rate": 8.723303902308003e-07, + "loss": 0.7305, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 8.723303902308002e-05, + "step": 7163 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1845, + "image_masks_loss": 0.0072, + "llm_learning_rate": 8.717007845504958e-07, + "loss": 0.7032, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.717007845504957e-05, + "step": 7164 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1845, + "image_masks_loss": 0.0067, + "llm_learning_rate": 8.710713581821275e-07, + "loss": 0.6476, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 8.710713581821275e-05, + "step": 7165 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1845, + "image_masks_loss": 0.0085, + "llm_learning_rate": 8.704421111950098e-07, + "loss": 0.6597, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 8.704421111950098e-05, + "step": 7166 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1844, + "image_masks_loss": 0.0087, + "llm_learning_rate": 8.698130436584365e-07, + "loss": 0.8258, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 8.698130436584364e-05, + "step": 7167 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1844, + "image_masks_loss": 0.0175, + "llm_learning_rate": 8.69184155641681e-07, + "loss": 0.713, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 8.69184155641681e-05, + "step": 7168 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1843, + "image_masks_loss": 0.0237, + "llm_learning_rate": 8.685554472139998e-07, + "loss": 0.6712, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.685554472139998e-05, + "step": 7169 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1843, + "image_masks_loss": 0.0056, + "llm_learning_rate": 8.679269184446266e-07, + "loss": 0.7068, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 8.679269184446265e-05, + "step": 7170 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1842, + "image_masks_loss": 0.0047, + "llm_learning_rate": 8.672985694027764e-07, + "loss": 0.7168, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 8.672985694027762e-05, + "step": 7171 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1842, + "image_masks_loss": 0.0058, + "llm_learning_rate": 8.666704001576443e-07, + "loss": 0.6524, + "output_text_mask_loss": 0.0526, + "predictor_learning_rate": 8.666704001576443e-05, + "step": 7172 + }, + { + "epoch": 0.73, + "gumbel_tau": 0.1841, + "image_masks_loss": 0.0067, + "llm_learning_rate": 8.66042410778406e-07, + "loss": 0.777, + "output_text_mask_loss": 0.042, + "predictor_learning_rate": 8.660424107784059e-05, + "step": 7173 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1841, + "image_masks_loss": 0.0109, + "llm_learning_rate": 8.654146013342169e-07, + "loss": 0.7208, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 8.654146013342168e-05, + "step": 7174 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1841, + "image_masks_loss": 0.01, + "llm_learning_rate": 8.64786971894212e-07, + "loss": 0.698, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 8.647869718942119e-05, + "step": 7175 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.184, + "image_masks_loss": 0.0083, + "llm_learning_rate": 8.641595225275096e-07, + "loss": 0.714, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.641595225275095e-05, + "step": 7176 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.184, + "image_masks_loss": 0.0066, + "llm_learning_rate": 8.635322533032034e-07, + "loss": 0.7809, + "output_text_mask_loss": 0.0173, + "predictor_learning_rate": 8.635322533032034e-05, + "step": 7177 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1839, + "image_masks_loss": 0.0059, + "llm_learning_rate": 8.629051642903708e-07, + "loss": 0.7473, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 8.629051642903707e-05, + "step": 7178 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1839, + "image_masks_loss": 0.0086, + "llm_learning_rate": 8.622782555580677e-07, + "loss": 0.6804, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.622782555580677e-05, + "step": 7179 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1838, + "image_masks_loss": 0.0048, + "llm_learning_rate": 8.616515271753311e-07, + "loss": 0.653, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 8.616515271753312e-05, + "step": 7180 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1838, + "image_masks_loss": 0.0045, + "llm_learning_rate": 8.610249792111769e-07, + "loss": 0.7475, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 8.610249792111769e-05, + "step": 7181 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1838, + "image_masks_loss": 0.0066, + "llm_learning_rate": 8.603986117346036e-07, + "loss": 0.6671, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 8.603986117346035e-05, + "step": 7182 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1837, + "image_masks_loss": 0.0142, + "llm_learning_rate": 8.59772424814587e-07, + "loss": 0.6959, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 8.597724248145869e-05, + "step": 7183 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1837, + "image_masks_loss": 0.0066, + "llm_learning_rate": 8.591464185200843e-07, + "loss": 0.7114, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 8.591464185200842e-05, + "step": 7184 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1836, + "image_masks_loss": 0.0079, + "llm_learning_rate": 8.585205929200333e-07, + "loss": 0.71, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 8.585205929200332e-05, + "step": 7185 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1836, + "image_masks_loss": 0.0122, + "llm_learning_rate": 8.578949480833503e-07, + "loss": 0.8019, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 8.578949480833503e-05, + "step": 7186 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1835, + "image_masks_loss": 0.0102, + "llm_learning_rate": 8.572694840789336e-07, + "loss": 0.5707, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.572694840789336e-05, + "step": 7187 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1835, + "image_masks_loss": 0.0207, + "llm_learning_rate": 8.566442009756596e-07, + "loss": 0.7448, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 8.566442009756594e-05, + "step": 7188 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1835, + "image_masks_loss": 0.0046, + "llm_learning_rate": 8.560190988423873e-07, + "loss": 0.7528, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 8.560190988423871e-05, + "step": 7189 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1834, + "image_masks_loss": 0.0122, + "llm_learning_rate": 8.553941777479538e-07, + "loss": 0.7567, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 8.553941777479537e-05, + "step": 7190 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1834, + "image_masks_loss": 0.0052, + "llm_learning_rate": 8.547694377611765e-07, + "loss": 0.6871, + "output_text_mask_loss": 0.0013, + "predictor_learning_rate": 8.547694377611765e-05, + "step": 7191 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1833, + "image_masks_loss": 0.0112, + "llm_learning_rate": 8.541448789508533e-07, + "loss": 0.7277, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.541448789508533e-05, + "step": 7192 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1833, + "image_masks_loss": 0.0226, + "llm_learning_rate": 8.535205013857623e-07, + "loss": 0.7768, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 8.535205013857623e-05, + "step": 7193 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1832, + "image_masks_loss": 0.0045, + "llm_learning_rate": 8.528963051346608e-07, + "loss": 0.706, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 8.528963051346608e-05, + "step": 7194 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1832, + "image_masks_loss": 0.0064, + "llm_learning_rate": 8.52272290266287e-07, + "loss": 0.781, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 8.522722902662869e-05, + "step": 7195 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1832, + "image_masks_loss": 0.0183, + "llm_learning_rate": 8.516484568493588e-07, + "loss": 0.7364, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.516484568493587e-05, + "step": 7196 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1831, + "image_masks_loss": 0.0061, + "llm_learning_rate": 8.510248049525741e-07, + "loss": 0.6551, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 8.51024804952574e-05, + "step": 7197 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1831, + "image_masks_loss": 0.0077, + "llm_learning_rate": 8.504013346446108e-07, + "loss": 0.7663, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 8.504013346446107e-05, + "step": 7198 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.183, + "image_masks_loss": 0.0035, + "llm_learning_rate": 8.497780459941269e-07, + "loss": 0.6565, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 8.497780459941267e-05, + "step": 7199 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.183, + "image_masks_loss": 0.0033, + "llm_learning_rate": 8.491549390697596e-07, + "loss": 0.7645, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 8.491549390697596e-05, + "step": 7200 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1829, + "image_masks_loss": 0.0013, + "llm_learning_rate": 8.485320139401282e-07, + "loss": 0.7839, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 8.485320139401282e-05, + "step": 7201 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1829, + "image_masks_loss": 0.0084, + "llm_learning_rate": 8.479092706738299e-07, + "loss": 0.6431, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 8.479092706738298e-05, + "step": 7202 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1828, + "image_masks_loss": 0.0076, + "llm_learning_rate": 8.472867093394427e-07, + "loss": 0.7122, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 8.472867093394426e-05, + "step": 7203 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1828, + "image_masks_loss": 0.0038, + "llm_learning_rate": 8.466643300055241e-07, + "loss": 0.7633, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 8.466643300055241e-05, + "step": 7204 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1828, + "image_masks_loss": 0.011, + "llm_learning_rate": 8.460421327406123e-07, + "loss": 0.6587, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.460421327406123e-05, + "step": 7205 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1827, + "image_masks_loss": 0.0046, + "llm_learning_rate": 8.454201176132251e-07, + "loss": 0.6575, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.454201176132251e-05, + "step": 7206 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1827, + "image_masks_loss": 0.0093, + "llm_learning_rate": 8.44798284691859e-07, + "loss": 0.6964, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 8.447982846918589e-05, + "step": 7207 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1826, + "image_masks_loss": 0.0061, + "llm_learning_rate": 8.441766340449934e-07, + "loss": 0.7149, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 8.441766340449933e-05, + "step": 7208 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1826, + "image_masks_loss": 0.0176, + "llm_learning_rate": 8.435551657410849e-07, + "loss": 0.7383, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 8.435551657410848e-05, + "step": 7209 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1825, + "image_masks_loss": 0.0128, + "llm_learning_rate": 8.429338798485715e-07, + "loss": 0.6925, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 8.429338798485713e-05, + "step": 7210 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1825, + "image_masks_loss": 0.015, + "llm_learning_rate": 8.423127764358699e-07, + "loss": 0.7489, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 8.423127764358698e-05, + "step": 7211 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1825, + "image_masks_loss": 0.0095, + "llm_learning_rate": 8.416918555713776e-07, + "loss": 0.6815, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 8.416918555713776e-05, + "step": 7212 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1824, + "image_masks_loss": 0.0043, + "llm_learning_rate": 8.410711173234723e-07, + "loss": 0.6574, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 8.410711173234722e-05, + "step": 7213 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1824, + "image_masks_loss": 0.008, + "llm_learning_rate": 8.404505617605105e-07, + "loss": 0.7399, + "output_text_mask_loss": 0.0225, + "predictor_learning_rate": 8.404505617605105e-05, + "step": 7214 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1823, + "image_masks_loss": 0.0102, + "llm_learning_rate": 8.398301889508292e-07, + "loss": 0.7375, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 8.398301889508293e-05, + "step": 7215 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1823, + "image_masks_loss": 0.009, + "llm_learning_rate": 8.392099989627456e-07, + "loss": 0.6928, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 8.392099989627455e-05, + "step": 7216 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1822, + "image_masks_loss": 0.0085, + "llm_learning_rate": 8.385899918645563e-07, + "loss": 0.823, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 8.385899918645562e-05, + "step": 7217 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1822, + "image_masks_loss": 0.0053, + "llm_learning_rate": 8.379701677245375e-07, + "loss": 0.7676, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 8.379701677245374e-05, + "step": 7218 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1822, + "image_masks_loss": 0.0085, + "llm_learning_rate": 8.373505266109458e-07, + "loss": 0.7728, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.373505266109457e-05, + "step": 7219 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1821, + "image_masks_loss": 0.0127, + "llm_learning_rate": 8.367310685920169e-07, + "loss": 0.7441, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 8.367310685920169e-05, + "step": 7220 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1821, + "image_masks_loss": 0.0051, + "llm_learning_rate": 8.361117937359681e-07, + "loss": 0.7014, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 8.36111793735968e-05, + "step": 7221 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.182, + "image_masks_loss": 0.0056, + "llm_learning_rate": 8.354927021109949e-07, + "loss": 0.681, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 8.354927021109948e-05, + "step": 7222 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.182, + "image_masks_loss": 0.0049, + "llm_learning_rate": 8.34873793785273e-07, + "loss": 0.6351, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 8.34873793785273e-05, + "step": 7223 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1819, + "image_masks_loss": 0.0119, + "llm_learning_rate": 8.342550688269577e-07, + "loss": 0.6871, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 8.342550688269576e-05, + "step": 7224 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1819, + "image_masks_loss": 0.0119, + "llm_learning_rate": 8.336365273041843e-07, + "loss": 0.7531, + "output_text_mask_loss": 0.0778, + "predictor_learning_rate": 8.336365273041843e-05, + "step": 7225 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1819, + "image_masks_loss": 0.007, + "llm_learning_rate": 8.330181692850686e-07, + "loss": 0.6295, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 8.330181692850685e-05, + "step": 7226 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1818, + "image_masks_loss": 0.0077, + "llm_learning_rate": 8.323999948377043e-07, + "loss": 0.6576, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 8.323999948377043e-05, + "step": 7227 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1818, + "image_masks_loss": 0.0079, + "llm_learning_rate": 8.317820040301675e-07, + "loss": 0.793, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 8.317820040301674e-05, + "step": 7228 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1817, + "image_masks_loss": 0.0185, + "llm_learning_rate": 8.311641969305126e-07, + "loss": 0.7746, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 8.311641969305125e-05, + "step": 7229 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1817, + "image_masks_loss": 0.0014, + "llm_learning_rate": 8.305465736067733e-07, + "loss": 0.7609, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 8.305465736067732e-05, + "step": 7230 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1816, + "image_masks_loss": 0.0055, + "llm_learning_rate": 8.299291341269639e-07, + "loss": 0.7083, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 8.299291341269639e-05, + "step": 7231 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1816, + "image_masks_loss": 0.004, + "llm_learning_rate": 8.29311878559079e-07, + "loss": 0.714, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 8.29311878559079e-05, + "step": 7232 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1816, + "image_masks_loss": 0.0024, + "llm_learning_rate": 8.286948069710896e-07, + "loss": 0.7792, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 8.286948069710894e-05, + "step": 7233 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1815, + "image_masks_loss": 0.0132, + "llm_learning_rate": 8.280779194309518e-07, + "loss": 0.7118, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.280779194309518e-05, + "step": 7234 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1815, + "image_masks_loss": 0.0095, + "llm_learning_rate": 8.274612160065973e-07, + "loss": 0.8014, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 8.274612160065973e-05, + "step": 7235 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1814, + "image_masks_loss": 0.0112, + "llm_learning_rate": 8.268446967659394e-07, + "loss": 0.6858, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.268446967659393e-05, + "step": 7236 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1814, + "image_masks_loss": 0.0022, + "llm_learning_rate": 8.262283617768701e-07, + "loss": 0.7249, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 8.262283617768701e-05, + "step": 7237 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1813, + "image_masks_loss": 0.0037, + "llm_learning_rate": 8.25612211107262e-07, + "loss": 0.757, + "output_text_mask_loss": 0.0309, + "predictor_learning_rate": 8.25612211107262e-05, + "step": 7238 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1813, + "image_masks_loss": 0.0053, + "llm_learning_rate": 8.249962448249665e-07, + "loss": 0.7449, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 8.249962448249665e-05, + "step": 7239 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1813, + "image_masks_loss": 0.0027, + "llm_learning_rate": 8.243804629978147e-07, + "loss": 0.7197, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 8.243804629978147e-05, + "step": 7240 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1812, + "image_masks_loss": 0.0058, + "llm_learning_rate": 8.237648656936195e-07, + "loss": 0.7324, + "output_text_mask_loss": 0.0033, + "predictor_learning_rate": 8.237648656936195e-05, + "step": 7241 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1812, + "image_masks_loss": 0.0072, + "llm_learning_rate": 8.231494529801709e-07, + "loss": 0.6232, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.231494529801708e-05, + "step": 7242 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1811, + "image_masks_loss": 0.0028, + "llm_learning_rate": 8.225342249252396e-07, + "loss": 0.8079, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 8.225342249252396e-05, + "step": 7243 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1811, + "image_masks_loss": 0.0041, + "llm_learning_rate": 8.21919181596576e-07, + "loss": 0.6535, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 8.21919181596576e-05, + "step": 7244 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.181, + "image_masks_loss": 0.006, + "llm_learning_rate": 8.213043230619097e-07, + "loss": 0.7272, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 8.213043230619097e-05, + "step": 7245 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.181, + "image_masks_loss": 0.0036, + "llm_learning_rate": 8.206896493889507e-07, + "loss": 0.7793, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 8.206896493889506e-05, + "step": 7246 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.181, + "image_masks_loss": 0.0086, + "llm_learning_rate": 8.200751606453872e-07, + "loss": 0.7052, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 8.200751606453871e-05, + "step": 7247 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1809, + "image_masks_loss": 0.0102, + "llm_learning_rate": 8.194608568988893e-07, + "loss": 0.8532, + "output_text_mask_loss": 0.0024, + "predictor_learning_rate": 8.194608568988893e-05, + "step": 7248 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1809, + "image_masks_loss": 0.0183, + "llm_learning_rate": 8.188467382171056e-07, + "loss": 0.7666, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 8.188467382171055e-05, + "step": 7249 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1808, + "image_masks_loss": 0.0033, + "llm_learning_rate": 8.182328046676643e-07, + "loss": 0.7195, + "output_text_mask_loss": 0.069, + "predictor_learning_rate": 8.182328046676644e-05, + "step": 7250 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1808, + "image_masks_loss": 0.0044, + "llm_learning_rate": 8.176190563181718e-07, + "loss": 0.742, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.176190563181718e-05, + "step": 7251 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1807, + "image_masks_loss": 0.0059, + "llm_learning_rate": 8.17005493236215e-07, + "loss": 0.6942, + "output_text_mask_loss": 0.0018, + "predictor_learning_rate": 8.170054932362151e-05, + "step": 7252 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1807, + "image_masks_loss": 0.0095, + "llm_learning_rate": 8.163921154893633e-07, + "loss": 0.6986, + "output_text_mask_loss": 0.0801, + "predictor_learning_rate": 8.163921154893634e-05, + "step": 7253 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1807, + "image_masks_loss": 0.0029, + "llm_learning_rate": 8.157789231451618e-07, + "loss": 0.7355, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.157789231451618e-05, + "step": 7254 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1806, + "image_masks_loss": 0.0057, + "llm_learning_rate": 8.151659162711367e-07, + "loss": 0.6758, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 8.151659162711367e-05, + "step": 7255 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1806, + "image_masks_loss": 0.0014, + "llm_learning_rate": 8.145530949347938e-07, + "loss": 0.7178, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 8.145530949347937e-05, + "step": 7256 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1805, + "image_masks_loss": 0.0073, + "llm_learning_rate": 8.139404592036182e-07, + "loss": 0.6965, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.139404592036182e-05, + "step": 7257 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1805, + "image_masks_loss": 0.0053, + "llm_learning_rate": 8.133280091450746e-07, + "loss": 0.7932, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 8.133280091450746e-05, + "step": 7258 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1804, + "image_masks_loss": 0.0106, + "llm_learning_rate": 8.127157448266069e-07, + "loss": 0.6299, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.127157448266068e-05, + "step": 7259 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1804, + "image_masks_loss": 0.0102, + "llm_learning_rate": 8.121036663156404e-07, + "loss": 0.7506, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 8.121036663156403e-05, + "step": 7260 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1804, + "image_masks_loss": 0.0074, + "llm_learning_rate": 8.11491773679578e-07, + "loss": 0.7113, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.11491773679578e-05, + "step": 7261 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1803, + "image_masks_loss": 0.0045, + "llm_learning_rate": 8.108800669858025e-07, + "loss": 0.733, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 8.108800669858024e-05, + "step": 7262 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1803, + "image_masks_loss": 0.004, + "llm_learning_rate": 8.10268546301676e-07, + "loss": 0.7508, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 8.10268546301676e-05, + "step": 7263 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1802, + "image_masks_loss": 0.0075, + "llm_learning_rate": 8.096572116945411e-07, + "loss": 0.7829, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 8.09657211694541e-05, + "step": 7264 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1802, + "image_masks_loss": 0.0058, + "llm_learning_rate": 8.090460632317193e-07, + "loss": 0.7555, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 8.090460632317192e-05, + "step": 7265 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1802, + "image_masks_loss": 0.0057, + "llm_learning_rate": 8.084351009805105e-07, + "loss": 0.6638, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 8.084351009805105e-05, + "step": 7266 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1801, + "image_masks_loss": 0.0076, + "llm_learning_rate": 8.078243250081971e-07, + "loss": 0.8201, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 8.07824325008197e-05, + "step": 7267 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1801, + "image_masks_loss": 0.0068, + "llm_learning_rate": 8.072137353820384e-07, + "loss": 0.7737, + "output_text_mask_loss": 0.0561, + "predictor_learning_rate": 8.072137353820383e-05, + "step": 7268 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.18, + "image_masks_loss": 0.0074, + "llm_learning_rate": 8.066033321692743e-07, + "loss": 0.8115, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 8.066033321692743e-05, + "step": 7269 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.18, + "image_masks_loss": 0.0043, + "llm_learning_rate": 8.059931154371223e-07, + "loss": 0.7215, + "output_text_mask_loss": 0.0763, + "predictor_learning_rate": 8.059931154371222e-05, + "step": 7270 + }, + { + "epoch": 0.74, + "gumbel_tau": 0.1799, + "image_masks_loss": 0.0086, + "llm_learning_rate": 8.053830852527816e-07, + "loss": 0.7788, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 8.053830852527815e-05, + "step": 7271 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1799, + "image_masks_loss": 0.0048, + "llm_learning_rate": 8.047732416834294e-07, + "loss": 0.731, + "output_text_mask_loss": 0.0877, + "predictor_learning_rate": 8.047732416834293e-05, + "step": 7272 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1799, + "image_masks_loss": 0.0162, + "llm_learning_rate": 8.041635847962248e-07, + "loss": 0.7671, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 8.041635847962248e-05, + "step": 7273 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1798, + "image_masks_loss": 0.0084, + "llm_learning_rate": 8.035541146583034e-07, + "loss": 0.8318, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.035541146583033e-05, + "step": 7274 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1798, + "image_masks_loss": 0.0057, + "llm_learning_rate": 8.029448313367815e-07, + "loss": 0.7949, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 8.029448313367815e-05, + "step": 7275 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1797, + "image_masks_loss": 0.0065, + "llm_learning_rate": 8.023357348987549e-07, + "loss": 0.7167, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 8.023357348987548e-05, + "step": 7276 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1797, + "image_masks_loss": 0.0079, + "llm_learning_rate": 8.017268254112986e-07, + "loss": 0.748, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 8.017268254112986e-05, + "step": 7277 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1796, + "image_masks_loss": 0.007, + "llm_learning_rate": 8.011181029414672e-07, + "loss": 0.6517, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 8.011181029414671e-05, + "step": 7278 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1796, + "image_masks_loss": 0.0081, + "llm_learning_rate": 8.005095675562932e-07, + "loss": 0.7277, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 8.005095675562932e-05, + "step": 7279 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1796, + "image_masks_loss": 0.0068, + "llm_learning_rate": 7.999012193227923e-07, + "loss": 0.6994, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 7.999012193227923e-05, + "step": 7280 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1795, + "image_masks_loss": 0.0129, + "llm_learning_rate": 7.992930583079558e-07, + "loss": 0.6137, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.992930583079558e-05, + "step": 7281 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1795, + "image_masks_loss": 0.0046, + "llm_learning_rate": 7.986850845787558e-07, + "loss": 0.754, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 7.986850845787558e-05, + "step": 7282 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1794, + "image_masks_loss": 0.0079, + "llm_learning_rate": 7.980772982021439e-07, + "loss": 0.7184, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.98077298202144e-05, + "step": 7283 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1794, + "image_masks_loss": 0.0114, + "llm_learning_rate": 7.974696992450509e-07, + "loss": 0.7639, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 7.974696992450508e-05, + "step": 7284 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1793, + "image_masks_loss": 0.0037, + "llm_learning_rate": 7.968622877743868e-07, + "loss": 0.7725, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 7.968622877743867e-05, + "step": 7285 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1793, + "image_masks_loss": 0.0088, + "llm_learning_rate": 7.962550638570402e-07, + "loss": 0.6264, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.962550638570401e-05, + "step": 7286 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1793, + "image_masks_loss": 0.0044, + "llm_learning_rate": 7.956480275598826e-07, + "loss": 0.737, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 7.956480275598827e-05, + "step": 7287 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1792, + "image_masks_loss": 0.0022, + "llm_learning_rate": 7.950411789497597e-07, + "loss": 0.7704, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 7.950411789497597e-05, + "step": 7288 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1792, + "image_masks_loss": 0.0157, + "llm_learning_rate": 7.944345180935001e-07, + "loss": 0.7084, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 7.944345180935e-05, + "step": 7289 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1791, + "image_masks_loss": 0.0127, + "llm_learning_rate": 7.9382804505791e-07, + "loss": 0.7592, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 7.9382804505791e-05, + "step": 7290 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1791, + "image_masks_loss": 0.0102, + "llm_learning_rate": 7.932217599097753e-07, + "loss": 0.6777, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 7.932217599097751e-05, + "step": 7291 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.179, + "image_masks_loss": 0.0071, + "llm_learning_rate": 7.926156627158629e-07, + "loss": 0.6792, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.926156627158629e-05, + "step": 7292 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.179, + "image_masks_loss": 0.0035, + "llm_learning_rate": 7.920097535429167e-07, + "loss": 0.7296, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 7.920097535429166e-05, + "step": 7293 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.179, + "image_masks_loss": 0.0074, + "llm_learning_rate": 7.914040324576608e-07, + "loss": 0.685, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.914040324576607e-05, + "step": 7294 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1789, + "image_masks_loss": 0.0042, + "llm_learning_rate": 7.907984995267987e-07, + "loss": 0.7057, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 7.907984995267986e-05, + "step": 7295 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1789, + "image_masks_loss": 0.0036, + "llm_learning_rate": 7.901931548170127e-07, + "loss": 0.7503, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 7.901931548170127e-05, + "step": 7296 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1788, + "image_masks_loss": 0.0072, + "llm_learning_rate": 7.895879983949653e-07, + "loss": 0.7718, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 7.895879983949653e-05, + "step": 7297 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1788, + "image_masks_loss": 0.0059, + "llm_learning_rate": 7.889830303272963e-07, + "loss": 0.7012, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 7.889830303272964e-05, + "step": 7298 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1788, + "image_masks_loss": 0.0124, + "llm_learning_rate": 7.883782506806278e-07, + "loss": 0.7637, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 7.883782506806278e-05, + "step": 7299 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1787, + "image_masks_loss": 0.0051, + "llm_learning_rate": 7.87773659521559e-07, + "loss": 0.7095, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 7.87773659521559e-05, + "step": 7300 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1787, + "image_masks_loss": 0.0076, + "llm_learning_rate": 7.871692569166689e-07, + "loss": 0.7887, + "output_text_mask_loss": 0.0303, + "predictor_learning_rate": 7.871692569166688e-05, + "step": 7301 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1786, + "image_masks_loss": 0.0086, + "llm_learning_rate": 7.86565042932515e-07, + "loss": 0.7237, + "output_text_mask_loss": 0.0725, + "predictor_learning_rate": 7.86565042932515e-05, + "step": 7302 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1786, + "image_masks_loss": 0.0029, + "llm_learning_rate": 7.859610176356352e-07, + "loss": 0.6921, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.859610176356352e-05, + "step": 7303 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1785, + "image_masks_loss": 0.0113, + "llm_learning_rate": 7.853571810925461e-07, + "loss": 0.6538, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 7.853571810925462e-05, + "step": 7304 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1785, + "image_masks_loss": 0.0042, + "llm_learning_rate": 7.847535333697426e-07, + "loss": 0.6841, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 7.847535333697426e-05, + "step": 7305 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1785, + "image_masks_loss": 0.0084, + "llm_learning_rate": 7.841500745337022e-07, + "loss": 0.7475, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.841500745337021e-05, + "step": 7306 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1784, + "image_masks_loss": 0.0073, + "llm_learning_rate": 7.835468046508768e-07, + "loss": 0.7607, + "output_text_mask_loss": 0.038, + "predictor_learning_rate": 7.835468046508768e-05, + "step": 7307 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1784, + "image_masks_loss": 0.0118, + "llm_learning_rate": 7.829437237877002e-07, + "loss": 0.6834, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 7.829437237877002e-05, + "step": 7308 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1783, + "image_masks_loss": 0.0095, + "llm_learning_rate": 7.823408320105855e-07, + "loss": 0.7734, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 7.823408320105855e-05, + "step": 7309 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1783, + "image_masks_loss": 0.0131, + "llm_learning_rate": 7.817381293859244e-07, + "loss": 0.6752, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 7.817381293859243e-05, + "step": 7310 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1782, + "image_masks_loss": 0.005, + "llm_learning_rate": 7.81135615980087e-07, + "loss": 0.6907, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 7.81135615980087e-05, + "step": 7311 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1782, + "image_masks_loss": 0.005, + "llm_learning_rate": 7.805332918594247e-07, + "loss": 0.807, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 7.805332918594247e-05, + "step": 7312 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1782, + "image_masks_loss": 0.0123, + "llm_learning_rate": 7.799311570902665e-07, + "loss": 0.6643, + "output_text_mask_loss": 0.0273, + "predictor_learning_rate": 7.799311570902664e-05, + "step": 7313 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1781, + "image_masks_loss": 0.0096, + "llm_learning_rate": 7.793292117389206e-07, + "loss": 0.7552, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 7.793292117389206e-05, + "step": 7314 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1781, + "image_masks_loss": 0.0067, + "llm_learning_rate": 7.787274558716745e-07, + "loss": 0.7167, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 7.787274558716745e-05, + "step": 7315 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.178, + "image_masks_loss": 0.0042, + "llm_learning_rate": 7.781258895547949e-07, + "loss": 0.7771, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 7.781258895547949e-05, + "step": 7316 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.178, + "image_masks_loss": 0.0078, + "llm_learning_rate": 7.775245128545278e-07, + "loss": 0.7517, + "output_text_mask_loss": 0.0359, + "predictor_learning_rate": 7.775245128545277e-05, + "step": 7317 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.178, + "image_masks_loss": 0.009, + "llm_learning_rate": 7.76923325837097e-07, + "loss": 0.7194, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 7.769233258370969e-05, + "step": 7318 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1779, + "image_masks_loss": 0.0083, + "llm_learning_rate": 7.763223285687085e-07, + "loss": 0.7744, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 7.763223285687085e-05, + "step": 7319 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1779, + "image_masks_loss": 0.007, + "llm_learning_rate": 7.757215211155445e-07, + "loss": 0.7136, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 7.757215211155444e-05, + "step": 7320 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1778, + "image_masks_loss": 0.0039, + "llm_learning_rate": 7.751209035437671e-07, + "loss": 0.7261, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 7.75120903543767e-05, + "step": 7321 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1778, + "image_masks_loss": 0.0194, + "llm_learning_rate": 7.745204759195179e-07, + "loss": 0.7377, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.745204759195179e-05, + "step": 7322 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1777, + "image_masks_loss": 0.0041, + "llm_learning_rate": 7.739202383089175e-07, + "loss": 0.7164, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 7.739202383089175e-05, + "step": 7323 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1777, + "image_masks_loss": 0.0079, + "llm_learning_rate": 7.733201907780649e-07, + "loss": 0.7716, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 7.733201907780648e-05, + "step": 7324 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1777, + "image_masks_loss": 0.0198, + "llm_learning_rate": 7.727203333930392e-07, + "loss": 0.7843, + "output_text_mask_loss": 0.0448, + "predictor_learning_rate": 7.727203333930391e-05, + "step": 7325 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1776, + "image_masks_loss": 0.0076, + "llm_learning_rate": 7.721206662198974e-07, + "loss": 0.6659, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 7.721206662198974e-05, + "step": 7326 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1776, + "image_masks_loss": 0.0189, + "llm_learning_rate": 7.715211893246766e-07, + "loss": 0.748, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 7.715211893246766e-05, + "step": 7327 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1775, + "image_masks_loss": 0.0061, + "llm_learning_rate": 7.709219027733928e-07, + "loss": 0.7922, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 7.709219027733927e-05, + "step": 7328 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1775, + "image_masks_loss": 0.0057, + "llm_learning_rate": 7.703228066320401e-07, + "loss": 0.8225, + "output_text_mask_loss": 0.1221, + "predictor_learning_rate": 7.7032280663204e-05, + "step": 7329 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1775, + "image_masks_loss": 0.0038, + "llm_learning_rate": 7.697239009665919e-07, + "loss": 0.8406, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 7.697239009665918e-05, + "step": 7330 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1774, + "image_masks_loss": 0.0033, + "llm_learning_rate": 7.691251858430026e-07, + "loss": 0.6926, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 7.691251858430026e-05, + "step": 7331 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1774, + "image_masks_loss": 0.0099, + "llm_learning_rate": 7.68526661327203e-07, + "loss": 0.6859, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 7.685266613272029e-05, + "step": 7332 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1773, + "image_masks_loss": 0.0051, + "llm_learning_rate": 7.679283274851046e-07, + "loss": 0.7129, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 7.679283274851045e-05, + "step": 7333 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1773, + "image_masks_loss": 0.0091, + "llm_learning_rate": 7.673301843825965e-07, + "loss": 0.8409, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 7.673301843825964e-05, + "step": 7334 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1772, + "image_masks_loss": 0.005, + "llm_learning_rate": 7.667322320855477e-07, + "loss": 0.7383, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 7.667322320855476e-05, + "step": 7335 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1772, + "image_masks_loss": 0.0077, + "llm_learning_rate": 7.661344706598068e-07, + "loss": 0.7778, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 7.661344706598066e-05, + "step": 7336 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1772, + "image_masks_loss": 0.0089, + "llm_learning_rate": 7.655369001711987e-07, + "loss": 0.6391, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.655369001711987e-05, + "step": 7337 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1771, + "image_masks_loss": 0.0056, + "llm_learning_rate": 7.649395206855315e-07, + "loss": 0.6933, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 7.649395206855315e-05, + "step": 7338 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1771, + "image_masks_loss": 0.0054, + "llm_learning_rate": 7.643423322685889e-07, + "loss": 0.7229, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 7.643423322685888e-05, + "step": 7339 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.177, + "image_masks_loss": 0.0122, + "llm_learning_rate": 7.637453349861348e-07, + "loss": 0.7658, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.637453349861348e-05, + "step": 7340 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.177, + "image_masks_loss": 0.0175, + "llm_learning_rate": 7.631485289039117e-07, + "loss": 0.7625, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 7.631485289039117e-05, + "step": 7341 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1769, + "image_masks_loss": 0.0046, + "llm_learning_rate": 7.625519140876414e-07, + "loss": 0.7864, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 7.625519140876413e-05, + "step": 7342 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1769, + "image_masks_loss": 0.0073, + "llm_learning_rate": 7.619554906030241e-07, + "loss": 0.7093, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.61955490603024e-05, + "step": 7343 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1769, + "image_masks_loss": 0.011, + "llm_learning_rate": 7.613592585157398e-07, + "loss": 0.5589, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.613592585157397e-05, + "step": 7344 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1768, + "image_masks_loss": 0.0089, + "llm_learning_rate": 7.607632178914464e-07, + "loss": 0.5571, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 7.607632178914464e-05, + "step": 7345 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1768, + "image_masks_loss": 0.01, + "llm_learning_rate": 7.601673687957817e-07, + "loss": 0.6598, + "output_text_mask_loss": 0.0408, + "predictor_learning_rate": 7.601673687957816e-05, + "step": 7346 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1767, + "image_masks_loss": 0.0121, + "llm_learning_rate": 7.595717112943615e-07, + "loss": 0.6896, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 7.595717112943615e-05, + "step": 7347 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1767, + "image_masks_loss": 0.0073, + "llm_learning_rate": 7.58976245452781e-07, + "loss": 0.7768, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 7.589762454527809e-05, + "step": 7348 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1767, + "image_masks_loss": 0.0084, + "llm_learning_rate": 7.583809713366144e-07, + "loss": 0.7284, + "output_text_mask_loss": 0.025, + "predictor_learning_rate": 7.583809713366144e-05, + "step": 7349 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1766, + "image_masks_loss": 0.0024, + "llm_learning_rate": 7.577858890114137e-07, + "loss": 0.7021, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 7.577858890114137e-05, + "step": 7350 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1766, + "image_masks_loss": 0.0159, + "llm_learning_rate": 7.571909985427125e-07, + "loss": 0.7933, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 7.571909985427125e-05, + "step": 7351 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1765, + "image_masks_loss": 0.0105, + "llm_learning_rate": 7.565962999960205e-07, + "loss": 0.7544, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 7.565962999960205e-05, + "step": 7352 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1765, + "image_masks_loss": 0.0132, + "llm_learning_rate": 7.560017934368272e-07, + "loss": 0.6835, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 7.560017934368271e-05, + "step": 7353 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1764, + "image_masks_loss": 0.0128, + "llm_learning_rate": 7.554074789306007e-07, + "loss": 0.6998, + "output_text_mask_loss": 0.0423, + "predictor_learning_rate": 7.554074789306006e-05, + "step": 7354 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1764, + "image_masks_loss": 0.0065, + "llm_learning_rate": 7.54813356542789e-07, + "loss": 0.7424, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.54813356542789e-05, + "step": 7355 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1764, + "image_masks_loss": 0.0136, + "llm_learning_rate": 7.542194263388175e-07, + "loss": 0.7442, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 7.542194263388174e-05, + "step": 7356 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1763, + "image_masks_loss": 0.0086, + "llm_learning_rate": 7.536256883840904e-07, + "loss": 0.7297, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 7.536256883840904e-05, + "step": 7357 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1763, + "image_masks_loss": 0.0051, + "llm_learning_rate": 7.530321427439935e-07, + "loss": 0.7598, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 7.530321427439935e-05, + "step": 7358 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1762, + "image_masks_loss": 0.0089, + "llm_learning_rate": 7.524387894838881e-07, + "loss": 0.6963, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 7.52438789483888e-05, + "step": 7359 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1762, + "image_masks_loss": 0.0101, + "llm_learning_rate": 7.518456286691156e-07, + "loss": 0.7057, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.518456286691156e-05, + "step": 7360 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1762, + "image_masks_loss": 0.0091, + "llm_learning_rate": 7.512526603649975e-07, + "loss": 0.7081, + "output_text_mask_loss": 0.0206, + "predictor_learning_rate": 7.512526603649975e-05, + "step": 7361 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1761, + "image_masks_loss": 0.0071, + "llm_learning_rate": 7.506598846368296e-07, + "loss": 0.7053, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 7.506598846368295e-05, + "step": 7362 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1761, + "image_masks_loss": 0.0082, + "llm_learning_rate": 7.500673015498927e-07, + "loss": 0.6467, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.500673015498927e-05, + "step": 7363 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.176, + "image_masks_loss": 0.0098, + "llm_learning_rate": 7.494749111694424e-07, + "loss": 0.7071, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 7.494749111694424e-05, + "step": 7364 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.176, + "image_masks_loss": 0.0034, + "llm_learning_rate": 7.488827135607141e-07, + "loss": 0.7984, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 7.488827135607141e-05, + "step": 7365 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.176, + "image_masks_loss": 0.0086, + "llm_learning_rate": 7.482907087889219e-07, + "loss": 0.7637, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 7.482907087889218e-05, + "step": 7366 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1759, + "image_masks_loss": 0.0133, + "llm_learning_rate": 7.476988969192584e-07, + "loss": 0.7488, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 7.476988969192583e-05, + "step": 7367 + }, + { + "epoch": 0.75, + "gumbel_tau": 0.1759, + "image_masks_loss": 0.0059, + "llm_learning_rate": 7.471072780168956e-07, + "loss": 0.7033, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.471072780168956e-05, + "step": 7368 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1758, + "image_masks_loss": 0.0109, + "llm_learning_rate": 7.465158521469829e-07, + "loss": 0.6377, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.465158521469828e-05, + "step": 7369 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1758, + "image_masks_loss": 0.0082, + "llm_learning_rate": 7.459246193746511e-07, + "loss": 0.7097, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 7.45924619374651e-05, + "step": 7370 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1757, + "image_masks_loss": 0.0105, + "llm_learning_rate": 7.453335797650072e-07, + "loss": 0.6551, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 7.453335797650071e-05, + "step": 7371 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1757, + "image_masks_loss": 0.0105, + "llm_learning_rate": 7.44742733383138e-07, + "loss": 0.7334, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.447427333831381e-05, + "step": 7372 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1757, + "image_masks_loss": 0.0189, + "llm_learning_rate": 7.441520802941085e-07, + "loss": 0.7934, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 7.441520802941085e-05, + "step": 7373 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1756, + "image_masks_loss": 0.0066, + "llm_learning_rate": 7.435616205629631e-07, + "loss": 0.763, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 7.435616205629631e-05, + "step": 7374 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1756, + "image_masks_loss": 0.0102, + "llm_learning_rate": 7.429713542547243e-07, + "loss": 0.6836, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 7.429713542547243e-05, + "step": 7375 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1755, + "image_masks_loss": 0.005, + "llm_learning_rate": 7.423812814343928e-07, + "loss": 0.7227, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.423812814343928e-05, + "step": 7376 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1755, + "image_masks_loss": 0.012, + "llm_learning_rate": 7.417914021669504e-07, + "loss": 0.739, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 7.417914021669503e-05, + "step": 7377 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1755, + "image_masks_loss": 0.0103, + "llm_learning_rate": 7.412017165173549e-07, + "loss": 0.8203, + "output_text_mask_loss": 0.0404, + "predictor_learning_rate": 7.412017165173548e-05, + "step": 7378 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1754, + "image_masks_loss": 0.0013, + "llm_learning_rate": 7.40612224550544e-07, + "loss": 0.7311, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.40612224550544e-05, + "step": 7379 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1754, + "image_masks_loss": 0.0082, + "llm_learning_rate": 7.400229263314346e-07, + "loss": 0.6091, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 7.400229263314345e-05, + "step": 7380 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1753, + "image_masks_loss": 0.0105, + "llm_learning_rate": 7.3943382192492e-07, + "loss": 0.5588, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 7.3943382192492e-05, + "step": 7381 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1753, + "image_masks_loss": 0.0126, + "llm_learning_rate": 7.388449113958738e-07, + "loss": 0.682, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.388449113958737e-05, + "step": 7382 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1752, + "image_masks_loss": 0.0079, + "llm_learning_rate": 7.382561948091493e-07, + "loss": 0.723, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.382561948091493e-05, + "step": 7383 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1752, + "image_masks_loss": 0.0106, + "llm_learning_rate": 7.376676722295767e-07, + "loss": 0.6785, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 7.376676722295766e-05, + "step": 7384 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1752, + "image_masks_loss": 0.0048, + "llm_learning_rate": 7.370793437219656e-07, + "loss": 0.762, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 7.370793437219655e-05, + "step": 7385 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1751, + "image_masks_loss": 0.0107, + "llm_learning_rate": 7.36491209351104e-07, + "loss": 0.6712, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 7.36491209351104e-05, + "step": 7386 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1751, + "image_masks_loss": 0.0083, + "llm_learning_rate": 7.359032691817586e-07, + "loss": 0.747, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 7.359032691817586e-05, + "step": 7387 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.175, + "image_masks_loss": 0.0074, + "llm_learning_rate": 7.353155232786743e-07, + "loss": 0.7079, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.353155232786743e-05, + "step": 7388 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.175, + "image_masks_loss": 0.0067, + "llm_learning_rate": 7.347279717065744e-07, + "loss": 0.766, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 7.347279717065744e-05, + "step": 7389 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.175, + "image_masks_loss": 0.0043, + "llm_learning_rate": 7.341406145301633e-07, + "loss": 0.585, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.341406145301633e-05, + "step": 7390 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1749, + "image_masks_loss": 0.0172, + "llm_learning_rate": 7.33553451814121e-07, + "loss": 0.7084, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.335534518141209e-05, + "step": 7391 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1749, + "image_masks_loss": 0.004, + "llm_learning_rate": 7.329664836231071e-07, + "loss": 0.7265, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.32966483623107e-05, + "step": 7392 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1748, + "image_masks_loss": 0.0034, + "llm_learning_rate": 7.3237971002176e-07, + "loss": 0.7273, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 7.3237971002176e-05, + "step": 7393 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1748, + "image_masks_loss": 0.0091, + "llm_learning_rate": 7.317931310746964e-07, + "loss": 0.7843, + "output_text_mask_loss": 0.0454, + "predictor_learning_rate": 7.317931310746964e-05, + "step": 7394 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1748, + "image_masks_loss": 0.0037, + "llm_learning_rate": 7.312067468465117e-07, + "loss": 0.7101, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 7.312067468465117e-05, + "step": 7395 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1747, + "image_masks_loss": 0.0069, + "llm_learning_rate": 7.306205574017791e-07, + "loss": 0.7667, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 7.306205574017791e-05, + "step": 7396 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1747, + "image_masks_loss": 0.0079, + "llm_learning_rate": 7.300345628050528e-07, + "loss": 0.7181, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 7.300345628050529e-05, + "step": 7397 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1746, + "image_masks_loss": 0.0039, + "llm_learning_rate": 7.294487631208636e-07, + "loss": 0.7555, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 7.294487631208636e-05, + "step": 7398 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1746, + "image_masks_loss": 0.0028, + "llm_learning_rate": 7.288631584137196e-07, + "loss": 0.8235, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 7.288631584137196e-05, + "step": 7399 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1745, + "image_masks_loss": 0.0138, + "llm_learning_rate": 7.282777487481094e-07, + "loss": 0.7096, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 7.282777487481094e-05, + "step": 7400 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1745, + "image_masks_loss": 0.0046, + "llm_learning_rate": 7.276925341884991e-07, + "loss": 0.6881, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 7.27692534188499e-05, + "step": 7401 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1745, + "image_masks_loss": 0.0097, + "llm_learning_rate": 7.271075147993356e-07, + "loss": 0.7994, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 7.271075147993355e-05, + "step": 7402 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1744, + "image_masks_loss": 0.0098, + "llm_learning_rate": 7.265226906450415e-07, + "loss": 0.7207, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.265226906450415e-05, + "step": 7403 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1744, + "image_masks_loss": 0.0028, + "llm_learning_rate": 7.259380617900188e-07, + "loss": 0.7203, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 7.259380617900188e-05, + "step": 7404 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1743, + "image_masks_loss": 0.0077, + "llm_learning_rate": 7.253536282986487e-07, + "loss": 0.6836, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.253536282986486e-05, + "step": 7405 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1743, + "image_masks_loss": 0.0057, + "llm_learning_rate": 7.247693902352895e-07, + "loss": 0.749, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 7.247693902352895e-05, + "step": 7406 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1743, + "image_masks_loss": 0.0042, + "llm_learning_rate": 7.241853476642793e-07, + "loss": 0.6861, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.241853476642793e-05, + "step": 7407 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1742, + "image_masks_loss": 0.0031, + "llm_learning_rate": 7.236015006499337e-07, + "loss": 0.6111, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 7.236015006499336e-05, + "step": 7408 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1742, + "image_masks_loss": 0.0049, + "llm_learning_rate": 7.230178492565479e-07, + "loss": 0.7452, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.230178492565479e-05, + "step": 7409 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1741, + "image_masks_loss": 0.0143, + "llm_learning_rate": 7.224343935483949e-07, + "loss": 0.7667, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 7.224343935483949e-05, + "step": 7410 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1741, + "image_masks_loss": 0.0073, + "llm_learning_rate": 7.218511335897255e-07, + "loss": 0.6735, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 7.218511335897255e-05, + "step": 7411 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1741, + "image_masks_loss": 0.0138, + "llm_learning_rate": 7.212680694447702e-07, + "loss": 0.7605, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 7.212680694447701e-05, + "step": 7412 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.174, + "image_masks_loss": 0.0105, + "llm_learning_rate": 7.206852011777371e-07, + "loss": 0.7186, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 7.206852011777371e-05, + "step": 7413 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.174, + "image_masks_loss": 0.0143, + "llm_learning_rate": 7.20102528852813e-07, + "loss": 0.7243, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 7.201025288528129e-05, + "step": 7414 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1739, + "image_masks_loss": 0.0118, + "llm_learning_rate": 7.19520052534162e-07, + "loss": 0.7887, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 7.19520052534162e-05, + "step": 7415 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1739, + "image_masks_loss": 0.0067, + "llm_learning_rate": 7.189377722859294e-07, + "loss": 0.6343, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.189377722859294e-05, + "step": 7416 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1738, + "image_masks_loss": 0.0153, + "llm_learning_rate": 7.183556881722375e-07, + "loss": 0.8203, + "output_text_mask_loss": 0.0011, + "predictor_learning_rate": 7.183556881722376e-05, + "step": 7417 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1738, + "image_masks_loss": 0.0021, + "llm_learning_rate": 7.177738002571849e-07, + "loss": 0.7299, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 7.177738002571848e-05, + "step": 7418 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1738, + "image_masks_loss": 0.0046, + "llm_learning_rate": 7.171921086048508e-07, + "loss": 0.7209, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.171921086048507e-05, + "step": 7419 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1737, + "image_masks_loss": 0.0062, + "llm_learning_rate": 7.166106132792932e-07, + "loss": 0.8125, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 7.166106132792932e-05, + "step": 7420 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1737, + "image_masks_loss": 0.0141, + "llm_learning_rate": 7.160293143445463e-07, + "loss": 0.6742, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.160293143445463e-05, + "step": 7421 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1736, + "image_masks_loss": 0.0056, + "llm_learning_rate": 7.154482118646255e-07, + "loss": 0.7024, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 7.154482118646254e-05, + "step": 7422 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1736, + "image_masks_loss": 0.0087, + "llm_learning_rate": 7.148673059035228e-07, + "loss": 0.7252, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 7.148673059035227e-05, + "step": 7423 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1736, + "image_masks_loss": 0.0062, + "llm_learning_rate": 7.142865965252086e-07, + "loss": 0.6838, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 7.142865965252085e-05, + "step": 7424 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1735, + "image_masks_loss": 0.005, + "llm_learning_rate": 7.137060837936316e-07, + "loss": 0.6971, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 7.137060837936315e-05, + "step": 7425 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1735, + "image_masks_loss": 0.0108, + "llm_learning_rate": 7.131257677727197e-07, + "loss": 0.7461, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.131257677727196e-05, + "step": 7426 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1734, + "image_masks_loss": 0.0072, + "llm_learning_rate": 7.125456485263785e-07, + "loss": 0.6653, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 7.125456485263784e-05, + "step": 7427 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1734, + "image_masks_loss": 0.0052, + "llm_learning_rate": 7.119657261184909e-07, + "loss": 0.6689, + "output_text_mask_loss": 0.0011, + "predictor_learning_rate": 7.119657261184909e-05, + "step": 7428 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1734, + "image_masks_loss": 0.009, + "llm_learning_rate": 7.113860006129209e-07, + "loss": 0.8171, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 7.113860006129209e-05, + "step": 7429 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1733, + "image_masks_loss": 0.01, + "llm_learning_rate": 7.108064720735086e-07, + "loss": 0.7515, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.108064720735085e-05, + "step": 7430 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1733, + "image_masks_loss": 0.0026, + "llm_learning_rate": 7.102271405640732e-07, + "loss": 0.8268, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 7.102271405640732e-05, + "step": 7431 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1732, + "image_masks_loss": 0.0054, + "llm_learning_rate": 7.096480061484112e-07, + "loss": 0.7647, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 7.096480061484111e-05, + "step": 7432 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1732, + "image_masks_loss": 0.0069, + "llm_learning_rate": 7.090690688902988e-07, + "loss": 0.724, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.090690688902987e-05, + "step": 7433 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1731, + "image_masks_loss": 0.0068, + "llm_learning_rate": 7.084903288534897e-07, + "loss": 0.6336, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 7.084903288534897e-05, + "step": 7434 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1731, + "image_masks_loss": 0.0113, + "llm_learning_rate": 7.07911786101716e-07, + "loss": 0.7363, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 7.07911786101716e-05, + "step": 7435 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1731, + "image_masks_loss": 0.0043, + "llm_learning_rate": 7.073334406986879e-07, + "loss": 0.7558, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 7.073334406986878e-05, + "step": 7436 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.173, + "image_masks_loss": 0.0057, + "llm_learning_rate": 7.067552927080945e-07, + "loss": 0.7706, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 7.067552927080944e-05, + "step": 7437 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.173, + "image_masks_loss": 0.009, + "llm_learning_rate": 7.061773421936027e-07, + "loss": 0.7987, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.061773421936027e-05, + "step": 7438 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1729, + "image_masks_loss": 0.0071, + "llm_learning_rate": 7.055995892188575e-07, + "loss": 0.7216, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.055995892188574e-05, + "step": 7439 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1729, + "image_masks_loss": 0.008, + "llm_learning_rate": 7.050220338474815e-07, + "loss": 0.7664, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.050220338474814e-05, + "step": 7440 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1729, + "image_masks_loss": 0.003, + "llm_learning_rate": 7.044446761430782e-07, + "loss": 0.8561, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 7.044446761430781e-05, + "step": 7441 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1728, + "image_masks_loss": 0.0111, + "llm_learning_rate": 7.038675161692265e-07, + "loss": 0.7483, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 7.038675161692264e-05, + "step": 7442 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1728, + "image_masks_loss": 0.0227, + "llm_learning_rate": 7.032905539894847e-07, + "loss": 0.8524, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 7.032905539894846e-05, + "step": 7443 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1727, + "image_masks_loss": 0.0069, + "llm_learning_rate": 7.027137896673891e-07, + "loss": 0.6926, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.027137896673891e-05, + "step": 7444 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1727, + "image_masks_loss": 0.0032, + "llm_learning_rate": 7.021372232664542e-07, + "loss": 0.7277, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 7.021372232664541e-05, + "step": 7445 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1727, + "image_masks_loss": 0.0059, + "llm_learning_rate": 7.015608548501729e-07, + "loss": 0.7402, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.015608548501728e-05, + "step": 7446 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1726, + "image_masks_loss": 0.0099, + "llm_learning_rate": 7.009846844820156e-07, + "loss": 0.684, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 7.009846844820155e-05, + "step": 7447 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1726, + "image_masks_loss": 0.0046, + "llm_learning_rate": 7.004087122254325e-07, + "loss": 0.6868, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 7.004087122254325e-05, + "step": 7448 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1725, + "image_masks_loss": 0.0028, + "llm_learning_rate": 6.998329381438504e-07, + "loss": 0.6897, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 6.998329381438504e-05, + "step": 7449 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1725, + "image_masks_loss": 0.0113, + "llm_learning_rate": 6.992573623006751e-07, + "loss": 0.7596, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 6.992573623006752e-05, + "step": 7450 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1725, + "image_masks_loss": 0.0089, + "llm_learning_rate": 6.986819847592899e-07, + "loss": 0.6665, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.986819847592899e-05, + "step": 7451 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1724, + "image_masks_loss": 0.0019, + "llm_learning_rate": 6.981068055830567e-07, + "loss": 0.7716, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 6.981068055830567e-05, + "step": 7452 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1724, + "image_masks_loss": 0.0031, + "llm_learning_rate": 6.975318248353158e-07, + "loss": 0.7157, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.975318248353158e-05, + "step": 7453 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1723, + "image_masks_loss": 0.0023, + "llm_learning_rate": 6.969570425793853e-07, + "loss": 0.742, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 6.969570425793854e-05, + "step": 7454 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1723, + "image_masks_loss": 0.0087, + "llm_learning_rate": 6.963824588785611e-07, + "loss": 0.7294, + "output_text_mask_loss": 0.0479, + "predictor_learning_rate": 6.963824588785611e-05, + "step": 7455 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1723, + "image_masks_loss": 0.0054, + "llm_learning_rate": 6.958080737961181e-07, + "loss": 0.7465, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 6.958080737961181e-05, + "step": 7456 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1722, + "image_masks_loss": 0.0061, + "llm_learning_rate": 6.952338873953088e-07, + "loss": 0.764, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 6.952338873953087e-05, + "step": 7457 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1722, + "image_masks_loss": 0.0034, + "llm_learning_rate": 6.946598997393636e-07, + "loss": 0.7052, + "output_text_mask_loss": 0.0486, + "predictor_learning_rate": 6.946598997393635e-05, + "step": 7458 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1721, + "image_masks_loss": 0.01, + "llm_learning_rate": 6.940861108914914e-07, + "loss": 0.7353, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 6.940861108914914e-05, + "step": 7459 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1721, + "image_masks_loss": 0.006, + "llm_learning_rate": 6.935125209148788e-07, + "loss": 0.703, + "output_text_mask_loss": 0.0372, + "predictor_learning_rate": 6.935125209148788e-05, + "step": 7460 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1721, + "image_masks_loss": 0.0133, + "llm_learning_rate": 6.929391298726917e-07, + "loss": 0.6793, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 6.929391298726918e-05, + "step": 7461 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.172, + "image_masks_loss": 0.0196, + "llm_learning_rate": 6.923659378280726e-07, + "loss": 0.7718, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 6.923659378280725e-05, + "step": 7462 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.172, + "image_masks_loss": 0.0078, + "llm_learning_rate": 6.91792944844143e-07, + "loss": 0.6899, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 6.91792944844143e-05, + "step": 7463 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1719, + "image_masks_loss": 0.0033, + "llm_learning_rate": 6.912201509840016e-07, + "loss": 0.7909, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 6.912201509840016e-05, + "step": 7464 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1719, + "image_masks_loss": 0.0042, + "llm_learning_rate": 6.906475563107264e-07, + "loss": 0.8361, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 6.906475563107265e-05, + "step": 7465 + }, + { + "epoch": 0.76, + "gumbel_tau": 0.1718, + "image_masks_loss": 0.0087, + "llm_learning_rate": 6.900751608873723e-07, + "loss": 0.7643, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 6.900751608873723e-05, + "step": 7466 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1718, + "image_masks_loss": 0.0175, + "llm_learning_rate": 6.895029647769724e-07, + "loss": 0.725, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 6.895029647769723e-05, + "step": 7467 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1718, + "image_masks_loss": 0.0093, + "llm_learning_rate": 6.889309680425394e-07, + "loss": 0.7026, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 6.889309680425394e-05, + "step": 7468 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1717, + "image_masks_loss": 0.0149, + "llm_learning_rate": 6.883591707470618e-07, + "loss": 0.77, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 6.883591707470619e-05, + "step": 7469 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1717, + "image_masks_loss": 0.0116, + "llm_learning_rate": 6.87787572953508e-07, + "loss": 0.7158, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 6.87787572953508e-05, + "step": 7470 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1716, + "image_masks_loss": 0.0069, + "llm_learning_rate": 6.872161747248232e-07, + "loss": 0.7529, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 6.872161747248232e-05, + "step": 7471 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1716, + "image_masks_loss": 0.0058, + "llm_learning_rate": 6.866449761239318e-07, + "loss": 0.774, + "output_text_mask_loss": 0.0219, + "predictor_learning_rate": 6.866449761239319e-05, + "step": 7472 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1716, + "image_masks_loss": 0.015, + "llm_learning_rate": 6.860739772137331e-07, + "loss": 0.7149, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.860739772137331e-05, + "step": 7473 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1715, + "image_masks_loss": 0.0061, + "llm_learning_rate": 6.855031780571092e-07, + "loss": 0.668, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 6.855031780571092e-05, + "step": 7474 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1715, + "image_masks_loss": 0.0046, + "llm_learning_rate": 6.84932578716917e-07, + "loss": 0.7476, + "output_text_mask_loss": 0.106, + "predictor_learning_rate": 6.849325787169169e-05, + "step": 7475 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1714, + "image_masks_loss": 0.0062, + "llm_learning_rate": 6.843621792559921e-07, + "loss": 0.7452, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 6.84362179255992e-05, + "step": 7476 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1714, + "image_masks_loss": 0.0055, + "llm_learning_rate": 6.837919797371481e-07, + "loss": 0.6884, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 6.83791979737148e-05, + "step": 7477 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1714, + "image_masks_loss": 0.0104, + "llm_learning_rate": 6.832219802231769e-07, + "loss": 0.6676, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.832219802231767e-05, + "step": 7478 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1713, + "image_masks_loss": 0.0093, + "llm_learning_rate": 6.82652180776848e-07, + "loss": 0.725, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 6.826521807768479e-05, + "step": 7479 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1713, + "image_masks_loss": 0.0078, + "llm_learning_rate": 6.82082581460908e-07, + "loss": 0.6513, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 6.82082581460908e-05, + "step": 7480 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1712, + "image_masks_loss": 0.0094, + "llm_learning_rate": 6.815131823380841e-07, + "loss": 0.7323, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 6.815131823380841e-05, + "step": 7481 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1712, + "image_masks_loss": 0.0108, + "llm_learning_rate": 6.809439834710793e-07, + "loss": 0.7879, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 6.809439834710793e-05, + "step": 7482 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1712, + "image_masks_loss": 0.0082, + "llm_learning_rate": 6.803749849225746e-07, + "loss": 0.6935, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 6.803749849225745e-05, + "step": 7483 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1711, + "image_masks_loss": 0.0092, + "llm_learning_rate": 6.7980618675523e-07, + "loss": 0.7429, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.7980618675523e-05, + "step": 7484 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1711, + "image_masks_loss": 0.0048, + "llm_learning_rate": 6.792375890316822e-07, + "loss": 0.611, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 6.79237589031682e-05, + "step": 7485 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.171, + "image_masks_loss": 0.0075, + "llm_learning_rate": 6.786691918145463e-07, + "loss": 0.7944, + "output_text_mask_loss": 0.0131, + "predictor_learning_rate": 6.786691918145463e-05, + "step": 7486 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.171, + "image_masks_loss": 0.0033, + "llm_learning_rate": 6.781009951664155e-07, + "loss": 0.7024, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 6.781009951664155e-05, + "step": 7487 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.171, + "image_masks_loss": 0.0059, + "llm_learning_rate": 6.77532999149862e-07, + "loss": 0.8019, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 6.775329991498618e-05, + "step": 7488 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1709, + "image_masks_loss": 0.0043, + "llm_learning_rate": 6.769652038274341e-07, + "loss": 0.6784, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.76965203827434e-05, + "step": 7489 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1709, + "image_masks_loss": 0.0037, + "llm_learning_rate": 6.763976092616583e-07, + "loss": 0.7236, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.763976092616583e-05, + "step": 7490 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1708, + "image_masks_loss": 0.0094, + "llm_learning_rate": 6.758302155150407e-07, + "loss": 0.7516, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 6.758302155150406e-05, + "step": 7491 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1708, + "image_masks_loss": 0.0158, + "llm_learning_rate": 6.752630226500614e-07, + "loss": 0.7859, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 6.752630226500615e-05, + "step": 7492 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1708, + "image_masks_loss": 0.0041, + "llm_learning_rate": 6.74696030729183e-07, + "loss": 0.752, + "output_text_mask_loss": 0.0197, + "predictor_learning_rate": 6.74696030729183e-05, + "step": 7493 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1707, + "image_masks_loss": 0.0107, + "llm_learning_rate": 6.741292398148436e-07, + "loss": 0.6452, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 6.741292398148435e-05, + "step": 7494 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1707, + "image_masks_loss": 0.0087, + "llm_learning_rate": 6.735626499694595e-07, + "loss": 0.8383, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 6.735626499694595e-05, + "step": 7495 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1706, + "image_masks_loss": 0.0076, + "llm_learning_rate": 6.729962612554242e-07, + "loss": 0.7636, + "output_text_mask_loss": 0.0595, + "predictor_learning_rate": 6.729962612554241e-05, + "step": 7496 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1706, + "image_masks_loss": 0.0048, + "llm_learning_rate": 6.724300737351103e-07, + "loss": 0.6333, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 6.724300737351103e-05, + "step": 7497 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1706, + "image_masks_loss": 0.0078, + "llm_learning_rate": 6.718640874708673e-07, + "loss": 0.6372, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.718640874708673e-05, + "step": 7498 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1705, + "image_masks_loss": 0.0058, + "llm_learning_rate": 6.712983025250225e-07, + "loss": 0.7451, + "output_text_mask_loss": 0.0538, + "predictor_learning_rate": 6.712983025250225e-05, + "step": 7499 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1705, + "image_masks_loss": 0.0056, + "llm_learning_rate": 6.707327189598825e-07, + "loss": 0.7152, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.707327189598824e-05, + "step": 7500 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1704, + "image_masks_loss": 0.0047, + "llm_learning_rate": 6.701673368377301e-07, + "loss": 0.6591, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 6.7016733683773e-05, + "step": 7501 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1704, + "image_masks_loss": 0.0085, + "llm_learning_rate": 6.696021562208263e-07, + "loss": 0.7989, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 6.696021562208263e-05, + "step": 7502 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1704, + "image_masks_loss": 0.0088, + "llm_learning_rate": 6.6903717717141e-07, + "loss": 0.7821, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 6.690371771714099e-05, + "step": 7503 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1703, + "image_masks_loss": 0.0068, + "llm_learning_rate": 6.684723997516982e-07, + "loss": 0.6889, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.684723997516981e-05, + "step": 7504 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1703, + "image_masks_loss": 0.0034, + "llm_learning_rate": 6.679078240238851e-07, + "loss": 0.7031, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 6.679078240238851e-05, + "step": 7505 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1702, + "image_masks_loss": 0.0036, + "llm_learning_rate": 6.673434500501428e-07, + "loss": 0.8307, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 6.673434500501426e-05, + "step": 7506 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1702, + "image_masks_loss": 0.0046, + "llm_learning_rate": 6.667792778926224e-07, + "loss": 0.6637, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.667792778926223e-05, + "step": 7507 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1702, + "image_masks_loss": 0.0142, + "llm_learning_rate": 6.662153076134515e-07, + "loss": 0.7271, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.662153076134514e-05, + "step": 7508 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1701, + "image_masks_loss": 0.0103, + "llm_learning_rate": 6.656515392747354e-07, + "loss": 0.6894, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 6.656515392747353e-05, + "step": 7509 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1701, + "image_masks_loss": 0.0058, + "llm_learning_rate": 6.650879729385581e-07, + "loss": 0.7801, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 6.650879729385581e-05, + "step": 7510 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.17, + "image_masks_loss": 0.005, + "llm_learning_rate": 6.645246086669799e-07, + "loss": 0.7784, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 6.645246086669799e-05, + "step": 7511 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.17, + "image_masks_loss": 0.0034, + "llm_learning_rate": 6.639614465220392e-07, + "loss": 0.7519, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 6.639614465220392e-05, + "step": 7512 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.17, + "image_masks_loss": 0.0083, + "llm_learning_rate": 6.633984865657547e-07, + "loss": 0.8424, + "output_text_mask_loss": 0.0144, + "predictor_learning_rate": 6.633984865657546e-05, + "step": 7513 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1699, + "image_masks_loss": 0.0063, + "llm_learning_rate": 6.628357288601195e-07, + "loss": 0.6983, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 6.628357288601195e-05, + "step": 7514 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1699, + "image_masks_loss": 0.0093, + "llm_learning_rate": 6.622731734671059e-07, + "loss": 0.8366, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 6.622731734671058e-05, + "step": 7515 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1698, + "image_masks_loss": 0.0084, + "llm_learning_rate": 6.617108204486642e-07, + "loss": 0.7544, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 6.617108204486641e-05, + "step": 7516 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1698, + "image_masks_loss": 0.0053, + "llm_learning_rate": 6.61148669866721e-07, + "loss": 0.7561, + "output_text_mask_loss": 0.2151, + "predictor_learning_rate": 6.61148669866721e-05, + "step": 7517 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1698, + "image_masks_loss": 0.0075, + "llm_learning_rate": 6.605867217831826e-07, + "loss": 0.7066, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.605867217831827e-05, + "step": 7518 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1697, + "image_masks_loss": 0.0051, + "llm_learning_rate": 6.600249762599306e-07, + "loss": 0.7074, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.600249762599306e-05, + "step": 7519 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1697, + "image_masks_loss": 0.003, + "llm_learning_rate": 6.594634333588273e-07, + "loss": 0.8607, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 6.594634333588274e-05, + "step": 7520 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1696, + "image_masks_loss": 0.0071, + "llm_learning_rate": 6.589020931417106e-07, + "loss": 0.6941, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 6.589020931417105e-05, + "step": 7521 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1696, + "image_masks_loss": 0.0057, + "llm_learning_rate": 6.58340955670396e-07, + "loss": 0.748, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.583409556703959e-05, + "step": 7522 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1696, + "image_masks_loss": 0.0092, + "llm_learning_rate": 6.577800210066776e-07, + "loss": 0.7049, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.577800210066775e-05, + "step": 7523 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1695, + "image_masks_loss": 0.0114, + "llm_learning_rate": 6.572192892123266e-07, + "loss": 0.7192, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 6.572192892123266e-05, + "step": 7524 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1695, + "image_masks_loss": 0.0113, + "llm_learning_rate": 6.566587603490924e-07, + "loss": 0.7979, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 6.566587603490922e-05, + "step": 7525 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1694, + "image_masks_loss": 0.0089, + "llm_learning_rate": 6.560984344787002e-07, + "loss": 0.6204, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 6.560984344787002e-05, + "step": 7526 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1694, + "image_masks_loss": 0.0065, + "llm_learning_rate": 6.555383116628566e-07, + "loss": 0.7395, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.555383116628566e-05, + "step": 7527 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1694, + "image_masks_loss": 0.0085, + "llm_learning_rate": 6.549783919632433e-07, + "loss": 0.8148, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 6.549783919632432e-05, + "step": 7528 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1693, + "image_masks_loss": 0.015, + "llm_learning_rate": 6.54418675441518e-07, + "loss": 0.7728, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 6.544186754415179e-05, + "step": 7529 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1693, + "image_masks_loss": 0.0028, + "llm_learning_rate": 6.538591621593194e-07, + "loss": 0.7384, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 6.538591621593192e-05, + "step": 7530 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1692, + "image_masks_loss": 0.0093, + "llm_learning_rate": 6.53299852178261e-07, + "loss": 0.7391, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 6.53299852178261e-05, + "step": 7531 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1692, + "image_masks_loss": 0.0042, + "llm_learning_rate": 6.527407455599372e-07, + "loss": 0.7342, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 6.527407455599372e-05, + "step": 7532 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1692, + "image_masks_loss": 0.007, + "llm_learning_rate": 6.521818423659168e-07, + "loss": 0.7659, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 6.521818423659168e-05, + "step": 7533 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1691, + "image_masks_loss": 0.0066, + "llm_learning_rate": 6.51623142657748e-07, + "loss": 0.7342, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 6.516231426577479e-05, + "step": 7534 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1691, + "image_masks_loss": 0.003, + "llm_learning_rate": 6.510646464969558e-07, + "loss": 0.7672, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 6.510646464969557e-05, + "step": 7535 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.169, + "image_masks_loss": 0.0066, + "llm_learning_rate": 6.505063539450429e-07, + "loss": 0.7774, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 6.505063539450429e-05, + "step": 7536 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.169, + "image_masks_loss": 0.0094, + "llm_learning_rate": 6.4994826506349e-07, + "loss": 0.8128, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 6.499482650634899e-05, + "step": 7537 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.169, + "image_masks_loss": 0.0095, + "llm_learning_rate": 6.493903799137544e-07, + "loss": 0.6781, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 6.493903799137544e-05, + "step": 7538 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1689, + "image_masks_loss": 0.0021, + "llm_learning_rate": 6.488326985572727e-07, + "loss": 0.708, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 6.488326985572726e-05, + "step": 7539 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1689, + "image_masks_loss": 0.0047, + "llm_learning_rate": 6.482752210554577e-07, + "loss": 0.7408, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 6.482752210554576e-05, + "step": 7540 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1688, + "image_masks_loss": 0.0032, + "llm_learning_rate": 6.477179474696998e-07, + "loss": 0.7559, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 6.477179474696998e-05, + "step": 7541 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1688, + "image_masks_loss": 0.0027, + "llm_learning_rate": 6.471608778613675e-07, + "loss": 0.77, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 6.471608778613674e-05, + "step": 7542 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1688, + "image_masks_loss": 0.0041, + "llm_learning_rate": 6.466040122918063e-07, + "loss": 0.7149, + "output_text_mask_loss": 0.0801, + "predictor_learning_rate": 6.466040122918062e-05, + "step": 7543 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1687, + "image_masks_loss": 0.011, + "llm_learning_rate": 6.460473508223397e-07, + "loss": 0.7439, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 6.460473508223397e-05, + "step": 7544 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1687, + "image_masks_loss": 0.0082, + "llm_learning_rate": 6.454908935142673e-07, + "loss": 0.6654, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 6.454908935142672e-05, + "step": 7545 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1686, + "image_masks_loss": 0.0154, + "llm_learning_rate": 6.449346404288694e-07, + "loss": 0.7504, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 6.449346404288694e-05, + "step": 7546 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1686, + "image_masks_loss": 0.0057, + "llm_learning_rate": 6.443785916274015e-07, + "loss": 0.7861, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 6.443785916274015e-05, + "step": 7547 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1686, + "image_masks_loss": 0.0085, + "llm_learning_rate": 6.438227471710959e-07, + "loss": 0.7473, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 6.438227471710958e-05, + "step": 7548 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1685, + "image_masks_loss": 0.0203, + "llm_learning_rate": 6.432671071211636e-07, + "loss": 0.67, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 6.432671071211635e-05, + "step": 7549 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1685, + "image_masks_loss": 0.0059, + "llm_learning_rate": 6.42711671538793e-07, + "loss": 0.7504, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 6.427116715387929e-05, + "step": 7550 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1684, + "image_masks_loss": 0.0089, + "llm_learning_rate": 6.421564404851496e-07, + "loss": 0.7478, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.421564404851495e-05, + "step": 7551 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1684, + "image_masks_loss": 0.0037, + "llm_learning_rate": 6.416014140213778e-07, + "loss": 0.7149, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 6.416014140213777e-05, + "step": 7552 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1684, + "image_masks_loss": 0.0038, + "llm_learning_rate": 6.410465922085979e-07, + "loss": 0.5885, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 6.410465922085978e-05, + "step": 7553 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1683, + "image_masks_loss": 0.0071, + "llm_learning_rate": 6.404919751079075e-07, + "loss": 0.7388, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 6.404919751079075e-05, + "step": 7554 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1683, + "image_masks_loss": 0.0148, + "llm_learning_rate": 6.399375627803831e-07, + "loss": 0.7602, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 6.39937562780383e-05, + "step": 7555 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1682, + "image_masks_loss": 0.0062, + "llm_learning_rate": 6.393833552870773e-07, + "loss": 0.6236, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 6.393833552870773e-05, + "step": 7556 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1682, + "image_masks_loss": 0.0069, + "llm_learning_rate": 6.388293526890207e-07, + "loss": 0.5855, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 6.388293526890207e-05, + "step": 7557 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1682, + "image_masks_loss": 0.0115, + "llm_learning_rate": 6.382755550472205e-07, + "loss": 0.5395, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.382755550472205e-05, + "step": 7558 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1681, + "image_masks_loss": 0.007, + "llm_learning_rate": 6.377219624226641e-07, + "loss": 0.7322, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 6.37721962422664e-05, + "step": 7559 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.1681, + "image_masks_loss": 0.0089, + "llm_learning_rate": 6.371685748763131e-07, + "loss": 0.7192, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 6.371685748763131e-05, + "step": 7560 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.168, + "image_masks_loss": 0.0082, + "llm_learning_rate": 6.366153924691079e-07, + "loss": 0.7628, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 6.366153924691078e-05, + "step": 7561 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.168, + "image_masks_loss": 0.0173, + "llm_learning_rate": 6.360624152619663e-07, + "loss": 0.6995, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 6.360624152619663e-05, + "step": 7562 + }, + { + "epoch": 0.77, + "gumbel_tau": 0.168, + "image_masks_loss": 0.0073, + "llm_learning_rate": 6.355096433157831e-07, + "loss": 0.64, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.355096433157831e-05, + "step": 7563 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1679, + "image_masks_loss": 0.0032, + "llm_learning_rate": 6.349570766914312e-07, + "loss": 0.7445, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 6.349570766914312e-05, + "step": 7564 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1679, + "image_masks_loss": 0.0087, + "llm_learning_rate": 6.344047154497599e-07, + "loss": 0.7266, + "output_text_mask_loss": 0.0315, + "predictor_learning_rate": 6.344047154497597e-05, + "step": 7565 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1678, + "image_masks_loss": 0.0073, + "llm_learning_rate": 6.338525596515971e-07, + "loss": 0.6893, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 6.33852559651597e-05, + "step": 7566 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1678, + "image_masks_loss": 0.0173, + "llm_learning_rate": 6.33300609357747e-07, + "loss": 0.7927, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 6.33300609357747e-05, + "step": 7567 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1678, + "image_masks_loss": 0.0073, + "llm_learning_rate": 6.327488646289917e-07, + "loss": 0.6471, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.327488646289917e-05, + "step": 7568 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1677, + "image_masks_loss": 0.0082, + "llm_learning_rate": 6.321973255260905e-07, + "loss": 0.7824, + "output_text_mask_loss": 0.0127, + "predictor_learning_rate": 6.321973255260905e-05, + "step": 7569 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1677, + "image_masks_loss": 0.0104, + "llm_learning_rate": 6.316459921097795e-07, + "loss": 0.723, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 6.316459921097795e-05, + "step": 7570 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1676, + "image_masks_loss": 0.0094, + "llm_learning_rate": 6.310948644407744e-07, + "loss": 0.7495, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 6.310948644407744e-05, + "step": 7571 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1676, + "image_masks_loss": 0.0168, + "llm_learning_rate": 6.305439425797655e-07, + "loss": 0.6921, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 6.305439425797655e-05, + "step": 7572 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1676, + "image_masks_loss": 0.0063, + "llm_learning_rate": 6.299932265874215e-07, + "loss": 0.7256, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 6.299932265874214e-05, + "step": 7573 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1675, + "image_masks_loss": 0.0073, + "llm_learning_rate": 6.294427165243888e-07, + "loss": 0.7084, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.294427165243888e-05, + "step": 7574 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1675, + "image_masks_loss": 0.0023, + "llm_learning_rate": 6.288924124512907e-07, + "loss": 0.7558, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 6.288924124512907e-05, + "step": 7575 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1674, + "image_masks_loss": 0.0117, + "llm_learning_rate": 6.283423144287279e-07, + "loss": 0.6346, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 6.283423144287279e-05, + "step": 7576 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1674, + "image_masks_loss": 0.0033, + "llm_learning_rate": 6.277924225172777e-07, + "loss": 0.7362, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 6.277924225172777e-05, + "step": 7577 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1674, + "image_masks_loss": 0.0063, + "llm_learning_rate": 6.272427367774966e-07, + "loss": 0.7183, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 6.272427367774966e-05, + "step": 7578 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1673, + "image_masks_loss": 0.0132, + "llm_learning_rate": 6.26693257269917e-07, + "loss": 0.7278, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.266932572699169e-05, + "step": 7579 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1673, + "image_masks_loss": 0.0111, + "llm_learning_rate": 6.261439840550485e-07, + "loss": 0.7187, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.261439840550484e-05, + "step": 7580 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1672, + "image_masks_loss": 0.0064, + "llm_learning_rate": 6.255949171933781e-07, + "loss": 0.7518, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.255949171933781e-05, + "step": 7581 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1672, + "image_masks_loss": 0.0043, + "llm_learning_rate": 6.250460567453706e-07, + "loss": 0.7279, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.250460567453705e-05, + "step": 7582 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1672, + "image_masks_loss": 0.0104, + "llm_learning_rate": 6.244974027714676e-07, + "loss": 0.6381, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 6.244974027714675e-05, + "step": 7583 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1671, + "image_masks_loss": 0.0014, + "llm_learning_rate": 6.239489553320883e-07, + "loss": 0.6679, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.239489553320881e-05, + "step": 7584 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1671, + "image_masks_loss": 0.0137, + "llm_learning_rate": 6.234007144876286e-07, + "loss": 0.6858, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 6.234007144876286e-05, + "step": 7585 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1671, + "image_masks_loss": 0.0085, + "llm_learning_rate": 6.228526802984622e-07, + "loss": 0.768, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 6.228526802984622e-05, + "step": 7586 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.167, + "image_masks_loss": 0.0125, + "llm_learning_rate": 6.223048528249398e-07, + "loss": 0.6881, + "output_text_mask_loss": 0.0206, + "predictor_learning_rate": 6.223048528249398e-05, + "step": 7587 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.167, + "image_masks_loss": 0.0122, + "llm_learning_rate": 6.217572321273896e-07, + "loss": 0.7015, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 6.217572321273896e-05, + "step": 7588 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1669, + "image_masks_loss": 0.0098, + "llm_learning_rate": 6.212098182661164e-07, + "loss": 0.7494, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 6.212098182661164e-05, + "step": 7589 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1669, + "image_masks_loss": 0.0152, + "llm_learning_rate": 6.206626113014022e-07, + "loss": 0.7342, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 6.206626113014021e-05, + "step": 7590 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1669, + "image_masks_loss": 0.0034, + "llm_learning_rate": 6.20115611293508e-07, + "loss": 0.734, + "output_text_mask_loss": 0.0839, + "predictor_learning_rate": 6.20115611293508e-05, + "step": 7591 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1668, + "image_masks_loss": 0.0216, + "llm_learning_rate": 6.195688183026702e-07, + "loss": 0.666, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 6.195688183026701e-05, + "step": 7592 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1668, + "image_masks_loss": 0.0051, + "llm_learning_rate": 6.190222323891024e-07, + "loss": 0.7756, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 6.190222323891023e-05, + "step": 7593 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1667, + "image_masks_loss": 0.0074, + "llm_learning_rate": 6.184758536129962e-07, + "loss": 0.7395, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 6.184758536129961e-05, + "step": 7594 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1667, + "image_masks_loss": 0.0045, + "llm_learning_rate": 6.179296820345199e-07, + "loss": 0.7661, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.179296820345198e-05, + "step": 7595 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1667, + "image_masks_loss": 0.0061, + "llm_learning_rate": 6.173837177138189e-07, + "loss": 0.7169, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 6.173837177138189e-05, + "step": 7596 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1666, + "image_masks_loss": 0.0207, + "llm_learning_rate": 6.168379607110159e-07, + "loss": 0.7449, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 6.168379607110159e-05, + "step": 7597 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1666, + "image_masks_loss": 0.0032, + "llm_learning_rate": 6.162924110862118e-07, + "loss": 0.6379, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.162924110862118e-05, + "step": 7598 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1665, + "image_masks_loss": 0.0049, + "llm_learning_rate": 6.157470688994832e-07, + "loss": 0.6984, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.157470688994831e-05, + "step": 7599 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1665, + "image_masks_loss": 0.004, + "llm_learning_rate": 6.152019342108845e-07, + "loss": 0.7625, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 6.152019342108844e-05, + "step": 7600 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1665, + "image_masks_loss": 0.0047, + "llm_learning_rate": 6.14657007080447e-07, + "loss": 0.6659, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 6.14657007080447e-05, + "step": 7601 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1664, + "image_masks_loss": 0.0078, + "llm_learning_rate": 6.141122875681791e-07, + "loss": 0.716, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 6.14112287568179e-05, + "step": 7602 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1664, + "image_masks_loss": 0.0084, + "llm_learning_rate": 6.135677757340668e-07, + "loss": 0.6911, + "output_text_mask_loss": 0.0755, + "predictor_learning_rate": 6.135677757340669e-05, + "step": 7603 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1663, + "image_masks_loss": 0.0042, + "llm_learning_rate": 6.130234716380731e-07, + "loss": 0.6967, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 6.130234716380731e-05, + "step": 7604 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1663, + "image_masks_loss": 0.0115, + "llm_learning_rate": 6.124793753401378e-07, + "loss": 0.7588, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 6.124793753401378e-05, + "step": 7605 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1663, + "image_masks_loss": 0.0069, + "llm_learning_rate": 6.119354869001778e-07, + "loss": 0.763, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 6.119354869001779e-05, + "step": 7606 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1662, + "image_masks_loss": 0.0088, + "llm_learning_rate": 6.113918063780875e-07, + "loss": 0.7147, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.113918063780873e-05, + "step": 7607 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1662, + "image_masks_loss": 0.0109, + "llm_learning_rate": 6.108483338337382e-07, + "loss": 0.7963, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 6.108483338337381e-05, + "step": 7608 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1661, + "image_masks_loss": 0.0059, + "llm_learning_rate": 6.103050693269777e-07, + "loss": 0.7781, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.103050693269777e-05, + "step": 7609 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1661, + "image_masks_loss": 0.0081, + "llm_learning_rate": 6.097620129176326e-07, + "loss": 0.7355, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 6.097620129176326e-05, + "step": 7610 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1661, + "image_masks_loss": 0.0088, + "llm_learning_rate": 6.092191646655052e-07, + "loss": 0.7319, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 6.092191646655051e-05, + "step": 7611 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.166, + "image_masks_loss": 0.0149, + "llm_learning_rate": 6.086765246303747e-07, + "loss": 0.7162, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.086765246303747e-05, + "step": 7612 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.166, + "image_masks_loss": 0.0083, + "llm_learning_rate": 6.081340928719981e-07, + "loss": 0.6634, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 6.08134092871998e-05, + "step": 7613 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.166, + "image_masks_loss": 0.0013, + "llm_learning_rate": 6.075918694501092e-07, + "loss": 0.8305, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 6.075918694501092e-05, + "step": 7614 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1659, + "image_masks_loss": 0.0065, + "llm_learning_rate": 6.070498544244191e-07, + "loss": 0.7419, + "output_text_mask_loss": 0.0196, + "predictor_learning_rate": 6.070498544244191e-05, + "step": 7615 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1659, + "image_masks_loss": 0.0077, + "llm_learning_rate": 6.065080478546145e-07, + "loss": 0.7149, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 6.065080478546145e-05, + "step": 7616 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1658, + "image_masks_loss": 0.0039, + "llm_learning_rate": 6.059664498003617e-07, + "loss": 0.64, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.059664498003617e-05, + "step": 7617 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1658, + "image_masks_loss": 0.0037, + "llm_learning_rate": 6.054250603213027e-07, + "loss": 0.724, + "output_text_mask_loss": 0.029, + "predictor_learning_rate": 6.0542506032130265e-05, + "step": 7618 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1658, + "image_masks_loss": 0.0089, + "llm_learning_rate": 6.048838794770556e-07, + "loss": 0.7862, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 6.048838794770556e-05, + "step": 7619 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1657, + "image_masks_loss": 0.0142, + "llm_learning_rate": 6.043429073272173e-07, + "loss": 0.6572, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.043429073272172e-05, + "step": 7620 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1657, + "image_masks_loss": 0.0116, + "llm_learning_rate": 6.03802143931361e-07, + "loss": 0.6716, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.03802143931361e-05, + "step": 7621 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1656, + "image_masks_loss": 0.0084, + "llm_learning_rate": 6.032615893490349e-07, + "loss": 0.6914, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 6.032615893490348e-05, + "step": 7622 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1656, + "image_masks_loss": 0.0045, + "llm_learning_rate": 6.027212436397678e-07, + "loss": 0.7213, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 6.0272124363976774e-05, + "step": 7623 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1656, + "image_masks_loss": 0.0188, + "llm_learning_rate": 6.021811068630637e-07, + "loss": 0.7591, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 6.0218110686306365e-05, + "step": 7624 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1655, + "image_masks_loss": 0.0093, + "llm_learning_rate": 6.016411790784032e-07, + "loss": 0.7562, + "output_text_mask_loss": 0.0614, + "predictor_learning_rate": 6.016411790784032e-05, + "step": 7625 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1655, + "image_masks_loss": 0.0088, + "llm_learning_rate": 6.011014603452444e-07, + "loss": 0.7476, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 6.0110146034524444e-05, + "step": 7626 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1654, + "image_masks_loss": 0.0104, + "llm_learning_rate": 6.005619507230226e-07, + "loss": 0.6999, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.0056195072302254e-05, + "step": 7627 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1654, + "image_masks_loss": 0.0193, + "llm_learning_rate": 6.000226502711495e-07, + "loss": 0.8067, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 6.000226502711495e-05, + "step": 7628 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1654, + "image_masks_loss": 0.0039, + "llm_learning_rate": 5.994835590490136e-07, + "loss": 0.7151, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.994835590490136e-05, + "step": 7629 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1653, + "image_masks_loss": 0.0149, + "llm_learning_rate": 5.989446771159821e-07, + "loss": 0.7798, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 5.9894467711598203e-05, + "step": 7630 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1653, + "image_masks_loss": 0.0061, + "llm_learning_rate": 5.984060045313972e-07, + "loss": 0.694, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 5.9840600453139714e-05, + "step": 7631 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1652, + "image_masks_loss": 0.0062, + "llm_learning_rate": 5.978675413545784e-07, + "loss": 0.6402, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.978675413545784e-05, + "step": 7632 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1652, + "image_masks_loss": 0.003, + "llm_learning_rate": 5.97329287644823e-07, + "loss": 0.6949, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 5.9732928764482304e-05, + "step": 7633 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1652, + "image_masks_loss": 0.0065, + "llm_learning_rate": 5.967912434614045e-07, + "loss": 0.6405, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 5.967912434614045e-05, + "step": 7634 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1651, + "image_masks_loss": 0.008, + "llm_learning_rate": 5.962534088635738e-07, + "loss": 0.6929, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 5.9625340886357376e-05, + "step": 7635 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1651, + "image_masks_loss": 0.0051, + "llm_learning_rate": 5.957157839105571e-07, + "loss": 0.7917, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 5.95715783910557e-05, + "step": 7636 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1651, + "image_masks_loss": 0.0047, + "llm_learning_rate": 5.951783686615606e-07, + "loss": 0.7845, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.951783686615606e-05, + "step": 7637 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.165, + "image_masks_loss": 0.0088, + "llm_learning_rate": 5.946411631757651e-07, + "loss": 0.6981, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 5.946411631757651e-05, + "step": 7638 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.165, + "image_masks_loss": 0.0061, + "llm_learning_rate": 5.941041675123294e-07, + "loss": 0.6792, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 5.941041675123293e-05, + "step": 7639 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1649, + "image_masks_loss": 0.0048, + "llm_learning_rate": 5.935673817303872e-07, + "loss": 0.745, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.935673817303872e-05, + "step": 7640 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1649, + "image_masks_loss": 0.007, + "llm_learning_rate": 5.930308058890505e-07, + "loss": 0.7323, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 5.930308058890505e-05, + "step": 7641 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1649, + "image_masks_loss": 0.0037, + "llm_learning_rate": 5.9249444004741e-07, + "loss": 0.7306, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.9249444004741e-05, + "step": 7642 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1648, + "image_masks_loss": 0.0038, + "llm_learning_rate": 5.919582842645305e-07, + "loss": 0.7655, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 5.9195828426453045e-05, + "step": 7643 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1648, + "image_masks_loss": 0.0149, + "llm_learning_rate": 5.914223385994547e-07, + "loss": 0.6732, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 5.9142233859945464e-05, + "step": 7644 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1647, + "image_masks_loss": 0.007, + "llm_learning_rate": 5.908866031112023e-07, + "loss": 0.7419, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 5.908866031112023e-05, + "step": 7645 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1647, + "image_masks_loss": 0.0065, + "llm_learning_rate": 5.903510778587695e-07, + "loss": 0.7029, + "output_text_mask_loss": 0.1175, + "predictor_learning_rate": 5.903510778587695e-05, + "step": 7646 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1647, + "image_masks_loss": 0.0084, + "llm_learning_rate": 5.898157629011295e-07, + "loss": 0.7657, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.8981576290112944e-05, + "step": 7647 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1646, + "image_masks_loss": 0.0029, + "llm_learning_rate": 5.892806582972319e-07, + "loss": 0.7476, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 5.8928065829723185e-05, + "step": 7648 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1646, + "image_masks_loss": 0.0175, + "llm_learning_rate": 5.887457641060045e-07, + "loss": 0.6338, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 5.8874576410600456e-05, + "step": 7649 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1645, + "image_masks_loss": 0.0041, + "llm_learning_rate": 5.882110803863511e-07, + "loss": 0.7598, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 5.8821108038635104e-05, + "step": 7650 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1645, + "image_masks_loss": 0.0066, + "llm_learning_rate": 5.876766071971518e-07, + "loss": 0.7017, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 5.8767660719715175e-05, + "step": 7651 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1645, + "image_masks_loss": 0.0021, + "llm_learning_rate": 5.87142344597264e-07, + "loss": 0.7019, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 5.87142344597264e-05, + "step": 7652 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1644, + "image_masks_loss": 0.0112, + "llm_learning_rate": 5.866082926455218e-07, + "loss": 0.6755, + "output_text_mask_loss": 0.0092, + "predictor_learning_rate": 5.8660829264552186e-05, + "step": 7653 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1644, + "image_masks_loss": 0.0094, + "llm_learning_rate": 5.860744514007366e-07, + "loss": 0.7392, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.8607445140073656e-05, + "step": 7654 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1644, + "image_masks_loss": 0.0098, + "llm_learning_rate": 5.855408209216951e-07, + "loss": 0.7689, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 5.8554082092169504e-05, + "step": 7655 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1643, + "image_masks_loss": 0.0099, + "llm_learning_rate": 5.850074012671634e-07, + "loss": 0.7021, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.850074012671633e-05, + "step": 7656 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1643, + "image_masks_loss": 0.0043, + "llm_learning_rate": 5.844741924958818e-07, + "loss": 0.7175, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 5.844741924958818e-05, + "step": 7657 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1642, + "image_masks_loss": 0.0148, + "llm_learning_rate": 5.839411946665699e-07, + "loss": 0.668, + "output_text_mask_loss": 0.0242, + "predictor_learning_rate": 5.839411946665699e-05, + "step": 7658 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1642, + "image_masks_loss": 0.0062, + "llm_learning_rate": 5.834084078379207e-07, + "loss": 0.7789, + "output_text_mask_loss": 0.0265, + "predictor_learning_rate": 5.834084078379206e-05, + "step": 7659 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1642, + "image_masks_loss": 0.0077, + "llm_learning_rate": 5.828758320686068e-07, + "loss": 0.7003, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 5.828758320686067e-05, + "step": 7660 + }, + { + "epoch": 0.78, + "gumbel_tau": 0.1641, + "image_masks_loss": 0.0089, + "llm_learning_rate": 5.823434674172757e-07, + "loss": 0.8453, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 5.823434674172756e-05, + "step": 7661 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1641, + "image_masks_loss": 0.0042, + "llm_learning_rate": 5.818113139425541e-07, + "loss": 0.7873, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 5.8181131394255404e-05, + "step": 7662 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.164, + "image_masks_loss": 0.0044, + "llm_learning_rate": 5.812793717030429e-07, + "loss": 0.7328, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 5.812793717030429e-05, + "step": 7663 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.164, + "image_masks_loss": 0.0069, + "llm_learning_rate": 5.807476407573215e-07, + "loss": 0.6758, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 5.807476407573214e-05, + "step": 7664 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.164, + "image_masks_loss": 0.007, + "llm_learning_rate": 5.802161211639445e-07, + "loss": 0.7984, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 5.802161211639445e-05, + "step": 7665 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1639, + "image_masks_loss": 0.0051, + "llm_learning_rate": 5.796848129814448e-07, + "loss": 0.6605, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 5.796848129814447e-05, + "step": 7666 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1639, + "image_masks_loss": 0.0063, + "llm_learning_rate": 5.791537162683305e-07, + "loss": 0.67, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 5.791537162683305e-05, + "step": 7667 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1639, + "image_masks_loss": 0.0021, + "llm_learning_rate": 5.786228310830869e-07, + "loss": 0.7434, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.7862283108308683e-05, + "step": 7668 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1638, + "image_masks_loss": 0.0044, + "llm_learning_rate": 5.780921574841774e-07, + "loss": 0.7401, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 5.780921574841774e-05, + "step": 7669 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1638, + "image_masks_loss": 0.0049, + "llm_learning_rate": 5.775616955300406e-07, + "loss": 0.7698, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 5.775616955300406e-05, + "step": 7670 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1637, + "image_masks_loss": 0.0124, + "llm_learning_rate": 5.77031445279092e-07, + "loss": 0.7179, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 5.77031445279092e-05, + "step": 7671 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1637, + "image_masks_loss": 0.0159, + "llm_learning_rate": 5.765014067897237e-07, + "loss": 0.7068, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 5.765014067897237e-05, + "step": 7672 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1637, + "image_masks_loss": 0.008, + "llm_learning_rate": 5.75971580120305e-07, + "loss": 0.6699, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 5.75971580120305e-05, + "step": 7673 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1636, + "image_masks_loss": 0.0057, + "llm_learning_rate": 5.754419653291815e-07, + "loss": 0.6126, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.7544196532918144e-05, + "step": 7674 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1636, + "image_masks_loss": 0.0093, + "llm_learning_rate": 5.749125624746749e-07, + "loss": 0.7506, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.749125624746748e-05, + "step": 7675 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1635, + "image_masks_loss": 0.0067, + "llm_learning_rate": 5.743833716150862e-07, + "loss": 0.7728, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 5.743833716150862e-05, + "step": 7676 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1635, + "image_masks_loss": 0.0093, + "llm_learning_rate": 5.738543928086892e-07, + "loss": 0.6741, + "output_text_mask_loss": 0.0469, + "predictor_learning_rate": 5.738543928086892e-05, + "step": 7677 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1635, + "image_masks_loss": 0.0115, + "llm_learning_rate": 5.733256261137366e-07, + "loss": 0.709, + "output_text_mask_loss": 0.0423, + "predictor_learning_rate": 5.733256261137365e-05, + "step": 7678 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1634, + "image_masks_loss": 0.009, + "llm_learning_rate": 5.727970715884576e-07, + "loss": 0.6358, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 5.7279707158845754e-05, + "step": 7679 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1634, + "image_masks_loss": 0.0068, + "llm_learning_rate": 5.722687292910572e-07, + "loss": 0.752, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.722687292910572e-05, + "step": 7680 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1633, + "image_masks_loss": 0.0129, + "llm_learning_rate": 5.717405992797187e-07, + "loss": 0.7032, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 5.7174059927971865e-05, + "step": 7681 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1633, + "image_masks_loss": 0.0079, + "llm_learning_rate": 5.712126816126004e-07, + "loss": 0.6633, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.7121268161260034e-05, + "step": 7682 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1633, + "image_masks_loss": 0.0074, + "llm_learning_rate": 5.706849763478381e-07, + "loss": 0.7282, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 5.706849763478381e-05, + "step": 7683 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1632, + "image_masks_loss": 0.0071, + "llm_learning_rate": 5.701574835435434e-07, + "loss": 0.7733, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.701574835435433e-05, + "step": 7684 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1632, + "image_masks_loss": 0.005, + "llm_learning_rate": 5.696302032578052e-07, + "loss": 0.7009, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.696302032578052e-05, + "step": 7685 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1632, + "image_masks_loss": 0.0087, + "llm_learning_rate": 5.691031355486887e-07, + "loss": 0.6225, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 5.6910313554868866e-05, + "step": 7686 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1631, + "image_masks_loss": 0.006, + "llm_learning_rate": 5.685762804742356e-07, + "loss": 0.7355, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 5.685762804742356e-05, + "step": 7687 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1631, + "image_masks_loss": 0.0045, + "llm_learning_rate": 5.680496380924644e-07, + "loss": 0.7929, + "output_text_mask_loss": 0.0553, + "predictor_learning_rate": 5.6804963809246436e-05, + "step": 7688 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.163, + "image_masks_loss": 0.0122, + "llm_learning_rate": 5.675232084613705e-07, + "loss": 0.7953, + "output_text_mask_loss": 0.0085, + "predictor_learning_rate": 5.675232084613705e-05, + "step": 7689 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.163, + "image_masks_loss": 0.0154, + "llm_learning_rate": 5.669969916389257e-07, + "loss": 0.7183, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 5.669969916389256e-05, + "step": 7690 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.163, + "image_masks_loss": 0.0125, + "llm_learning_rate": 5.664709876830773e-07, + "loss": 0.8218, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.664709876830773e-05, + "step": 7691 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1629, + "image_masks_loss": 0.0107, + "llm_learning_rate": 5.65945196651751e-07, + "loss": 0.6869, + "output_text_mask_loss": 0.0412, + "predictor_learning_rate": 5.6594519665175096e-05, + "step": 7692 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1629, + "image_masks_loss": 0.0174, + "llm_learning_rate": 5.654196186028474e-07, + "loss": 0.6507, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 5.654196186028473e-05, + "step": 7693 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1628, + "image_masks_loss": 0.0114, + "llm_learning_rate": 5.648942535942445e-07, + "loss": 0.7238, + "output_text_mask_loss": 0.0587, + "predictor_learning_rate": 5.648942535942445e-05, + "step": 7694 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1628, + "image_masks_loss": 0.0174, + "llm_learning_rate": 5.643691016837966e-07, + "loss": 0.7075, + "output_text_mask_loss": 0.0538, + "predictor_learning_rate": 5.643691016837965e-05, + "step": 7695 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1628, + "image_masks_loss": 0.0041, + "llm_learning_rate": 5.638441629293348e-07, + "loss": 0.7447, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 5.6384416292933476e-05, + "step": 7696 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1627, + "image_masks_loss": 0.0086, + "llm_learning_rate": 5.633194373886663e-07, + "loss": 0.7532, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.633194373886663e-05, + "step": 7697 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1627, + "image_masks_loss": 0.0184, + "llm_learning_rate": 5.627949251195753e-07, + "loss": 0.714, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 5.627949251195752e-05, + "step": 7698 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1627, + "image_masks_loss": 0.0067, + "llm_learning_rate": 5.622706261798217e-07, + "loss": 0.7366, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 5.622706261798216e-05, + "step": 7699 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1626, + "image_masks_loss": 0.0078, + "llm_learning_rate": 5.617465406271425e-07, + "loss": 0.7806, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 5.617465406271424e-05, + "step": 7700 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1626, + "image_masks_loss": 0.0087, + "llm_learning_rate": 5.61222668519252e-07, + "loss": 0.7174, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 5.61222668519252e-05, + "step": 7701 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1625, + "image_masks_loss": 0.008, + "llm_learning_rate": 5.606990099138401e-07, + "loss": 0.7315, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 5.606990099138401e-05, + "step": 7702 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1625, + "image_masks_loss": 0.0083, + "llm_learning_rate": 5.601755648685725e-07, + "loss": 0.8205, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 5.601755648685725e-05, + "step": 7703 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1625, + "image_masks_loss": 0.0135, + "llm_learning_rate": 5.596523334410927e-07, + "loss": 0.7535, + "output_text_mask_loss": 0.0227, + "predictor_learning_rate": 5.596523334410927e-05, + "step": 7704 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1624, + "image_masks_loss": 0.0054, + "llm_learning_rate": 5.591293156890198e-07, + "loss": 0.6465, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.5912931568901984e-05, + "step": 7705 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1624, + "image_masks_loss": 0.009, + "llm_learning_rate": 5.586065116699499e-07, + "loss": 0.6916, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.586065116699498e-05, + "step": 7706 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1623, + "image_masks_loss": 0.0171, + "llm_learning_rate": 5.580839214414543e-07, + "loss": 0.6612, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 5.580839214414543e-05, + "step": 7707 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1623, + "image_masks_loss": 0.0068, + "llm_learning_rate": 5.575615450610836e-07, + "loss": 0.6819, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 5.575615450610835e-05, + "step": 7708 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1623, + "image_masks_loss": 0.009, + "llm_learning_rate": 5.570393825863621e-07, + "loss": 0.709, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 5.57039382586362e-05, + "step": 7709 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1622, + "image_masks_loss": 0.0059, + "llm_learning_rate": 5.565174340747914e-07, + "loss": 0.72, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 5.565174340747914e-05, + "step": 7710 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1622, + "image_masks_loss": 0.0052, + "llm_learning_rate": 5.5599569958385e-07, + "loss": 0.7291, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 5.5599569958385e-05, + "step": 7711 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1622, + "image_masks_loss": 0.0174, + "llm_learning_rate": 5.554741791709922e-07, + "loss": 0.692, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.554741791709922e-05, + "step": 7712 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1621, + "image_masks_loss": 0.0026, + "llm_learning_rate": 5.54952872893649e-07, + "loss": 0.7361, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.54952872893649e-05, + "step": 7713 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1621, + "image_masks_loss": 0.0032, + "llm_learning_rate": 5.544317808092281e-07, + "loss": 0.6632, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 5.544317808092281e-05, + "step": 7714 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.162, + "image_masks_loss": 0.0017, + "llm_learning_rate": 5.539109029751131e-07, + "loss": 0.7516, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 5.53910902975113e-05, + "step": 7715 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.162, + "image_masks_loss": 0.0114, + "llm_learning_rate": 5.533902394486637e-07, + "loss": 0.7182, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 5.533902394486637e-05, + "step": 7716 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.162, + "image_masks_loss": 0.0075, + "llm_learning_rate": 5.528697902872174e-07, + "loss": 0.6993, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.528697902872173e-05, + "step": 7717 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1619, + "image_masks_loss": 0.0237, + "llm_learning_rate": 5.523495555480868e-07, + "loss": 0.7764, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 5.5234955554808673e-05, + "step": 7718 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1619, + "image_masks_loss": 0.003, + "llm_learning_rate": 5.518295352885614e-07, + "loss": 0.7847, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.5182953528856135e-05, + "step": 7719 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1619, + "image_masks_loss": 0.0052, + "llm_learning_rate": 5.513097295659062e-07, + "loss": 0.7136, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 5.5130972956590616e-05, + "step": 7720 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1618, + "image_masks_loss": 0.0109, + "llm_learning_rate": 5.507901384373648e-07, + "loss": 0.7476, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 5.507901384373648e-05, + "step": 7721 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1618, + "image_masks_loss": 0.005, + "llm_learning_rate": 5.502707619601552e-07, + "loss": 0.7743, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 5.502707619601552e-05, + "step": 7722 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1617, + "image_masks_loss": 0.007, + "llm_learning_rate": 5.497516001914721e-07, + "loss": 0.7256, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 5.4975160019147206e-05, + "step": 7723 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1617, + "image_masks_loss": 0.0094, + "llm_learning_rate": 5.492326531884867e-07, + "loss": 0.8054, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 5.492326531884867e-05, + "step": 7724 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1617, + "image_masks_loss": 0.0102, + "llm_learning_rate": 5.487139210083467e-07, + "loss": 0.7883, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 5.487139210083467e-05, + "step": 7725 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1616, + "image_masks_loss": 0.003, + "llm_learning_rate": 5.48195403708176e-07, + "loss": 0.7242, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 5.48195403708176e-05, + "step": 7726 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1616, + "image_masks_loss": 0.0066, + "llm_learning_rate": 5.476771013450744e-07, + "loss": 0.7955, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 5.476771013450743e-05, + "step": 7727 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1615, + "image_masks_loss": 0.0224, + "llm_learning_rate": 5.471590139761196e-07, + "loss": 0.7952, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 5.4715901397611955e-05, + "step": 7728 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1615, + "image_masks_loss": 0.0154, + "llm_learning_rate": 5.466411416583639e-07, + "loss": 0.7156, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.466411416583639e-05, + "step": 7729 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1615, + "image_masks_loss": 0.0128, + "llm_learning_rate": 5.461234844488366e-07, + "loss": 0.7891, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.461234844488366e-05, + "step": 7730 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1614, + "image_masks_loss": 0.0166, + "llm_learning_rate": 5.456060424045437e-07, + "loss": 0.718, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.456060424045436e-05, + "step": 7731 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1614, + "image_masks_loss": 0.0169, + "llm_learning_rate": 5.450888155824663e-07, + "loss": 0.7374, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 5.450888155824663e-05, + "step": 7732 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1614, + "image_masks_loss": 0.0204, + "llm_learning_rate": 5.445718040395628e-07, + "loss": 0.8058, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 5.445718040395628e-05, + "step": 7733 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1613, + "image_masks_loss": 0.0221, + "llm_learning_rate": 5.44055007832768e-07, + "loss": 0.752, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 5.4405500783276804e-05, + "step": 7734 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1613, + "image_masks_loss": 0.0233, + "llm_learning_rate": 5.435384270189922e-07, + "loss": 0.7236, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.435384270189922e-05, + "step": 7735 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1612, + "image_masks_loss": 0.0072, + "llm_learning_rate": 5.430220616551229e-07, + "loss": 0.6566, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.430220616551229e-05, + "step": 7736 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1612, + "image_masks_loss": 0.0194, + "llm_learning_rate": 5.425059117980234e-07, + "loss": 0.679, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.425059117980233e-05, + "step": 7737 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1612, + "image_masks_loss": 0.0162, + "llm_learning_rate": 5.419899775045325e-07, + "loss": 0.7681, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 5.4198997750453254e-05, + "step": 7738 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1611, + "image_masks_loss": 0.0135, + "llm_learning_rate": 5.414742588314661e-07, + "loss": 0.702, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.4147425883146615e-05, + "step": 7739 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1611, + "image_masks_loss": 0.0265, + "llm_learning_rate": 5.409587558356175e-07, + "loss": 0.7104, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 5.4095875583561746e-05, + "step": 7740 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1611, + "image_masks_loss": 0.0094, + "llm_learning_rate": 5.404434685737545e-07, + "loss": 0.7487, + "output_text_mask_loss": 0.0296, + "predictor_learning_rate": 5.404434685737544e-05, + "step": 7741 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.161, + "image_masks_loss": 0.0135, + "llm_learning_rate": 5.399283971026212e-07, + "loss": 0.7379, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.399283971026211e-05, + "step": 7742 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.161, + "image_masks_loss": 0.0158, + "llm_learning_rate": 5.394135414789387e-07, + "loss": 0.7151, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 5.394135414789386e-05, + "step": 7743 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1609, + "image_masks_loss": 0.0082, + "llm_learning_rate": 5.38898901759404e-07, + "loss": 0.7485, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 5.38898901759404e-05, + "step": 7744 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1609, + "image_masks_loss": 0.0125, + "llm_learning_rate": 5.383844780006906e-07, + "loss": 0.7072, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 5.3838447800069054e-05, + "step": 7745 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1609, + "image_masks_loss": 0.0034, + "llm_learning_rate": 5.37870270259447e-07, + "loss": 0.6923, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.3787027025944706e-05, + "step": 7746 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1608, + "image_masks_loss": 0.0044, + "llm_learning_rate": 5.373562785923006e-07, + "loss": 0.7247, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.373562785923006e-05, + "step": 7747 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1608, + "image_masks_loss": 0.013, + "llm_learning_rate": 5.368425030558522e-07, + "loss": 0.6317, + "output_text_mask_loss": 0.0296, + "predictor_learning_rate": 5.3684250305585215e-05, + "step": 7748 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1607, + "image_masks_loss": 0.0125, + "llm_learning_rate": 5.363289437066801e-07, + "loss": 0.6962, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.363289437066801e-05, + "step": 7749 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1607, + "image_masks_loss": 0.0188, + "llm_learning_rate": 5.358156006013396e-07, + "loss": 0.7715, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.358156006013395e-05, + "step": 7750 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1607, + "image_masks_loss": 0.0261, + "llm_learning_rate": 5.353024737963592e-07, + "loss": 0.6942, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 5.353024737963591e-05, + "step": 7751 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1606, + "image_masks_loss": 0.038, + "llm_learning_rate": 5.347895633482462e-07, + "loss": 0.6947, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.347895633482461e-05, + "step": 7752 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1606, + "image_masks_loss": 0.0234, + "llm_learning_rate": 5.342768693134842e-07, + "loss": 0.749, + "output_text_mask_loss": 0.0622, + "predictor_learning_rate": 5.342768693134842e-05, + "step": 7753 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1606, + "image_masks_loss": 0.0092, + "llm_learning_rate": 5.337643917485319e-07, + "loss": 0.7347, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 5.337643917485319e-05, + "step": 7754 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1605, + "image_masks_loss": 0.0114, + "llm_learning_rate": 5.332521307098246e-07, + "loss": 0.7749, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.3325213070982457e-05, + "step": 7755 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1605, + "image_masks_loss": 0.0075, + "llm_learning_rate": 5.327400862537732e-07, + "loss": 0.7917, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 5.327400862537732e-05, + "step": 7756 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1604, + "image_masks_loss": 0.0228, + "llm_learning_rate": 5.322282584367655e-07, + "loss": 0.6131, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.3222825843676545e-05, + "step": 7757 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1604, + "image_masks_loss": 0.0088, + "llm_learning_rate": 5.317166473151649e-07, + "loss": 0.734, + "output_text_mask_loss": 0.0591, + "predictor_learning_rate": 5.31716647315165e-05, + "step": 7758 + }, + { + "epoch": 0.79, + "gumbel_tau": 0.1604, + "image_masks_loss": 0.0198, + "llm_learning_rate": 5.312052529453107e-07, + "loss": 0.8049, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.312052529453107e-05, + "step": 7759 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1603, + "image_masks_loss": 0.0307, + "llm_learning_rate": 5.306940753835202e-07, + "loss": 0.7066, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 5.306940753835201e-05, + "step": 7760 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1603, + "image_masks_loss": 0.0154, + "llm_learning_rate": 5.301831146860842e-07, + "loss": 0.6992, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 5.301831146860842e-05, + "step": 7761 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1603, + "image_masks_loss": 0.0237, + "llm_learning_rate": 5.296723709092716e-07, + "loss": 0.7293, + "output_text_mask_loss": 0.1144, + "predictor_learning_rate": 5.2967237090927155e-05, + "step": 7762 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1602, + "image_masks_loss": 0.0077, + "llm_learning_rate": 5.29161844109326e-07, + "loss": 0.7718, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.29161844109326e-05, + "step": 7763 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1602, + "image_masks_loss": 0.0196, + "llm_learning_rate": 5.286515343424681e-07, + "loss": 0.7303, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.2865153434246804e-05, + "step": 7764 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1601, + "image_masks_loss": 0.0138, + "llm_learning_rate": 5.281414416648941e-07, + "loss": 0.8089, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 5.2814144166489415e-05, + "step": 7765 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1601, + "image_masks_loss": 0.0169, + "llm_learning_rate": 5.276315661327763e-07, + "loss": 0.7234, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 5.276315661327763e-05, + "step": 7766 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1601, + "image_masks_loss": 0.0126, + "llm_learning_rate": 5.27121907802264e-07, + "loss": 0.679, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 5.271219078022641e-05, + "step": 7767 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.16, + "image_masks_loss": 0.0126, + "llm_learning_rate": 5.26612466729482e-07, + "loss": 0.7155, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 5.266124667294819e-05, + "step": 7768 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.16, + "image_masks_loss": 0.0081, + "llm_learning_rate": 5.261032429705312e-07, + "loss": 0.729, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 5.261032429705312e-05, + "step": 7769 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.16, + "image_masks_loss": 0.0108, + "llm_learning_rate": 5.255942365814873e-07, + "loss": 0.6899, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 5.255942365814873e-05, + "step": 7770 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1599, + "image_masks_loss": 0.0189, + "llm_learning_rate": 5.250854476184034e-07, + "loss": 0.7061, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 5.250854476184033e-05, + "step": 7771 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1599, + "image_masks_loss": 0.0212, + "llm_learning_rate": 5.245768761373093e-07, + "loss": 0.6159, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 5.245768761373093e-05, + "step": 7772 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1598, + "image_masks_loss": 0.008, + "llm_learning_rate": 5.240685221942099e-07, + "loss": 0.8171, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 5.240685221942099e-05, + "step": 7773 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1598, + "image_masks_loss": 0.003, + "llm_learning_rate": 5.235603858450861e-07, + "loss": 0.7207, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.2356038584508606e-05, + "step": 7774 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1598, + "image_masks_loss": 0.0102, + "llm_learning_rate": 5.230524671458947e-07, + "loss": 0.7455, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 5.2305246714589474e-05, + "step": 7775 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1597, + "image_masks_loss": 0.0033, + "llm_learning_rate": 5.225447661525693e-07, + "loss": 0.6537, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.225447661525692e-05, + "step": 7776 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1597, + "image_masks_loss": 0.0094, + "llm_learning_rate": 5.220372829210188e-07, + "loss": 0.7456, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.220372829210188e-05, + "step": 7777 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1597, + "image_masks_loss": 0.0175, + "llm_learning_rate": 5.215300175071278e-07, + "loss": 0.7646, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 5.2153001750712775e-05, + "step": 7778 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1596, + "image_masks_loss": 0.0057, + "llm_learning_rate": 5.210229699667588e-07, + "loss": 0.7515, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 5.210229699667587e-05, + "step": 7779 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1596, + "image_masks_loss": 0.0101, + "llm_learning_rate": 5.205161403557483e-07, + "loss": 0.7547, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 5.205161403557482e-05, + "step": 7780 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1595, + "image_masks_loss": 0.0036, + "llm_learning_rate": 5.200095287299095e-07, + "loss": 0.6763, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.200095287299095e-05, + "step": 7781 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1595, + "image_masks_loss": 0.0307, + "llm_learning_rate": 5.19503135145032e-07, + "loss": 0.7153, + "output_text_mask_loss": 0.0383, + "predictor_learning_rate": 5.1950313514503186e-05, + "step": 7782 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1595, + "image_masks_loss": 0.0123, + "llm_learning_rate": 5.189969596568803e-07, + "loss": 0.6826, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.1899695965688026e-05, + "step": 7783 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1594, + "image_masks_loss": 0.0114, + "llm_learning_rate": 5.184910023211959e-07, + "loss": 0.7145, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 5.1849100232119585e-05, + "step": 7784 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1594, + "image_masks_loss": 0.0083, + "llm_learning_rate": 5.179852631936954e-07, + "loss": 0.614, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.1798526319369544e-05, + "step": 7785 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1594, + "image_masks_loss": 0.0343, + "llm_learning_rate": 5.174797423300732e-07, + "loss": 0.7558, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 5.1747974233007314e-05, + "step": 7786 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1593, + "image_masks_loss": 0.0161, + "llm_learning_rate": 5.169744397859983e-07, + "loss": 0.7437, + "output_text_mask_loss": 0.028, + "predictor_learning_rate": 5.1697443978599826e-05, + "step": 7787 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1593, + "image_masks_loss": 0.0123, + "llm_learning_rate": 5.164693556171144e-07, + "loss": 0.654, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 5.1646935561711435e-05, + "step": 7788 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1592, + "image_masks_loss": 0.0317, + "llm_learning_rate": 5.159644898790433e-07, + "loss": 0.8172, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 5.159644898790433e-05, + "step": 7789 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1592, + "image_masks_loss": 0.0233, + "llm_learning_rate": 5.154598426273816e-07, + "loss": 0.7553, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 5.1545984262738155e-05, + "step": 7790 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1592, + "image_masks_loss": 0.0151, + "llm_learning_rate": 5.149554139177021e-07, + "loss": 0.785, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 5.149554139177021e-05, + "step": 7791 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1591, + "image_masks_loss": 0.0078, + "llm_learning_rate": 5.144512038055544e-07, + "loss": 0.7158, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 5.144512038055543e-05, + "step": 7792 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1591, + "image_masks_loss": 0.0109, + "llm_learning_rate": 5.139472123464629e-07, + "loss": 0.6907, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 5.139472123464628e-05, + "step": 7793 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1591, + "image_masks_loss": 0.0121, + "llm_learning_rate": 5.134434395959279e-07, + "loss": 0.7514, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 5.1344343959592784e-05, + "step": 7794 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.159, + "image_masks_loss": 0.0212, + "llm_learning_rate": 5.129398856094262e-07, + "loss": 0.6701, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 5.129398856094261e-05, + "step": 7795 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.159, + "image_masks_loss": 0.0172, + "llm_learning_rate": 5.124365504424106e-07, + "loss": 0.7653, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 5.1243655044241055e-05, + "step": 7796 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1589, + "image_masks_loss": 0.011, + "llm_learning_rate": 5.11933434150309e-07, + "loss": 0.7753, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 5.1193343415030905e-05, + "step": 7797 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1589, + "image_masks_loss": 0.0135, + "llm_learning_rate": 5.114305367885253e-07, + "loss": 0.6907, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 5.114305367885252e-05, + "step": 7798 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1589, + "image_masks_loss": 0.0098, + "llm_learning_rate": 5.109278584124408e-07, + "loss": 0.6908, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.109278584124408e-05, + "step": 7799 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1588, + "image_masks_loss": 0.0311, + "llm_learning_rate": 5.10425399077411e-07, + "loss": 0.6997, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 5.1042539907741104e-05, + "step": 7800 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1588, + "image_masks_loss": 0.0154, + "llm_learning_rate": 5.099231588387679e-07, + "loss": 0.7244, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.099231588387679e-05, + "step": 7801 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1588, + "image_masks_loss": 0.0092, + "llm_learning_rate": 5.094211377518193e-07, + "loss": 0.6668, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.094211377518193e-05, + "step": 7802 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1587, + "image_masks_loss": 0.0113, + "llm_learning_rate": 5.089193358718488e-07, + "loss": 0.7514, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 5.0891933587184886e-05, + "step": 7803 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1587, + "image_masks_loss": 0.0161, + "llm_learning_rate": 5.084177532541162e-07, + "loss": 0.5697, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 5.0841775325411623e-05, + "step": 7804 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1586, + "image_masks_loss": 0.0032, + "llm_learning_rate": 5.079163899538559e-07, + "loss": 0.6326, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 5.079163899538558e-05, + "step": 7805 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1586, + "image_masks_loss": 0.018, + "llm_learning_rate": 5.07415246026281e-07, + "loss": 0.7326, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.0741524602628096e-05, + "step": 7806 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1586, + "image_masks_loss": 0.0087, + "llm_learning_rate": 5.06914321526577e-07, + "loss": 0.783, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 5.06914321526577e-05, + "step": 7807 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1585, + "image_masks_loss": 0.0128, + "llm_learning_rate": 5.064136165099076e-07, + "loss": 0.7539, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 5.0641361650990756e-05, + "step": 7808 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1585, + "image_masks_loss": 0.0202, + "llm_learning_rate": 5.059131310314111e-07, + "loss": 0.7151, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 5.059131310314111e-05, + "step": 7809 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1585, + "image_masks_loss": 0.013, + "llm_learning_rate": 5.054128651462017e-07, + "loss": 0.7711, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 5.054128651462017e-05, + "step": 7810 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1584, + "image_masks_loss": 0.0324, + "llm_learning_rate": 5.049128189093708e-07, + "loss": 0.6563, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.0491281890937074e-05, + "step": 7811 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1584, + "image_masks_loss": 0.0135, + "llm_learning_rate": 5.044129923759844e-07, + "loss": 0.575, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.044129923759844e-05, + "step": 7812 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1583, + "image_masks_loss": 0.0092, + "llm_learning_rate": 5.039133856010841e-07, + "loss": 0.7626, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 5.0391338560108414e-05, + "step": 7813 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1583, + "image_masks_loss": 0.0216, + "llm_learning_rate": 5.034139986396879e-07, + "loss": 0.7653, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 5.03413998639688e-05, + "step": 7814 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1583, + "image_masks_loss": 0.0093, + "llm_learning_rate": 5.029148315467894e-07, + "loss": 0.683, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.0291483154678934e-05, + "step": 7815 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1582, + "image_masks_loss": 0.0157, + "llm_learning_rate": 5.024158843773582e-07, + "loss": 0.7804, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 5.024158843773582e-05, + "step": 7816 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1582, + "image_masks_loss": 0.0112, + "llm_learning_rate": 5.019171571863385e-07, + "loss": 0.767, + "output_text_mask_loss": 0.0668, + "predictor_learning_rate": 5.019171571863385e-05, + "step": 7817 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1582, + "image_masks_loss": 0.0366, + "llm_learning_rate": 5.014186500286525e-07, + "loss": 0.8307, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.014186500286525e-05, + "step": 7818 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1581, + "image_masks_loss": 0.0056, + "llm_learning_rate": 5.009203629591966e-07, + "loss": 0.7465, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 5.009203629591966e-05, + "step": 7819 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1581, + "image_masks_loss": 0.0081, + "llm_learning_rate": 5.004222960328428e-07, + "loss": 0.7075, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 5.004222960328428e-05, + "step": 7820 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.158, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.9992444930444e-07, + "loss": 0.8363, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 4.9992444930443995e-05, + "step": 7821 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.158, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.994268228288115e-07, + "loss": 0.8092, + "output_text_mask_loss": 0.0026, + "predictor_learning_rate": 4.994268228288115e-05, + "step": 7822 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.158, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.989294166607578e-07, + "loss": 0.702, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 4.989294166607577e-05, + "step": 7823 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1579, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.984322308550538e-07, + "loss": 0.6187, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.9843223085505374e-05, + "step": 7824 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1579, + "image_masks_loss": 0.0055, + "llm_learning_rate": 4.979352654664507e-07, + "loss": 0.788, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 4.979352654664507e-05, + "step": 7825 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1579, + "image_masks_loss": 0.0024, + "llm_learning_rate": 4.974385205496756e-07, + "loss": 0.7505, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.9743852054967565e-05, + "step": 7826 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1578, + "image_masks_loss": 0.0048, + "llm_learning_rate": 4.969419961594313e-07, + "loss": 0.7453, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 4.969419961594313e-05, + "step": 7827 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1578, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.96445692350396e-07, + "loss": 0.7327, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.96445692350396e-05, + "step": 7828 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1577, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.959496091772237e-07, + "loss": 0.7294, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.9594960917722366e-05, + "step": 7829 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1577, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.954537466945436e-07, + "loss": 0.7664, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 4.954537466945436e-05, + "step": 7830 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1577, + "image_masks_loss": 0.0166, + "llm_learning_rate": 4.949581049569627e-07, + "loss": 0.7315, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 4.9495810495696266e-05, + "step": 7831 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1576, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.944626840190611e-07, + "loss": 0.6851, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.944626840190611e-05, + "step": 7832 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1576, + "image_masks_loss": 0.0269, + "llm_learning_rate": 4.939674839353961e-07, + "loss": 0.7219, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.939674839353961e-05, + "step": 7833 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1576, + "image_masks_loss": 0.0056, + "llm_learning_rate": 4.934725047604999e-07, + "loss": 0.7482, + "output_text_mask_loss": 0.0265, + "predictor_learning_rate": 4.934725047604999e-05, + "step": 7834 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1575, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.929777465488808e-07, + "loss": 0.6175, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 4.929777465488808e-05, + "step": 7835 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1575, + "image_masks_loss": 0.0059, + "llm_learning_rate": 4.924832093550231e-07, + "loss": 0.7016, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.92483209355023e-05, + "step": 7836 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1574, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.91988893233385e-07, + "loss": 0.801, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 4.9198889323338494e-05, + "step": 7837 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1574, + "image_masks_loss": 0.0029, + "llm_learning_rate": 4.91494798238403e-07, + "loss": 0.7307, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 4.91494798238403e-05, + "step": 7838 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1574, + "image_masks_loss": 0.0045, + "llm_learning_rate": 4.910009244244881e-07, + "loss": 0.8029, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 4.910009244244881e-05, + "step": 7839 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1573, + "image_masks_loss": 0.0086, + "llm_learning_rate": 4.905072718460261e-07, + "loss": 0.7149, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 4.90507271846026e-05, + "step": 7840 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1573, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.900138405573793e-07, + "loss": 0.7413, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 4.900138405573792e-05, + "step": 7841 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1573, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.895206306128852e-07, + "loss": 0.8022, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 4.895206306128852e-05, + "step": 7842 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1572, + "image_masks_loss": 0.0194, + "llm_learning_rate": 4.890276420668577e-07, + "loss": 0.6776, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 4.890276420668577e-05, + "step": 7843 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1572, + "image_masks_loss": 0.0248, + "llm_learning_rate": 4.885348749735854e-07, + "loss": 0.6789, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 4.885348749735854e-05, + "step": 7844 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1571, + "image_masks_loss": 0.0186, + "llm_learning_rate": 4.880423293873332e-07, + "loss": 0.701, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 4.880423293873332e-05, + "step": 7845 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1571, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.875500053623411e-07, + "loss": 0.8105, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.8755000536234106e-05, + "step": 7846 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1571, + "image_masks_loss": 0.0126, + "llm_learning_rate": 4.87057902952825e-07, + "loss": 0.6128, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.870579029528249e-05, + "step": 7847 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.157, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.865660222129759e-07, + "loss": 0.7411, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.8656602221297586e-05, + "step": 7848 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.157, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.860743631969608e-07, + "loss": 0.7568, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.860743631969608e-05, + "step": 7849 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.157, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.855829259589234e-07, + "loss": 0.7483, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 4.855829259589234e-05, + "step": 7850 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1569, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.850917105529812e-07, + "loss": 0.6408, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.850917105529812e-05, + "step": 7851 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1569, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.846007170332276e-07, + "loss": 0.72, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 4.846007170332276e-05, + "step": 7852 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1569, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.841099454537324e-07, + "loss": 0.5783, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.8410994545373234e-05, + "step": 7853 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1568, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.8361939586854e-07, + "loss": 0.7295, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.8361939586854006e-05, + "step": 7854 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1568, + "image_masks_loss": 0.007, + "llm_learning_rate": 4.831290683316714e-07, + "loss": 0.6194, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 4.831290683316714e-05, + "step": 7855 + }, + { + "epoch": 0.8, + "gumbel_tau": 0.1567, + "image_masks_loss": 0.0252, + "llm_learning_rate": 4.826389628971212e-07, + "loss": 0.6751, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.8263896289712115e-05, + "step": 7856 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1567, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.821490796188627e-07, + "loss": 0.7022, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.821490796188627e-05, + "step": 7857 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1567, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.816594185508419e-07, + "loss": 0.716, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 4.816594185508419e-05, + "step": 7858 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1566, + "image_masks_loss": 0.0056, + "llm_learning_rate": 4.811699797469818e-07, + "loss": 0.7535, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.811699797469818e-05, + "step": 7859 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1566, + "image_masks_loss": 0.0282, + "llm_learning_rate": 4.806807632611804e-07, + "loss": 0.7595, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 4.8068076326118035e-05, + "step": 7860 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1566, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.801917691473113e-07, + "loss": 0.6758, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 4.8019176914731124e-05, + "step": 7861 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1565, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.797029974592233e-07, + "loss": 0.6438, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 4.797029974592232e-05, + "step": 7862 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1565, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.792144482507413e-07, + "loss": 0.7671, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 4.792144482507413e-05, + "step": 7863 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1564, + "image_masks_loss": 0.0063, + "llm_learning_rate": 4.787261215756653e-07, + "loss": 0.7993, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 4.7872612157566525e-05, + "step": 7864 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1564, + "image_masks_loss": 0.0059, + "llm_learning_rate": 4.78238017487771e-07, + "loss": 0.7098, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 4.7823801748777095e-05, + "step": 7865 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1564, + "image_masks_loss": 0.0071, + "llm_learning_rate": 4.777501360408096e-07, + "loss": 0.7694, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 4.777501360408096e-05, + "step": 7866 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1563, + "image_masks_loss": 0.0139, + "llm_learning_rate": 4.772624772885076e-07, + "loss": 0.6399, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.7726247728850756e-05, + "step": 7867 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1563, + "image_masks_loss": 0.0771, + "llm_learning_rate": 4.7677504128456704e-07, + "loss": 0.7855, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 4.7677504128456703e-05, + "step": 7868 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1563, + "image_masks_loss": 0.0022, + "llm_learning_rate": 4.7628782808266496e-07, + "loss": 0.6545, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.762878280826649e-05, + "step": 7869 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1562, + "image_masks_loss": 0.0128, + "llm_learning_rate": 4.7580083773645563e-07, + "loss": 0.6898, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 4.758008377364556e-05, + "step": 7870 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1562, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.753140702995665e-07, + "loss": 0.7307, + "output_text_mask_loss": 0.0288, + "predictor_learning_rate": 4.7531407029956644e-05, + "step": 7871 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1562, + "image_masks_loss": 0.0098, + "llm_learning_rate": 4.748275258256019e-07, + "loss": 0.7445, + "output_text_mask_loss": 0.0226, + "predictor_learning_rate": 4.7482752582560185e-05, + "step": 7872 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1561, + "image_masks_loss": 0.0149, + "llm_learning_rate": 4.7434120436814087e-07, + "loss": 0.6704, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 4.743412043681408e-05, + "step": 7873 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1561, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.7385510598073865e-07, + "loss": 0.7405, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 4.738551059807386e-05, + "step": 7874 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.156, + "image_masks_loss": 0.0043, + "llm_learning_rate": 4.733692307169252e-07, + "loss": 0.6437, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.7336923071692517e-05, + "step": 7875 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.156, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.728835786302052e-07, + "loss": 0.7363, + "output_text_mask_loss": 0.0057, + "predictor_learning_rate": 4.728835786302052e-05, + "step": 7876 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.156, + "image_masks_loss": 0.0117, + "llm_learning_rate": 4.723981497740618e-07, + "loss": 0.7622, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 4.723981497740618e-05, + "step": 7877 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1559, + "image_masks_loss": 0.0034, + "llm_learning_rate": 4.719129442019504e-07, + "loss": 0.8053, + "output_text_mask_loss": 0.0225, + "predictor_learning_rate": 4.719129442019504e-05, + "step": 7878 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1559, + "image_masks_loss": 0.023, + "llm_learning_rate": 4.714279619673029e-07, + "loss": 0.8005, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.714279619673029e-05, + "step": 7879 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1559, + "image_masks_loss": 0.0035, + "llm_learning_rate": 4.7094320312352724e-07, + "loss": 0.6947, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 4.7094320312352726e-05, + "step": 7880 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1558, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.7045866772400446e-07, + "loss": 0.868, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 4.7045866772400444e-05, + "step": 7881 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1558, + "image_masks_loss": 0.018, + "llm_learning_rate": 4.6997435582209427e-07, + "loss": 0.766, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 4.6997435582209425e-05, + "step": 7882 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1557, + "image_masks_loss": 0.0071, + "llm_learning_rate": 4.694902674711299e-07, + "loss": 0.7399, + "output_text_mask_loss": 0.0553, + "predictor_learning_rate": 4.694902674711299e-05, + "step": 7883 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1557, + "image_masks_loss": 0.0074, + "llm_learning_rate": 4.690064027244201e-07, + "loss": 0.7402, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 4.690064027244201e-05, + "step": 7884 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1557, + "image_masks_loss": 0.0138, + "llm_learning_rate": 4.685227616352489e-07, + "loss": 0.7733, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 4.685227616352489e-05, + "step": 7885 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1556, + "image_masks_loss": 0.0175, + "llm_learning_rate": 4.680393442568762e-07, + "loss": 0.6583, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.6803934425687616e-05, + "step": 7886 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1556, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.675561506425369e-07, + "loss": 0.6919, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 4.675561506425369e-05, + "step": 7887 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1556, + "image_masks_loss": 0.0184, + "llm_learning_rate": 4.6707318084544096e-07, + "loss": 0.7324, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.6707318084544095e-05, + "step": 7888 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1555, + "image_masks_loss": 0.0288, + "llm_learning_rate": 4.66590434918775e-07, + "loss": 0.7099, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 4.6659043491877493e-05, + "step": 7889 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1555, + "image_masks_loss": 0.0091, + "llm_learning_rate": 4.661079129156995e-07, + "loss": 0.6975, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.661079129156995e-05, + "step": 7890 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1555, + "image_masks_loss": 0.0155, + "llm_learning_rate": 4.656256148893509e-07, + "loss": 0.7054, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 4.656256148893509e-05, + "step": 7891 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1554, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.651435408928412e-07, + "loss": 0.7278, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 4.6514354089284114e-05, + "step": 7892 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1554, + "image_masks_loss": 0.0237, + "llm_learning_rate": 4.646616909792573e-07, + "loss": 0.7665, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.6466169097925724e-05, + "step": 7893 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1553, + "image_masks_loss": 0.0084, + "llm_learning_rate": 4.641800652016612e-07, + "loss": 0.7122, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 4.6418006520166116e-05, + "step": 7894 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1553, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.636986636130905e-07, + "loss": 0.6797, + "output_text_mask_loss": 0.0037, + "predictor_learning_rate": 4.636986636130905e-05, + "step": 7895 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1553, + "image_masks_loss": 0.0037, + "llm_learning_rate": 4.6321748626655923e-07, + "loss": 0.7271, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.632174862665592e-05, + "step": 7896 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1552, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.627365332150552e-07, + "loss": 0.7146, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 4.627365332150551e-05, + "step": 7897 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1552, + "image_masks_loss": 0.0177, + "llm_learning_rate": 4.6225580451154187e-07, + "loss": 0.7041, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.622558045115419e-05, + "step": 7898 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1552, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.6177530020895895e-07, + "loss": 0.7148, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.6177530020895895e-05, + "step": 7899 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1551, + "image_masks_loss": 0.0041, + "llm_learning_rate": 4.6129502036021926e-07, + "loss": 0.7443, + "output_text_mask_loss": 0.008, + "predictor_learning_rate": 4.6129502036021924e-05, + "step": 7900 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1551, + "image_masks_loss": 0.0067, + "llm_learning_rate": 4.608149650182123e-07, + "loss": 0.7735, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.6081496501821225e-05, + "step": 7901 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.155, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.603351342358042e-07, + "loss": 0.6544, + "output_text_mask_loss": 0.002, + "predictor_learning_rate": 4.603351342358042e-05, + "step": 7902 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.155, + "image_masks_loss": 0.008, + "llm_learning_rate": 4.598555280658343e-07, + "loss": 0.6723, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 4.598555280658342e-05, + "step": 7903 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.155, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.5937614656111795e-07, + "loss": 0.7437, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 4.593761465611179e-05, + "step": 7904 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1549, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.588969897744455e-07, + "loss": 0.7049, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 4.588969897744455e-05, + "step": 7905 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1549, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.5841805775858285e-07, + "loss": 0.7276, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 4.584180577585828e-05, + "step": 7906 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1549, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.5793935056627134e-07, + "loss": 0.8029, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 4.5793935056627125e-05, + "step": 7907 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1548, + "image_masks_loss": 0.0075, + "llm_learning_rate": 4.574608682502263e-07, + "loss": 0.6611, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 4.574608682502263e-05, + "step": 7908 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1548, + "image_masks_loss": 0.0053, + "llm_learning_rate": 4.569826108631406e-07, + "loss": 0.7809, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 4.5698261086314056e-05, + "step": 7909 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1548, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.565045784576808e-07, + "loss": 0.747, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 4.565045784576807e-05, + "step": 7910 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1547, + "image_masks_loss": 0.0052, + "llm_learning_rate": 4.5602677108648823e-07, + "loss": 0.7016, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 4.560267710864882e-05, + "step": 7911 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1547, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.5554918880218043e-07, + "loss": 0.7113, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.555491888021804e-05, + "step": 7912 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1546, + "image_masks_loss": 0.0135, + "llm_learning_rate": 4.5507183165734964e-07, + "loss": 0.8133, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 4.550718316573496e-05, + "step": 7913 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1546, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.5459469970456385e-07, + "loss": 0.7757, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.5459469970456384e-05, + "step": 7914 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1546, + "image_masks_loss": 0.0058, + "llm_learning_rate": 4.54117792996365e-07, + "loss": 0.7518, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.5411779299636496e-05, + "step": 7915 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1545, + "image_masks_loss": 0.0106, + "llm_learning_rate": 4.5364111158527233e-07, + "loss": 0.7247, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.5364111158527236e-05, + "step": 7916 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1545, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.531646555237795e-07, + "loss": 0.7513, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 4.531646555237795e-05, + "step": 7917 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1545, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.526884248643529e-07, + "loss": 0.6865, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 4.5268842486435285e-05, + "step": 7918 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1544, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.522124196594374e-07, + "loss": 0.6852, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 4.522124196594374e-05, + "step": 7919 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1544, + "image_masks_loss": 0.0085, + "llm_learning_rate": 4.5173663996145146e-07, + "loss": 0.7887, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 4.5173663996145145e-05, + "step": 7920 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1544, + "image_masks_loss": 0.0054, + "llm_learning_rate": 4.5126108582278864e-07, + "loss": 0.7758, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 4.5126108582278864e-05, + "step": 7921 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1543, + "image_masks_loss": 0.0051, + "llm_learning_rate": 4.507857572958188e-07, + "loss": 0.7735, + "output_text_mask_loss": 0.0018, + "predictor_learning_rate": 4.507857572958188e-05, + "step": 7922 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1543, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.503106544328858e-07, + "loss": 0.6748, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.503106544328858e-05, + "step": 7923 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1542, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.4983577728630936e-07, + "loss": 0.7319, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 4.498357772863093e-05, + "step": 7924 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1542, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.493611259083833e-07, + "loss": 0.7535, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.493611259083832e-05, + "step": 7925 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1542, + "image_masks_loss": 0.0145, + "llm_learning_rate": 4.488867003513778e-07, + "loss": 0.7007, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 4.488867003513777e-05, + "step": 7926 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1541, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.484125006675377e-07, + "loss": 0.6827, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 4.484125006675377e-05, + "step": 7927 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1541, + "image_masks_loss": 0.0024, + "llm_learning_rate": 4.479385269090819e-07, + "loss": 0.7318, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.479385269090819e-05, + "step": 7928 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1541, + "image_masks_loss": 0.0118, + "llm_learning_rate": 4.474647791282069e-07, + "loss": 0.7328, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 4.474647791282069e-05, + "step": 7929 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.154, + "image_masks_loss": 0.0093, + "llm_learning_rate": 4.4699125737708267e-07, + "loss": 0.7754, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 4.4699125737708264e-05, + "step": 7930 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.154, + "image_masks_loss": 0.0164, + "llm_learning_rate": 4.4651796170785354e-07, + "loss": 0.7109, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 4.465179617078535e-05, + "step": 7931 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.154, + "image_masks_loss": 0.0197, + "llm_learning_rate": 4.4604489217264067e-07, + "loss": 0.8083, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.4604489217264066e-05, + "step": 7932 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1539, + "image_masks_loss": 0.0045, + "llm_learning_rate": 4.4557204882353933e-07, + "loss": 0.7731, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 4.455720488235393e-05, + "step": 7933 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1539, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.4509943171262e-07, + "loss": 0.7184, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 4.4509943171262e-05, + "step": 7934 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1538, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.4462704089192785e-07, + "loss": 0.7779, + "output_text_mask_loss": 0.0097, + "predictor_learning_rate": 4.446270408919279e-05, + "step": 7935 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1538, + "image_masks_loss": 0.003, + "llm_learning_rate": 4.441548764134851e-07, + "loss": 0.8447, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 4.4415487641348506e-05, + "step": 7936 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1538, + "image_masks_loss": 0.0064, + "llm_learning_rate": 4.436829383292862e-07, + "loss": 0.6595, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 4.4368293832928616e-05, + "step": 7937 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1537, + "image_masks_loss": 0.0102, + "llm_learning_rate": 4.432112266913022e-07, + "loss": 0.6895, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.4321122669130216e-05, + "step": 7938 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1537, + "image_masks_loss": 0.0118, + "llm_learning_rate": 4.4273974155147945e-07, + "loss": 0.7469, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 4.427397415514795e-05, + "step": 7939 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1537, + "image_masks_loss": 0.0076, + "llm_learning_rate": 4.42268482961738e-07, + "loss": 0.7343, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 4.42268482961738e-05, + "step": 7940 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1536, + "image_masks_loss": 0.0072, + "llm_learning_rate": 4.4179745097397504e-07, + "loss": 0.6771, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.41797450973975e-05, + "step": 7941 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1536, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.4132664564006167e-07, + "loss": 0.7588, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.4132664564006165e-05, + "step": 7942 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1536, + "image_masks_loss": 0.0362, + "llm_learning_rate": 4.4085606701184334e-07, + "loss": 0.8492, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.4085606701184336e-05, + "step": 7943 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1535, + "image_masks_loss": 0.011, + "llm_learning_rate": 4.403857151411414e-07, + "loss": 0.6967, + "output_text_mask_loss": 0.0587, + "predictor_learning_rate": 4.403857151411414e-05, + "step": 7944 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1535, + "image_masks_loss": 0.0074, + "llm_learning_rate": 4.399155900797522e-07, + "loss": 0.6614, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.399155900797522e-05, + "step": 7945 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1534, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.394456918794468e-07, + "loss": 0.6491, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 4.394456918794468e-05, + "step": 7946 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1534, + "image_masks_loss": 0.0079, + "llm_learning_rate": 4.3897602059197104e-07, + "loss": 0.7886, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 4.38976020591971e-05, + "step": 7947 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1534, + "image_masks_loss": 0.0052, + "llm_learning_rate": 4.3850657626904715e-07, + "loss": 0.7388, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 4.385065762690471e-05, + "step": 7948 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1533, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.38037358962371e-07, + "loss": 0.7306, + "output_text_mask_loss": 0.0483, + "predictor_learning_rate": 4.3803735896237096e-05, + "step": 7949 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1533, + "image_masks_loss": 0.0089, + "llm_learning_rate": 4.3756836872361376e-07, + "loss": 0.6507, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 4.375683687236137e-05, + "step": 7950 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1533, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.370996056044216e-07, + "loss": 0.7248, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 4.370996056044216e-05, + "step": 7951 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1532, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.366310696564158e-07, + "loss": 0.7226, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 4.366310696564158e-05, + "step": 7952 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1532, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.3616276093119246e-07, + "loss": 0.6681, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 4.3616276093119243e-05, + "step": 7953 + }, + { + "epoch": 0.81, + "gumbel_tau": 0.1532, + "image_masks_loss": 0.0151, + "llm_learning_rate": 4.3569467948032294e-07, + "loss": 0.7769, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 4.3569467948032295e-05, + "step": 7954 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1531, + "image_masks_loss": 0.0081, + "llm_learning_rate": 4.352268253553535e-07, + "loss": 0.6469, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.352268253553535e-05, + "step": 7955 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1531, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.347591986078048e-07, + "loss": 0.7647, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 4.3475919860780475e-05, + "step": 7956 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1531, + "image_masks_loss": 0.0035, + "llm_learning_rate": 4.3429179928917373e-07, + "loss": 0.7505, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 4.3429179928917374e-05, + "step": 7957 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.153, + "image_masks_loss": 0.0199, + "llm_learning_rate": 4.338246274509306e-07, + "loss": 0.6718, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 4.338246274509306e-05, + "step": 7958 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.153, + "image_masks_loss": 0.0257, + "llm_learning_rate": 4.3335768314452184e-07, + "loss": 0.7279, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 4.3335768314452185e-05, + "step": 7959 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1529, + "image_masks_loss": 0.0058, + "llm_learning_rate": 4.328909664213679e-07, + "loss": 0.7494, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.328909664213679e-05, + "step": 7960 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1529, + "image_masks_loss": 0.0061, + "llm_learning_rate": 4.3242447733286536e-07, + "loss": 0.5738, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 4.324244773328653e-05, + "step": 7961 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1529, + "image_masks_loss": 0.008, + "llm_learning_rate": 4.319582159303848e-07, + "loss": 0.7814, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.319582159303848e-05, + "step": 7962 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1528, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.314921822652718e-07, + "loss": 0.8222, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 4.314921822652718e-05, + "step": 7963 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1528, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.3102637638884725e-07, + "loss": 0.7248, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 4.310263763888472e-05, + "step": 7964 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1528, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.3056079835240657e-07, + "loss": 0.6836, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 4.3056079835240655e-05, + "step": 7965 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1527, + "image_masks_loss": 0.0168, + "llm_learning_rate": 4.3009544820722007e-07, + "loss": 0.7474, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 4.3009544820722e-05, + "step": 7966 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1527, + "image_masks_loss": 0.0143, + "llm_learning_rate": 4.296303260045329e-07, + "loss": 0.7918, + "output_text_mask_loss": 0.0079, + "predictor_learning_rate": 4.2963032600453284e-05, + "step": 7967 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1527, + "image_masks_loss": 0.0045, + "llm_learning_rate": 4.291654317955665e-07, + "loss": 0.7508, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 4.291654317955665e-05, + "step": 7968 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1526, + "image_masks_loss": 0.02, + "llm_learning_rate": 4.2870076563151536e-07, + "loss": 0.7417, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.287007656315153e-05, + "step": 7969 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1526, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.282363275635493e-07, + "loss": 0.8024, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 4.282363275635492e-05, + "step": 7970 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1525, + "image_masks_loss": 0.0218, + "llm_learning_rate": 4.2777211764281376e-07, + "loss": 0.7518, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 4.277721176428137e-05, + "step": 7971 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1525, + "image_masks_loss": 0.0179, + "llm_learning_rate": 4.2730813592042813e-07, + "loss": 0.7501, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 4.273081359204281e-05, + "step": 7972 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1525, + "image_masks_loss": 0.0075, + "llm_learning_rate": 4.2684438244748744e-07, + "loss": 0.8467, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 4.268443824474874e-05, + "step": 7973 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1524, + "image_masks_loss": 0.0107, + "llm_learning_rate": 4.2638085727506105e-07, + "loss": 0.7261, + "output_text_mask_loss": 0.0406, + "predictor_learning_rate": 4.26380857275061e-05, + "step": 7974 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1524, + "image_masks_loss": 0.0039, + "llm_learning_rate": 4.2591756045419345e-07, + "loss": 0.7079, + "output_text_mask_loss": 0.0102, + "predictor_learning_rate": 4.2591756045419344e-05, + "step": 7975 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1524, + "image_masks_loss": 0.0202, + "llm_learning_rate": 4.254544920359041e-07, + "loss": 0.719, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 4.254544920359041e-05, + "step": 7976 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1523, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.2499165207118693e-07, + "loss": 0.7458, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.249916520711869e-05, + "step": 7977 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1523, + "image_masks_loss": 0.0097, + "llm_learning_rate": 4.2452904061101083e-07, + "loss": 0.6628, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 4.245290406110108e-05, + "step": 7978 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1523, + "image_masks_loss": 0.0054, + "llm_learning_rate": 4.240666577063191e-07, + "loss": 0.6357, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 4.24066657706319e-05, + "step": 7979 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1522, + "image_masks_loss": 0.0296, + "llm_learning_rate": 4.236045034080316e-07, + "loss": 0.7386, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 4.2360450340803156e-05, + "step": 7980 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1522, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.2314257776704116e-07, + "loss": 0.6495, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 4.231425777670411e-05, + "step": 7981 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1522, + "image_masks_loss": 0.0192, + "llm_learning_rate": 4.2268088083421616e-07, + "loss": 0.8183, + "output_text_mask_loss": 0.0196, + "predictor_learning_rate": 4.2268088083421615e-05, + "step": 7982 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1521, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.222194126603996e-07, + "loss": 0.7077, + "output_text_mask_loss": 0.0423, + "predictor_learning_rate": 4.222194126603995e-05, + "step": 7983 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1521, + "image_masks_loss": 0.0092, + "llm_learning_rate": 4.217581732964093e-07, + "loss": 0.7011, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 4.2175817329640934e-05, + "step": 7984 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.152, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.2129716279303817e-07, + "loss": 0.7284, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 4.212971627930382e-05, + "step": 7985 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.152, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.2083638120105296e-07, + "loss": 0.6975, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 4.2083638120105296e-05, + "step": 7986 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.152, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.203758285711973e-07, + "loss": 0.735, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 4.203758285711973e-05, + "step": 7987 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1519, + "image_masks_loss": 0.0309, + "llm_learning_rate": 4.1991550495418723e-07, + "loss": 0.7188, + "output_text_mask_loss": 0.0252, + "predictor_learning_rate": 4.1991550495418725e-05, + "step": 7988 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1519, + "image_masks_loss": 0.0124, + "llm_learning_rate": 4.1945541040071535e-07, + "loss": 0.7653, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 4.1945541040071535e-05, + "step": 7989 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1519, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.1899554496144774e-07, + "loss": 0.8018, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 4.1899554496144774e-05, + "step": 7990 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1518, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.1853590868702664e-07, + "loss": 0.6507, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 4.185359086870266e-05, + "step": 7991 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1518, + "image_masks_loss": 0.0226, + "llm_learning_rate": 4.1807650162806636e-07, + "loss": 0.6804, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 4.180765016280663e-05, + "step": 7992 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1518, + "image_masks_loss": 0.0227, + "llm_learning_rate": 4.176173238351594e-07, + "loss": 0.7539, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 4.176173238351594e-05, + "step": 7993 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1517, + "image_masks_loss": 0.0064, + "llm_learning_rate": 4.1715837535887123e-07, + "loss": 0.7149, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.1715837535887116e-05, + "step": 7994 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1517, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.166996562497422e-07, + "loss": 0.717, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 4.166996562497422e-05, + "step": 7995 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1516, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.1624116655828743e-07, + "loss": 0.7389, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.162411665582874e-05, + "step": 7996 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1516, + "image_masks_loss": 0.0066, + "llm_learning_rate": 4.1578290633499665e-07, + "loss": 0.7511, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 4.157829063349966e-05, + "step": 7997 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1516, + "image_masks_loss": 0.0188, + "llm_learning_rate": 4.1532487563033465e-07, + "loss": 0.7714, + "output_text_mask_loss": 0.0568, + "predictor_learning_rate": 4.153248756303346e-05, + "step": 7998 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1515, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.1486707449474e-07, + "loss": 0.6152, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 4.1486707449474e-05, + "step": 7999 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1515, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.144095029786285e-07, + "loss": 0.7395, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 4.144095029786285e-05, + "step": 8000 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1515, + "image_masks_loss": 0.0412, + "llm_learning_rate": 4.1395216113238794e-07, + "loss": 0.7534, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.1395216113238795e-05, + "step": 8001 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1514, + "image_masks_loss": 0.0311, + "llm_learning_rate": 4.134950490063816e-07, + "loss": 0.7603, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.134950490063816e-05, + "step": 8002 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1514, + "image_masks_loss": 0.0061, + "llm_learning_rate": 4.130381666509484e-07, + "loss": 0.79, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 4.130381666509484e-05, + "step": 8003 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1514, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.1258151411640053e-07, + "loss": 0.8268, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 4.125815141164005e-05, + "step": 8004 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1513, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.121250914530259e-07, + "loss": 0.7157, + "output_text_mask_loss": 0.0671, + "predictor_learning_rate": 4.121250914530258e-05, + "step": 8005 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1513, + "image_masks_loss": 0.025, + "llm_learning_rate": 4.116688987110862e-07, + "loss": 0.8052, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 4.1166889871108616e-05, + "step": 8006 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1513, + "image_masks_loss": 0.0357, + "llm_learning_rate": 4.112129359408193e-07, + "loss": 0.7258, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 4.112129359408193e-05, + "step": 8007 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1512, + "image_masks_loss": 0.0378, + "llm_learning_rate": 4.1075720319243675e-07, + "loss": 0.7615, + "output_text_mask_loss": 0.1938, + "predictor_learning_rate": 4.107572031924367e-05, + "step": 8008 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1512, + "image_masks_loss": 0.0115, + "llm_learning_rate": 4.1030170051612474e-07, + "loss": 0.7012, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 4.103017005161247e-05, + "step": 8009 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1511, + "image_masks_loss": 0.0071, + "llm_learning_rate": 4.098464279620443e-07, + "loss": 0.6662, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.098464279620443e-05, + "step": 8010 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1511, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.0939138558033e-07, + "loss": 0.7484, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.0939138558032996e-05, + "step": 8011 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1511, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.0893657342109345e-07, + "loss": 0.6749, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 4.0893657342109343e-05, + "step": 8012 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.151, + "image_masks_loss": 0.0311, + "llm_learning_rate": 4.08481991534419e-07, + "loss": 0.7105, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 4.08481991534419e-05, + "step": 8013 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.151, + "image_masks_loss": 0.0136, + "llm_learning_rate": 4.0802763997036644e-07, + "loss": 0.7184, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 4.080276399703664e-05, + "step": 8014 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.151, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.0757351877896977e-07, + "loss": 0.6572, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.0757351877896974e-05, + "step": 8015 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1509, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.0711962801023796e-07, + "loss": 0.7401, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.071196280102379e-05, + "step": 8016 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1509, + "image_masks_loss": 0.0125, + "llm_learning_rate": 4.0666596771415435e-07, + "loss": 0.6255, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.066659677141543e-05, + "step": 8017 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1509, + "image_masks_loss": 0.0069, + "llm_learning_rate": 4.0621253794067644e-07, + "loss": 0.7642, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.062125379406764e-05, + "step": 8018 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1508, + "image_masks_loss": 0.006, + "llm_learning_rate": 4.057593387397382e-07, + "loss": 0.7474, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 4.057593387397382e-05, + "step": 8019 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1508, + "image_masks_loss": 0.0294, + "llm_learning_rate": 4.053063701612464e-07, + "loss": 0.7466, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 4.053063701612464e-05, + "step": 8020 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1508, + "image_masks_loss": 0.0094, + "llm_learning_rate": 4.0485363225508304e-07, + "loss": 0.7766, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.04853632255083e-05, + "step": 8021 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1507, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.0440112507110413e-07, + "loss": 0.8023, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 4.044011250711041e-05, + "step": 8022 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1507, + "image_masks_loss": 0.019, + "llm_learning_rate": 4.039488486591414e-07, + "loss": 0.7185, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.0394884865914136e-05, + "step": 8023 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1507, + "image_masks_loss": 0.0067, + "llm_learning_rate": 4.034968030690001e-07, + "loss": 0.7804, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 4.03496803069e-05, + "step": 8024 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1506, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.0304498835045975e-07, + "loss": 0.7517, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 4.030449883504597e-05, + "step": 8025 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1506, + "image_masks_loss": 0.0061, + "llm_learning_rate": 4.02593404553277e-07, + "loss": 0.7764, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 4.0259340455327696e-05, + "step": 8026 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1505, + "image_masks_loss": 0.0137, + "llm_learning_rate": 4.021420517271801e-07, + "loss": 0.6961, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 4.021420517271801e-05, + "step": 8027 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1505, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.016909299218741e-07, + "loss": 0.6664, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.0169092992187404e-05, + "step": 8028 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1505, + "image_masks_loss": 0.0064, + "llm_learning_rate": 4.012400391870361e-07, + "loss": 0.7098, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 4.01240039187036e-05, + "step": 8029 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1504, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.0078937957231973e-07, + "loss": 0.6088, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.007893795723197e-05, + "step": 8030 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1504, + "image_masks_loss": 0.01, + "llm_learning_rate": 4.00338951127352e-07, + "loss": 0.7691, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 4.0033895112735196e-05, + "step": 8031 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1504, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.9988875390173623e-07, + "loss": 0.723, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 3.998887539017362e-05, + "step": 8032 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1503, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.99438787945049e-07, + "loss": 0.7377, + "output_text_mask_loss": 0.0347, + "predictor_learning_rate": 3.994387879450489e-05, + "step": 8033 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1503, + "image_masks_loss": 0.0056, + "llm_learning_rate": 3.989890533068411e-07, + "loss": 0.6649, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 3.989890533068411e-05, + "step": 8034 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1503, + "image_masks_loss": 0.022, + "llm_learning_rate": 3.985395500366385e-07, + "loss": 0.7524, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 3.985395500366384e-05, + "step": 8035 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1502, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.9809027818394156e-07, + "loss": 0.6859, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 3.980902781839416e-05, + "step": 8036 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1502, + "image_masks_loss": 0.0079, + "llm_learning_rate": 3.97641237798225e-07, + "loss": 0.7122, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.97641237798225e-05, + "step": 8037 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1502, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.971924289289375e-07, + "loss": 0.7267, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 3.971924289289375e-05, + "step": 8038 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1501, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.96743851625504e-07, + "loss": 0.7705, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 3.96743851625504e-05, + "step": 8039 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1501, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.9629550593732245e-07, + "loss": 0.7466, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 3.9629550593732245e-05, + "step": 8040 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.15, + "image_masks_loss": 0.0136, + "llm_learning_rate": 3.9584739191376564e-07, + "loss": 0.6753, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.9584739191376565e-05, + "step": 8041 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.15, + "image_masks_loss": 0.0035, + "llm_learning_rate": 3.95399509604181e-07, + "loss": 0.7176, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.95399509604181e-05, + "step": 8042 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.15, + "image_masks_loss": 0.0028, + "llm_learning_rate": 3.9495185905788993e-07, + "loss": 0.6976, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.949518590578899e-05, + "step": 8043 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1499, + "image_masks_loss": 0.0259, + "llm_learning_rate": 3.9450444032418905e-07, + "loss": 0.7775, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.94504440324189e-05, + "step": 8044 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1499, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.940572534523485e-07, + "loss": 0.7578, + "output_text_mask_loss": 0.0809, + "predictor_learning_rate": 3.9405725345234844e-05, + "step": 8045 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1499, + "image_masks_loss": 0.0209, + "llm_learning_rate": 3.936102984916146e-07, + "loss": 0.7476, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.936102984916146e-05, + "step": 8046 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1498, + "image_masks_loss": 0.017, + "llm_learning_rate": 3.931635754912069e-07, + "loss": 0.6293, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 3.931635754912069e-05, + "step": 8047 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1498, + "image_masks_loss": 0.0205, + "llm_learning_rate": 3.9271708450031864e-07, + "loss": 0.7328, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 3.9271708450031865e-05, + "step": 8048 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1498, + "image_masks_loss": 0.019, + "llm_learning_rate": 3.9227082556811874e-07, + "loss": 0.6609, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 3.922708255681187e-05, + "step": 8049 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1497, + "image_masks_loss": 0.0129, + "llm_learning_rate": 3.9182479874374957e-07, + "loss": 0.6343, + "output_text_mask_loss": 0.0183, + "predictor_learning_rate": 3.9182479874374956e-05, + "step": 8050 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1497, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.913790040763299e-07, + "loss": 0.6761, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 3.913790040763299e-05, + "step": 8051 + }, + { + "epoch": 0.82, + "gumbel_tau": 0.1497, + "image_masks_loss": 0.0307, + "llm_learning_rate": 3.909334416149513e-07, + "loss": 0.6629, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.9093344161495126e-05, + "step": 8052 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1496, + "image_masks_loss": 0.0194, + "llm_learning_rate": 3.9048811140867977e-07, + "loss": 0.7354, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 3.904881114086797e-05, + "step": 8053 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1496, + "image_masks_loss": 0.0224, + "llm_learning_rate": 3.900430135065564e-07, + "loss": 0.6888, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 3.900430135065563e-05, + "step": 8054 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1496, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.8959814795759584e-07, + "loss": 0.7226, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 3.895981479575958e-05, + "step": 8055 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1495, + "image_masks_loss": 0.0218, + "llm_learning_rate": 3.8915351481078804e-07, + "loss": 0.7703, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 3.89153514810788e-05, + "step": 8056 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1495, + "image_masks_loss": 0.008, + "llm_learning_rate": 3.8870911411509617e-07, + "loss": 0.673, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.887091141150961e-05, + "step": 8057 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1494, + "image_masks_loss": 0.0061, + "llm_learning_rate": 3.8826494591946014e-07, + "loss": 0.6441, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 3.882649459194601e-05, + "step": 8058 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1494, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.8782101027279194e-07, + "loss": 0.6264, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 3.8782101027279196e-05, + "step": 8059 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1494, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.873773072239789e-07, + "loss": 0.6134, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.873773072239789e-05, + "step": 8060 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1493, + "image_masks_loss": 0.0143, + "llm_learning_rate": 3.8693383682188224e-07, + "loss": 0.7367, + "output_text_mask_loss": 0.0435, + "predictor_learning_rate": 3.8693383682188225e-05, + "step": 8061 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1493, + "image_masks_loss": 0.0107, + "llm_learning_rate": 3.864905991153384e-07, + "loss": 0.7833, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 3.864905991153383e-05, + "step": 8062 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1493, + "image_masks_loss": 0.005, + "llm_learning_rate": 3.8604759415315716e-07, + "loss": 0.6573, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 3.860475941531572e-05, + "step": 8063 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1492, + "image_masks_loss": 0.0191, + "llm_learning_rate": 3.85604821984123e-07, + "loss": 0.7483, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.856048219841229e-05, + "step": 8064 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1492, + "image_masks_loss": 0.0047, + "llm_learning_rate": 3.851622826569967e-07, + "loss": 0.7241, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.851622826569967e-05, + "step": 8065 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1492, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.8471997622050977e-07, + "loss": 0.7395, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 3.847199762205097e-05, + "step": 8066 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1491, + "image_masks_loss": 0.0118, + "llm_learning_rate": 3.842779027233709e-07, + "loss": 0.7469, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 3.842779027233709e-05, + "step": 8067 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1491, + "image_masks_loss": 0.0132, + "llm_learning_rate": 3.8383606221426173e-07, + "loss": 0.8092, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 3.838360622142617e-05, + "step": 8068 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1491, + "image_masks_loss": 0.0123, + "llm_learning_rate": 3.833944547418392e-07, + "loss": 0.7332, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 3.8339445474183915e-05, + "step": 8069 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.149, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.829530803547329e-07, + "loss": 0.7617, + "output_text_mask_loss": 0.0484, + "predictor_learning_rate": 3.8295308035473294e-05, + "step": 8070 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.149, + "image_masks_loss": 0.0065, + "llm_learning_rate": 3.8251193910155e-07, + "loss": 0.6914, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.8251193910155e-05, + "step": 8071 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.149, + "image_masks_loss": 0.0119, + "llm_learning_rate": 3.820710310308684e-07, + "loss": 0.6414, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 3.820710310308684e-05, + "step": 8072 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1489, + "image_masks_loss": 0.0181, + "llm_learning_rate": 3.8163035619124265e-07, + "loss": 0.7291, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.816303561912426e-05, + "step": 8073 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1489, + "image_masks_loss": 0.0129, + "llm_learning_rate": 3.8118991463120063e-07, + "loss": 0.5774, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.811899146312006e-05, + "step": 8074 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1488, + "image_masks_loss": 0.0091, + "llm_learning_rate": 3.8074970639924433e-07, + "loss": 0.7926, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 3.8074970639924435e-05, + "step": 8075 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1488, + "image_masks_loss": 0.0059, + "llm_learning_rate": 3.8030973154385113e-07, + "loss": 0.7768, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 3.803097315438511e-05, + "step": 8076 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1488, + "image_masks_loss": 0.0042, + "llm_learning_rate": 3.798699901134711e-07, + "loss": 0.7577, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 3.798699901134711e-05, + "step": 8077 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1487, + "image_masks_loss": 0.017, + "llm_learning_rate": 3.7943048215653063e-07, + "loss": 0.7064, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.794304821565306e-05, + "step": 8078 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1487, + "image_masks_loss": 0.0095, + "llm_learning_rate": 3.789912077214289e-07, + "loss": 0.6785, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.789912077214289e-05, + "step": 8079 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1487, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.7855216685653976e-07, + "loss": 0.6789, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 3.7855216685653974e-05, + "step": 8080 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1486, + "image_masks_loss": 0.0123, + "llm_learning_rate": 3.7811335961021127e-07, + "loss": 0.8166, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.781133596102112e-05, + "step": 8081 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1486, + "image_masks_loss": 0.0215, + "llm_learning_rate": 3.776747860307661e-07, + "loss": 0.6366, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.776747860307661e-05, + "step": 8082 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1486, + "image_masks_loss": 0.0115, + "llm_learning_rate": 3.7723644616650066e-07, + "loss": 0.7055, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 3.772364461665007e-05, + "step": 8083 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1485, + "image_masks_loss": 0.0246, + "llm_learning_rate": 3.7679834006568603e-07, + "loss": 0.6387, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 3.76798340065686e-05, + "step": 8084 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1485, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.763604677765673e-07, + "loss": 0.7065, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.763604677765672e-05, + "step": 8085 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1485, + "image_masks_loss": 0.0081, + "llm_learning_rate": 3.7592282934736433e-07, + "loss": 0.6734, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 3.759228293473643e-05, + "step": 8086 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1484, + "image_masks_loss": 0.0091, + "llm_learning_rate": 3.7548542482627043e-07, + "loss": 0.8088, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 3.7548542482627044e-05, + "step": 8087 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1484, + "image_masks_loss": 0.0136, + "llm_learning_rate": 3.750482542614539e-07, + "loss": 0.6993, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.750482542614539e-05, + "step": 8088 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1484, + "image_masks_loss": 0.0105, + "llm_learning_rate": 3.7461131770105603e-07, + "loss": 0.7466, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.74611317701056e-05, + "step": 8089 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1483, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.7417461519319464e-07, + "loss": 0.7609, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.741746151931946e-05, + "step": 8090 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1483, + "image_masks_loss": 0.0102, + "llm_learning_rate": 3.7373814678595966e-07, + "loss": 0.6405, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 3.737381467859596e-05, + "step": 8091 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1483, + "image_masks_loss": 0.0039, + "llm_learning_rate": 3.7330191252741596e-07, + "loss": 0.6256, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.733019125274159e-05, + "step": 8092 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1482, + "image_masks_loss": 0.0026, + "llm_learning_rate": 3.728659124656028e-07, + "loss": 0.7816, + "output_text_mask_loss": 0.0075, + "predictor_learning_rate": 3.728659124656028e-05, + "step": 8093 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1482, + "image_masks_loss": 0.0152, + "llm_learning_rate": 3.724301466485333e-07, + "loss": 0.7945, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 3.724301466485333e-05, + "step": 8094 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1481, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.719946151241949e-07, + "loss": 0.6798, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 3.719946151241949e-05, + "step": 8095 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1481, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.7155931794054913e-07, + "loss": 0.7596, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.715593179405491e-05, + "step": 8096 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1481, + "image_masks_loss": 0.0147, + "llm_learning_rate": 3.711242551455324e-07, + "loss": 0.7631, + "output_text_mask_loss": 0.0033, + "predictor_learning_rate": 3.711242551455324e-05, + "step": 8097 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.148, + "image_masks_loss": 0.018, + "llm_learning_rate": 3.7068942678705505e-07, + "loss": 0.7961, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.70689426787055e-05, + "step": 8098 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.148, + "image_masks_loss": 0.0177, + "llm_learning_rate": 3.7025483291300043e-07, + "loss": 0.6821, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.702548329130004e-05, + "step": 8099 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.148, + "image_masks_loss": 0.0129, + "llm_learning_rate": 3.6982047357122767e-07, + "loss": 0.7633, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 3.6982047357122766e-05, + "step": 8100 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1479, + "image_masks_loss": 0.0056, + "llm_learning_rate": 3.693863488095692e-07, + "loss": 0.7123, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 3.693863488095692e-05, + "step": 8101 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1479, + "image_masks_loss": 0.012, + "llm_learning_rate": 3.689524586758317e-07, + "loss": 0.7593, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 3.689524586758317e-05, + "step": 8102 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1479, + "image_masks_loss": 0.012, + "llm_learning_rate": 3.685188032177961e-07, + "loss": 0.6043, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.685188032177961e-05, + "step": 8103 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1478, + "image_masks_loss": 0.0091, + "llm_learning_rate": 3.6808538248321757e-07, + "loss": 0.7145, + "output_text_mask_loss": 0.0024, + "predictor_learning_rate": 3.680853824832175e-05, + "step": 8104 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1478, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.676521965198254e-07, + "loss": 0.7053, + "output_text_mask_loss": 0.0123, + "predictor_learning_rate": 3.676521965198254e-05, + "step": 8105 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1478, + "image_masks_loss": 0.0192, + "llm_learning_rate": 3.672192453753229e-07, + "loss": 0.8305, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 3.672192453753229e-05, + "step": 8106 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1477, + "image_masks_loss": 0.0061, + "llm_learning_rate": 3.6678652909738796e-07, + "loss": 0.7742, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 3.667865290973879e-05, + "step": 8107 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1477, + "image_masks_loss": 0.0151, + "llm_learning_rate": 3.6635404773367163e-07, + "loss": 0.7548, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 3.663540477336716e-05, + "step": 8108 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1477, + "image_masks_loss": 0.0219, + "llm_learning_rate": 3.659218013317997e-07, + "loss": 0.6933, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.659218013317997e-05, + "step": 8109 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1476, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.6548978993937304e-07, + "loss": 0.6572, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.6548978993937305e-05, + "step": 8110 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1476, + "image_masks_loss": 0.013, + "llm_learning_rate": 3.65058013603965e-07, + "loss": 0.736, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 3.6505801360396495e-05, + "step": 8111 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1476, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.646264723731241e-07, + "loss": 0.7994, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 3.646264723731241e-05, + "step": 8112 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1475, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.641951662943724e-07, + "loss": 0.7464, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.641951662943724e-05, + "step": 8113 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1475, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.637640954152061e-07, + "loss": 0.769, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 3.637640954152061e-05, + "step": 8114 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1475, + "image_masks_loss": 0.0109, + "llm_learning_rate": 3.63333259783096e-07, + "loss": 0.748, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 3.63333259783096e-05, + "step": 8115 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1474, + "image_masks_loss": 0.0167, + "llm_learning_rate": 3.62902659445486e-07, + "loss": 0.6961, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 3.6290265944548596e-05, + "step": 8116 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1474, + "image_masks_loss": 0.0103, + "llm_learning_rate": 3.6247229444979604e-07, + "loss": 0.7294, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 3.6247229444979604e-05, + "step": 8117 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1473, + "image_masks_loss": 0.004, + "llm_learning_rate": 3.62042164843418e-07, + "loss": 0.6677, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 3.62042164843418e-05, + "step": 8118 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1473, + "image_masks_loss": 0.0177, + "llm_learning_rate": 3.6161227067371904e-07, + "loss": 0.7351, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 3.6161227067371905e-05, + "step": 8119 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1473, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.611826119880399e-07, + "loss": 0.7205, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.611826119880399e-05, + "step": 8120 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1472, + "image_masks_loss": 0.025, + "llm_learning_rate": 3.6075318883369543e-07, + "loss": 0.7022, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 3.607531888336954e-05, + "step": 8121 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1472, + "image_masks_loss": 0.0095, + "llm_learning_rate": 3.603240012579748e-07, + "loss": 0.7728, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.6032400125797475e-05, + "step": 8122 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1472, + "image_masks_loss": 0.0515, + "llm_learning_rate": 3.598950493081413e-07, + "loss": 0.7152, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 3.5989504930814124e-05, + "step": 8123 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1471, + "image_masks_loss": 0.0095, + "llm_learning_rate": 3.5946633303143155e-07, + "loss": 0.7058, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 3.5946633303143154e-05, + "step": 8124 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1471, + "image_masks_loss": 0.007, + "llm_learning_rate": 3.590378524750573e-07, + "loss": 0.7832, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 3.590378524750573e-05, + "step": 8125 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1471, + "image_masks_loss": 0.0074, + "llm_learning_rate": 3.586096076862033e-07, + "loss": 0.7588, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.5860960768620334e-05, + "step": 8126 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.147, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.5818159871202936e-07, + "loss": 0.7574, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 3.5818159871202934e-05, + "step": 8127 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.147, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.5775382559966766e-07, + "loss": 0.5915, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 3.577538255996676e-05, + "step": 8128 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.147, + "image_masks_loss": 0.0056, + "llm_learning_rate": 3.5732628839622694e-07, + "loss": 0.6642, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 3.57326288396227e-05, + "step": 8129 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1469, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.568989871487882e-07, + "loss": 0.6805, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 3.568989871487882e-05, + "step": 8130 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1469, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.5647192190440635e-07, + "loss": 0.7146, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 3.564719219044063e-05, + "step": 8131 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1469, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.5604509271011113e-07, + "loss": 0.6786, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 3.560450927101111e-05, + "step": 8132 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1468, + "image_masks_loss": 0.0156, + "llm_learning_rate": 3.556184996129056e-07, + "loss": 0.6084, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.556184996129055e-05, + "step": 8133 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1468, + "image_masks_loss": 0.006, + "llm_learning_rate": 3.551921426597674e-07, + "loss": 0.7796, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 3.5519214265976736e-05, + "step": 8134 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1468, + "image_masks_loss": 0.0101, + "llm_learning_rate": 3.547660218976473e-07, + "loss": 0.7461, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 3.5476602189764725e-05, + "step": 8135 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1467, + "image_masks_loss": 0.0164, + "llm_learning_rate": 3.543401373734717e-07, + "loss": 0.755, + "output_text_mask_loss": 0.0652, + "predictor_learning_rate": 3.543401373734717e-05, + "step": 8136 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1467, + "image_masks_loss": 0.004, + "llm_learning_rate": 3.5391448913413953e-07, + "loss": 0.6667, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 3.539144891341395e-05, + "step": 8137 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1467, + "image_masks_loss": 0.029, + "llm_learning_rate": 3.53489077226524e-07, + "loss": 0.7381, + "output_text_mask_loss": 0.0044, + "predictor_learning_rate": 3.5348907722652395e-05, + "step": 8138 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1466, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.5306390169747295e-07, + "loss": 0.6928, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 3.530639016974729e-05, + "step": 8139 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1466, + "image_masks_loss": 0.0072, + "llm_learning_rate": 3.5263896259380647e-07, + "loss": 0.6528, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.526389625938065e-05, + "step": 8140 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1465, + "image_masks_loss": 0.0086, + "llm_learning_rate": 3.5221425996232016e-07, + "loss": 0.685, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.522142599623202e-05, + "step": 8141 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1465, + "image_masks_loss": 0.0067, + "llm_learning_rate": 3.517897938497841e-07, + "loss": 0.678, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.517897938497841e-05, + "step": 8142 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1465, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.5136556430294097e-07, + "loss": 0.7424, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 3.513655643029409e-05, + "step": 8143 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1464, + "image_masks_loss": 0.0096, + "llm_learning_rate": 3.5094157136850743e-07, + "loss": 0.7458, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 3.509415713685074e-05, + "step": 8144 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1464, + "image_masks_loss": 0.0073, + "llm_learning_rate": 3.505178150931751e-07, + "loss": 0.7293, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 3.505178150931751e-05, + "step": 8145 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1464, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.500942955236089e-07, + "loss": 0.6595, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 3.500942955236089e-05, + "step": 8146 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1463, + "image_masks_loss": 0.0064, + "llm_learning_rate": 3.4967101270644734e-07, + "loss": 0.7651, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 3.496710127064473e-05, + "step": 8147 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1463, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.4924796668830306e-07, + "loss": 0.719, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.4924796668830304e-05, + "step": 8148 + }, + { + "epoch": 0.83, + "gumbel_tau": 0.1463, + "image_masks_loss": 0.0036, + "llm_learning_rate": 3.488251575157639e-07, + "loss": 0.7294, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 3.488251575157639e-05, + "step": 8149 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1462, + "image_masks_loss": 0.0169, + "llm_learning_rate": 3.4840258523539006e-07, + "loss": 0.8495, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 3.4840258523539e-05, + "step": 8150 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1462, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.47980249893716e-07, + "loss": 0.7538, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 3.4798024989371596e-05, + "step": 8151 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1462, + "image_masks_loss": 0.0068, + "llm_learning_rate": 3.4755815153725e-07, + "loss": 0.7022, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.4755815153725e-05, + "step": 8152 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1461, + "image_masks_loss": 0.0047, + "llm_learning_rate": 3.4713629021247505e-07, + "loss": 0.7003, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 3.47136290212475e-05, + "step": 8153 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1461, + "image_masks_loss": 0.0083, + "llm_learning_rate": 3.4671466596584683e-07, + "loss": 0.8209, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 3.467146659658468e-05, + "step": 8154 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1461, + "image_masks_loss": 0.0081, + "llm_learning_rate": 3.462932788437956e-07, + "loss": 0.7254, + "output_text_mask_loss": 0.011, + "predictor_learning_rate": 3.4629327884379556e-05, + "step": 8155 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.146, + "image_masks_loss": 0.0046, + "llm_learning_rate": 3.458721288927261e-07, + "loss": 0.7248, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 3.458721288927261e-05, + "step": 8156 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.146, + "image_masks_loss": 0.0117, + "llm_learning_rate": 3.4545121615901596e-07, + "loss": 0.7473, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 3.454512161590159e-05, + "step": 8157 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.146, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.4503054068901753e-07, + "loss": 0.7718, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.450305406890175e-05, + "step": 8158 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1459, + "image_masks_loss": 0.0048, + "llm_learning_rate": 3.446101025290552e-07, + "loss": 0.8358, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 3.446101025290552e-05, + "step": 8159 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1459, + "image_masks_loss": 0.0167, + "llm_learning_rate": 3.441899017254294e-07, + "loss": 0.7314, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 3.441899017254294e-05, + "step": 8160 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1459, + "image_masks_loss": 0.0062, + "llm_learning_rate": 3.437699383244131e-07, + "loss": 0.7227, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 3.437699383244131e-05, + "step": 8161 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1458, + "image_masks_loss": 0.0213, + "llm_learning_rate": 3.4335021237225456e-07, + "loss": 0.8354, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 3.4335021237225455e-05, + "step": 8162 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1458, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.4293072391517414e-07, + "loss": 0.7088, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.429307239151741e-05, + "step": 8163 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1458, + "image_masks_loss": 0.0128, + "llm_learning_rate": 3.425114729993673e-07, + "loss": 0.7411, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 3.425114729993672e-05, + "step": 8164 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1457, + "image_masks_loss": 0.02, + "llm_learning_rate": 3.4209245967100265e-07, + "loss": 0.7297, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 3.420924596710026e-05, + "step": 8165 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1457, + "image_masks_loss": 0.0085, + "llm_learning_rate": 3.4167368397622276e-07, + "loss": 0.7779, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.416736839762227e-05, + "step": 8166 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1457, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.4125514596114463e-07, + "loss": 0.8044, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 3.412551459611446e-05, + "step": 8167 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1456, + "image_masks_loss": 0.0071, + "llm_learning_rate": 3.4083684567185723e-07, + "loss": 0.6908, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.408368456718572e-05, + "step": 8168 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1456, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.4041878315442645e-07, + "loss": 0.7081, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.404187831544264e-05, + "step": 8169 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1456, + "image_masks_loss": 0.0101, + "llm_learning_rate": 3.400009584548897e-07, + "loss": 0.6697, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 3.400009584548897e-05, + "step": 8170 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1455, + "image_masks_loss": 0.024, + "llm_learning_rate": 3.395833716192584e-07, + "loss": 0.7398, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.395833716192584e-05, + "step": 8171 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1455, + "image_masks_loss": 0.0037, + "llm_learning_rate": 3.3916602269351866e-07, + "loss": 0.7342, + "output_text_mask_loss": 0.0098, + "predictor_learning_rate": 3.391660226935187e-05, + "step": 8172 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1454, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.387489117236292e-07, + "loss": 0.7057, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.387489117236292e-05, + "step": 8173 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1454, + "image_masks_loss": 0.0145, + "llm_learning_rate": 3.3833203875552387e-07, + "loss": 0.7175, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 3.3833203875552385e-05, + "step": 8174 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1454, + "image_masks_loss": 0.0142, + "llm_learning_rate": 3.3791540383510873e-07, + "loss": 0.7081, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 3.379154038351087e-05, + "step": 8175 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1453, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.374990070082662e-07, + "loss": 0.7794, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 3.374990070082662e-05, + "step": 8176 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1453, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.370828483208491e-07, + "loss": 0.6548, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 3.370828483208491e-05, + "step": 8177 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1453, + "image_masks_loss": 0.0096, + "llm_learning_rate": 3.366669278186868e-07, + "loss": 0.7856, + "output_text_mask_loss": 0.0267, + "predictor_learning_rate": 3.366669278186868e-05, + "step": 8178 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1452, + "image_masks_loss": 0.0154, + "llm_learning_rate": 3.3625124554758083e-07, + "loss": 0.736, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.362512455475808e-05, + "step": 8179 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1452, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.358358015533067e-07, + "loss": 0.7965, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 3.358358015533067e-05, + "step": 8180 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1452, + "image_masks_loss": 0.0078, + "llm_learning_rate": 3.3542059588161535e-07, + "loss": 0.7646, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 3.354205958816153e-05, + "step": 8181 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1451, + "image_masks_loss": 0.0086, + "llm_learning_rate": 3.350056285782291e-07, + "loss": 0.7308, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 3.3500562857822905e-05, + "step": 8182 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1451, + "image_masks_loss": 0.0123, + "llm_learning_rate": 3.345908996888456e-07, + "loss": 0.7262, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 3.3459089968884555e-05, + "step": 8183 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1451, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.341764092591354e-07, + "loss": 0.7212, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.341764092591354e-05, + "step": 8184 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.145, + "image_masks_loss": 0.0065, + "llm_learning_rate": 3.3376215733474327e-07, + "loss": 0.6792, + "output_text_mask_loss": 0.0591, + "predictor_learning_rate": 3.3376215733474325e-05, + "step": 8185 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.145, + "image_masks_loss": 0.0131, + "llm_learning_rate": 3.3334814396128736e-07, + "loss": 0.7322, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.333481439612873e-05, + "step": 8186 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.145, + "image_masks_loss": 0.005, + "llm_learning_rate": 3.3293436918435923e-07, + "loss": 0.7508, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.329343691843592e-05, + "step": 8187 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1449, + "image_masks_loss": 0.0186, + "llm_learning_rate": 3.325208330495261e-07, + "loss": 0.7257, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 3.325208330495261e-05, + "step": 8188 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1449, + "image_masks_loss": 0.0084, + "llm_learning_rate": 3.321075356023265e-07, + "loss": 0.66, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.321075356023265e-05, + "step": 8189 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1449, + "image_masks_loss": 0.0123, + "llm_learning_rate": 3.3169447688827404e-07, + "loss": 0.7062, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 3.31694476888274e-05, + "step": 8190 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1448, + "image_masks_loss": 0.0187, + "llm_learning_rate": 3.3128165695285543e-07, + "loss": 0.7648, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.312816569528554e-05, + "step": 8191 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1448, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.3086907584153145e-07, + "loss": 0.7251, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.3086907584153144e-05, + "step": 8192 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1448, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.304567335997361e-07, + "loss": 0.7058, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 3.304567335997361e-05, + "step": 8193 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1447, + "image_masks_loss": 0.0267, + "llm_learning_rate": 3.300446302728774e-07, + "loss": 0.8069, + "output_text_mask_loss": 0.0366, + "predictor_learning_rate": 3.300446302728774e-05, + "step": 8194 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1447, + "image_masks_loss": 0.0042, + "llm_learning_rate": 3.2963276590633864e-07, + "loss": 0.6874, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 3.296327659063386e-05, + "step": 8195 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1447, + "image_masks_loss": 0.0052, + "llm_learning_rate": 3.2922114054547316e-07, + "loss": 0.7817, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 3.2922114054547315e-05, + "step": 8196 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1446, + "image_masks_loss": 0.0049, + "llm_learning_rate": 3.288097542356111e-07, + "loss": 0.709, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.288097542356111e-05, + "step": 8197 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1446, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.283986070220549e-07, + "loss": 0.7379, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 3.2839860702205486e-05, + "step": 8198 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1446, + "image_masks_loss": 0.0077, + "llm_learning_rate": 3.279876989500813e-07, + "loss": 0.8031, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 3.279876989500813e-05, + "step": 8199 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1445, + "image_masks_loss": 0.007, + "llm_learning_rate": 3.2757703006493974e-07, + "loss": 0.7298, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.275770300649397e-05, + "step": 8200 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1445, + "image_masks_loss": 0.0164, + "llm_learning_rate": 3.2716660041185483e-07, + "loss": 0.7067, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 3.271666004118548e-05, + "step": 8201 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1445, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.267564100360238e-07, + "loss": 0.7472, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.267564100360237e-05, + "step": 8202 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1444, + "image_masks_loss": 0.0157, + "llm_learning_rate": 3.2634645898261764e-07, + "loss": 0.7664, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.263464589826176e-05, + "step": 8203 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1444, + "image_masks_loss": 0.0035, + "llm_learning_rate": 3.2593674729678115e-07, + "loss": 0.7088, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 3.259367472967811e-05, + "step": 8204 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1444, + "image_masks_loss": 0.0147, + "llm_learning_rate": 3.255272750236324e-07, + "loss": 0.5608, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.255272750236324e-05, + "step": 8205 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1443, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.251180422082637e-07, + "loss": 0.7927, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 3.251180422082636e-05, + "step": 8206 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1443, + "image_masks_loss": 0.0296, + "llm_learning_rate": 3.2470904889573966e-07, + "loss": 0.7728, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 3.247090488957396e-05, + "step": 8207 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1443, + "image_masks_loss": 0.0073, + "llm_learning_rate": 3.243002951311014e-07, + "loss": 0.7144, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 3.2430029513110136e-05, + "step": 8208 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1442, + "image_masks_loss": 0.0078, + "llm_learning_rate": 3.238917809593606e-07, + "loss": 0.6971, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 3.2389178095936054e-05, + "step": 8209 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1442, + "image_masks_loss": 0.0037, + "llm_learning_rate": 3.2348350642550403e-07, + "loss": 0.7483, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.23483506425504e-05, + "step": 8210 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1441, + "image_masks_loss": 0.0063, + "llm_learning_rate": 3.230754715744919e-07, + "loss": 0.8513, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 3.2307547157449185e-05, + "step": 8211 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1441, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.2266767645125734e-07, + "loss": 0.656, + "output_text_mask_loss": 0.0067, + "predictor_learning_rate": 3.2266767645125735e-05, + "step": 8212 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1441, + "image_masks_loss": 0.0039, + "llm_learning_rate": 3.222601211007084e-07, + "loss": 0.792, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 3.222601211007084e-05, + "step": 8213 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.144, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.218528055677256e-07, + "loss": 0.7767, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 3.2185280556772554e-05, + "step": 8214 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.144, + "image_masks_loss": 0.0112, + "llm_learning_rate": 3.214457298971635e-07, + "loss": 0.7741, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.2144572989716346e-05, + "step": 8215 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.144, + "image_masks_loss": 0.0068, + "llm_learning_rate": 3.210388941338499e-07, + "loss": 0.7913, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 3.210388941338499e-05, + "step": 8216 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1439, + "image_masks_loss": 0.0124, + "llm_learning_rate": 3.206322983225871e-07, + "loss": 0.7099, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 3.206322983225871e-05, + "step": 8217 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1439, + "image_masks_loss": 0.0063, + "llm_learning_rate": 3.202259425081497e-07, + "loss": 0.7214, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.202259425081497e-05, + "step": 8218 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1439, + "image_masks_loss": 0.0153, + "llm_learning_rate": 3.1981982673528645e-07, + "loss": 0.7362, + "output_text_mask_loss": 0.0125, + "predictor_learning_rate": 3.198198267352864e-05, + "step": 8219 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1438, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.194139510487204e-07, + "loss": 0.6683, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.194139510487204e-05, + "step": 8220 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1438, + "image_masks_loss": 0.0136, + "llm_learning_rate": 3.1900831549314727e-07, + "loss": 0.787, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.1900831549314727e-05, + "step": 8221 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1438, + "image_masks_loss": 0.0071, + "llm_learning_rate": 3.1860292011323625e-07, + "loss": 0.6521, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 3.186029201132362e-05, + "step": 8222 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1437, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.1819776495363053e-07, + "loss": 0.7855, + "output_text_mask_loss": 0.0094, + "predictor_learning_rate": 3.181977649536305e-05, + "step": 8223 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1437, + "image_masks_loss": 0.0087, + "llm_learning_rate": 3.1779285005894693e-07, + "loss": 0.7557, + "output_text_mask_loss": 0.0152, + "predictor_learning_rate": 3.1779285005894693e-05, + "step": 8224 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1437, + "image_masks_loss": 0.0133, + "llm_learning_rate": 3.173881754737751e-07, + "loss": 0.7167, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.173881754737751e-05, + "step": 8225 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1436, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.169837412426785e-07, + "loss": 0.7132, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 3.169837412426785e-05, + "step": 8226 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1436, + "image_masks_loss": 0.0116, + "llm_learning_rate": 3.1657954741019544e-07, + "loss": 0.6535, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 3.165795474101954e-05, + "step": 8227 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1436, + "image_masks_loss": 0.0167, + "llm_learning_rate": 3.161755940208358e-07, + "loss": 0.7271, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 3.161755940208358e-05, + "step": 8228 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1435, + "image_masks_loss": 0.0155, + "llm_learning_rate": 3.1577188111908407e-07, + "loss": 0.7743, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.157718811190841e-05, + "step": 8229 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1435, + "image_masks_loss": 0.0282, + "llm_learning_rate": 3.1536840874939826e-07, + "loss": 0.7437, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 3.1536840874939825e-05, + "step": 8230 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1435, + "image_masks_loss": 0.0143, + "llm_learning_rate": 3.14965176956209e-07, + "loss": 0.7426, + "output_text_mask_loss": 0.0648, + "predictor_learning_rate": 3.1496517695620894e-05, + "step": 8231 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1434, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.145621857839215e-07, + "loss": 0.7108, + "output_text_mask_loss": 0.0908, + "predictor_learning_rate": 3.1456218578392146e-05, + "step": 8232 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1434, + "image_masks_loss": 0.0138, + "llm_learning_rate": 3.1415943527691425e-07, + "loss": 0.686, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 3.141594352769142e-05, + "step": 8233 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1434, + "image_masks_loss": 0.007, + "llm_learning_rate": 3.137569254795383e-07, + "loss": 0.7865, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.137569254795383e-05, + "step": 8234 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1433, + "image_masks_loss": 0.0036, + "llm_learning_rate": 3.133546564361198e-07, + "loss": 0.7039, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 3.133546564361198e-05, + "step": 8235 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1433, + "image_masks_loss": 0.005, + "llm_learning_rate": 3.12952628190957e-07, + "loss": 0.6831, + "output_text_mask_loss": 0.0104, + "predictor_learning_rate": 3.12952628190957e-05, + "step": 8236 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1433, + "image_masks_loss": 0.0099, + "llm_learning_rate": 3.125508407883221e-07, + "loss": 0.7026, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 3.1255084078832206e-05, + "step": 8237 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1432, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.1214929427246104e-07, + "loss": 0.6778, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.12149294272461e-05, + "step": 8238 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1432, + "image_masks_loss": 0.0058, + "llm_learning_rate": 3.1174798868759237e-07, + "loss": 0.7137, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.1174798868759234e-05, + "step": 8239 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1432, + "image_masks_loss": 0.0091, + "llm_learning_rate": 3.113469240779099e-07, + "loss": 0.8276, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 3.1134692407790985e-05, + "step": 8240 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1431, + "image_masks_loss": 0.0074, + "llm_learning_rate": 3.109461004875791e-07, + "loss": 0.7041, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.109461004875791e-05, + "step": 8241 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1431, + "image_masks_loss": 0.0136, + "llm_learning_rate": 3.1054551796073974e-07, + "loss": 0.7194, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.105455179607397e-05, + "step": 8242 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1431, + "image_masks_loss": 0.0057, + "llm_learning_rate": 3.1014517654150475e-07, + "loss": 0.7561, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 3.101451765415047e-05, + "step": 8243 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.143, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.0974507627396064e-07, + "loss": 0.8011, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 3.097450762739606e-05, + "step": 8244 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.143, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.093452172021674e-07, + "loss": 0.7106, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 3.0934521720216744e-05, + "step": 8245 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.143, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.0894559937015763e-07, + "loss": 0.7526, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 3.089455993701576e-05, + "step": 8246 + }, + { + "epoch": 0.84, + "gumbel_tau": 0.1429, + "image_masks_loss": 0.018, + "llm_learning_rate": 3.0854622282193933e-07, + "loss": 0.7227, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 3.0854622282193934e-05, + "step": 8247 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1429, + "image_masks_loss": 0.0092, + "llm_learning_rate": 3.081470876014925e-07, + "loss": 0.5913, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 3.0814708760149244e-05, + "step": 8248 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1429, + "image_masks_loss": 0.0102, + "llm_learning_rate": 3.077481937527704e-07, + "loss": 0.7635, + "output_text_mask_loss": 0.0094, + "predictor_learning_rate": 3.077481937527704e-05, + "step": 8249 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1428, + "image_masks_loss": 0.0165, + "llm_learning_rate": 3.073495413197003e-07, + "loss": 0.7087, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.073495413197003e-05, + "step": 8250 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1428, + "image_masks_loss": 0.0114, + "llm_learning_rate": 3.069511303461825e-07, + "loss": 0.6567, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 3.069511303461825e-05, + "step": 8251 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1428, + "image_masks_loss": 0.0007, + "llm_learning_rate": 3.0655296087609115e-07, + "loss": 0.6852, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 3.065529608760911e-05, + "step": 8252 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1427, + "image_masks_loss": 0.0164, + "llm_learning_rate": 3.0615503295327366e-07, + "loss": 0.6677, + "output_text_mask_loss": 0.041, + "predictor_learning_rate": 3.061550329532736e-05, + "step": 8253 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1427, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.057573466215502e-07, + "loss": 0.7526, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 3.057573466215502e-05, + "step": 8254 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1427, + "image_masks_loss": 0.0041, + "llm_learning_rate": 3.053599019247153e-07, + "loss": 0.6776, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.053599019247153e-05, + "step": 8255 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1426, + "image_masks_loss": 0.0065, + "llm_learning_rate": 3.049626989065366e-07, + "loss": 0.7457, + "output_text_mask_loss": 0.0387, + "predictor_learning_rate": 3.0496269890653657e-05, + "step": 8256 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1426, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.045657376107544e-07, + "loss": 0.754, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 3.045657376107544e-05, + "step": 8257 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1426, + "image_masks_loss": 0.0074, + "llm_learning_rate": 3.04169018081083e-07, + "loss": 0.7419, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 3.04169018081083e-05, + "step": 8258 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1425, + "image_masks_loss": 0.0108, + "llm_learning_rate": 3.0377254036121103e-07, + "loss": 0.8382, + "output_text_mask_loss": 0.0542, + "predictor_learning_rate": 3.03772540361211e-05, + "step": 8259 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1425, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.0337630449479847e-07, + "loss": 0.6738, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 3.0337630449479847e-05, + "step": 8260 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1425, + "image_masks_loss": 0.0101, + "llm_learning_rate": 3.029803105254803e-07, + "loss": 0.6851, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.0298031052548025e-05, + "step": 8261 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1424, + "image_masks_loss": 0.0161, + "llm_learning_rate": 3.02584558496864e-07, + "loss": 0.6913, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 3.0258455849686395e-05, + "step": 8262 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1424, + "image_masks_loss": 0.0041, + "llm_learning_rate": 3.0218904845253055e-07, + "loss": 0.6424, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.0218904845253054e-05, + "step": 8263 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1424, + "image_masks_loss": 0.0149, + "llm_learning_rate": 3.0179378043603436e-07, + "loss": 0.6953, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.0179378043603434e-05, + "step": 8264 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1423, + "image_masks_loss": 0.0059, + "llm_learning_rate": 3.0139875449090305e-07, + "loss": 0.7422, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 3.0139875449090303e-05, + "step": 8265 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1423, + "image_masks_loss": 0.0063, + "llm_learning_rate": 3.010039706606388e-07, + "loss": 0.7223, + "output_text_mask_loss": 0.0057, + "predictor_learning_rate": 3.010039706606388e-05, + "step": 8266 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1423, + "image_masks_loss": 0.0057, + "llm_learning_rate": 3.006094289887151e-07, + "loss": 0.6905, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.0060942898871513e-05, + "step": 8267 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1422, + "image_masks_loss": 0.0087, + "llm_learning_rate": 3.0021512951858013e-07, + "loss": 0.7362, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.002151295185801e-05, + "step": 8268 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1422, + "image_masks_loss": 0.0055, + "llm_learning_rate": 2.998210722936554e-07, + "loss": 0.7445, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.9982107229365536e-05, + "step": 8269 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1422, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.9942725735733454e-07, + "loss": 0.6375, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 2.994272573573345e-05, + "step": 8270 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1421, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.9903368475298484e-07, + "loss": 0.7224, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.9903368475298483e-05, + "step": 8271 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1421, + "image_masks_loss": 0.0017, + "llm_learning_rate": 2.986403545239491e-07, + "loss": 0.7627, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 2.986403545239491e-05, + "step": 8272 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1421, + "image_masks_loss": 0.0135, + "llm_learning_rate": 2.9824726671354055e-07, + "loss": 0.6724, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.9824726671354052e-05, + "step": 8273 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.142, + "image_masks_loss": 0.0132, + "llm_learning_rate": 2.978544213650475e-07, + "loss": 0.6687, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.978544213650475e-05, + "step": 8274 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.142, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.974618185217304e-07, + "loss": 0.8003, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 2.974618185217304e-05, + "step": 8275 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.142, + "image_masks_loss": 0.0111, + "llm_learning_rate": 2.970694582268241e-07, + "loss": 0.7562, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 2.970694582268241e-05, + "step": 8276 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1419, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.966773405235357e-07, + "loss": 0.7411, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 2.966773405235357e-05, + "step": 8277 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1419, + "image_masks_loss": 0.0129, + "llm_learning_rate": 2.9628546545504566e-07, + "loss": 0.7181, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.9628546545504565e-05, + "step": 8278 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1419, + "image_masks_loss": 0.0154, + "llm_learning_rate": 2.9589383306450947e-07, + "loss": 0.7068, + "output_text_mask_loss": 0.0679, + "predictor_learning_rate": 2.9589383306450945e-05, + "step": 8279 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1418, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.955024433950537e-07, + "loss": 0.753, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 2.955024433950537e-05, + "step": 8280 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1418, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.9511129648977927e-07, + "loss": 0.7753, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 2.9511129648977923e-05, + "step": 8281 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1418, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.947203923917599e-07, + "loss": 0.7024, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 2.947203923917599e-05, + "step": 8282 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1417, + "image_masks_loss": 0.0278, + "llm_learning_rate": 2.9432973114404324e-07, + "loss": 0.6565, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 2.9432973114404322e-05, + "step": 8283 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1417, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.939393127896492e-07, + "loss": 0.7967, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.939393127896492e-05, + "step": 8284 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1417, + "image_masks_loss": 0.0043, + "llm_learning_rate": 2.935491373715715e-07, + "loss": 0.7335, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 2.9354913737157153e-05, + "step": 8285 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1416, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.931592049327781e-07, + "loss": 0.7372, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 2.9315920493277802e-05, + "step": 8286 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1416, + "image_masks_loss": 0.0038, + "llm_learning_rate": 2.927695155162086e-07, + "loss": 0.6499, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 2.9276951551620856e-05, + "step": 8287 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1416, + "image_masks_loss": 0.0057, + "llm_learning_rate": 2.9238006916477683e-07, + "loss": 0.7123, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 2.9238006916477682e-05, + "step": 8288 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1415, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.9199086592136853e-07, + "loss": 0.7172, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 2.9199086592136852e-05, + "step": 8289 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1415, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.9160190582884383e-07, + "loss": 0.6909, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.9160190582884383e-05, + "step": 8290 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1415, + "image_masks_loss": 0.003, + "llm_learning_rate": 2.912131889300368e-07, + "loss": 0.8316, + "output_text_mask_loss": 0.0561, + "predictor_learning_rate": 2.912131889300368e-05, + "step": 8291 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1414, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.908247152677535e-07, + "loss": 0.7236, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.9082471526775345e-05, + "step": 8292 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1414, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.9043648488477293e-07, + "loss": 0.7388, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.9043648488477293e-05, + "step": 8293 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1414, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.900484978238485e-07, + "loss": 0.7886, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 2.900484978238485e-05, + "step": 8294 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1413, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.896607541277058e-07, + "loss": 0.6946, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.8966075412770575e-05, + "step": 8295 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1413, + "image_masks_loss": 0.0147, + "llm_learning_rate": 2.892732538390444e-07, + "loss": 0.7523, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 2.892732538390444e-05, + "step": 8296 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1413, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.888859970005356e-07, + "loss": 0.7747, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.8888599700053558e-05, + "step": 8297 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1412, + "image_masks_loss": 0.0026, + "llm_learning_rate": 2.8849898365482686e-07, + "loss": 0.6194, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.8849898365482686e-05, + "step": 8298 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1412, + "image_masks_loss": 0.0032, + "llm_learning_rate": 2.881122138445355e-07, + "loss": 0.7123, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 2.881122138445355e-05, + "step": 8299 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1412, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.877256876122542e-07, + "loss": 0.7131, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.8772568761225415e-05, + "step": 8300 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1411, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.873394050005479e-07, + "loss": 0.7006, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.873394050005479e-05, + "step": 8301 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1411, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.8695336605195466e-07, + "loss": 0.7336, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 2.8695336605195467e-05, + "step": 8302 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1411, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.8656757080898627e-07, + "loss": 0.6964, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 2.8656757080898628e-05, + "step": 8303 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.141, + "image_masks_loss": 0.0091, + "llm_learning_rate": 2.861820193141265e-07, + "loss": 0.8073, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 2.861820193141265e-05, + "step": 8304 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.141, + "image_masks_loss": 0.0175, + "llm_learning_rate": 2.857967116098345e-07, + "loss": 0.7488, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.857967116098345e-05, + "step": 8305 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.141, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.85411647738541e-07, + "loss": 0.7228, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.85411647738541e-05, + "step": 8306 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1409, + "image_masks_loss": 0.0144, + "llm_learning_rate": 2.8502682774264935e-07, + "loss": 0.5893, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.850268277426493e-05, + "step": 8307 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1409, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.8464225166453695e-07, + "loss": 0.7369, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 2.8464225166453693e-05, + "step": 8308 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1409, + "image_masks_loss": 0.0096, + "llm_learning_rate": 2.842579195465542e-07, + "loss": 0.7593, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 2.8425791954655416e-05, + "step": 8309 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1408, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.8387383143102435e-07, + "loss": 0.696, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 2.8387383143102434e-05, + "step": 8310 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1408, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.834899873602448e-07, + "loss": 0.775, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.834899873602448e-05, + "step": 8311 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1408, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.83106387376485e-07, + "loss": 0.6836, + "output_text_mask_loss": 0.0341, + "predictor_learning_rate": 2.83106387376485e-05, + "step": 8312 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1407, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.8272303152198783e-07, + "loss": 0.8006, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 2.827230315219878e-05, + "step": 8313 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1407, + "image_masks_loss": 0.0182, + "llm_learning_rate": 2.8233991983896924e-07, + "loss": 0.7679, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.8233991983896924e-05, + "step": 8314 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1407, + "image_masks_loss": 0.0182, + "llm_learning_rate": 2.819570523696183e-07, + "loss": 0.7727, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 2.8195705236961826e-05, + "step": 8315 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1406, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.815744291560971e-07, + "loss": 0.6754, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 2.8157442915609704e-05, + "step": 8316 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1406, + "image_masks_loss": 0.0055, + "llm_learning_rate": 2.8119205024054075e-07, + "loss": 0.6629, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.8119205024054073e-05, + "step": 8317 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1406, + "image_masks_loss": 0.0257, + "llm_learning_rate": 2.8080991566505845e-07, + "loss": 0.6732, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 2.8080991566505843e-05, + "step": 8318 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1405, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.804280254717312e-07, + "loss": 0.7283, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.8042802547173117e-05, + "step": 8319 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1405, + "image_masks_loss": 0.0032, + "llm_learning_rate": 2.8004637970261374e-07, + "loss": 0.7485, + "output_text_mask_loss": 0.0117, + "predictor_learning_rate": 2.800463797026137e-05, + "step": 8320 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1405, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.7966497839973376e-07, + "loss": 0.6801, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 2.7966497839973377e-05, + "step": 8321 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1404, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.7928382160509195e-07, + "loss": 0.6933, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.7928382160509192e-05, + "step": 8322 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1404, + "image_masks_loss": 0.0191, + "llm_learning_rate": 2.789029093606624e-07, + "loss": 0.7346, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 2.7890290936066236e-05, + "step": 8323 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1404, + "image_masks_loss": 0.0091, + "llm_learning_rate": 2.7852224170839076e-07, + "loss": 0.7866, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 2.7852224170839074e-05, + "step": 8324 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1403, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.7814181869019935e-07, + "loss": 0.696, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 2.7814181869019934e-05, + "step": 8325 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1403, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.777616403479794e-07, + "loss": 0.7432, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.7776164034797935e-05, + "step": 8326 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1403, + "image_masks_loss": 0.0053, + "llm_learning_rate": 2.7738170672359733e-07, + "loss": 0.791, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 2.773817067235973e-05, + "step": 8327 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1402, + "image_masks_loss": 0.0045, + "llm_learning_rate": 2.7700201785889254e-07, + "loss": 0.6893, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 2.770020178588925e-05, + "step": 8328 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1402, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.7662257379567624e-07, + "loss": 0.7148, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 2.7662257379567624e-05, + "step": 8329 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1402, + "image_masks_loss": 0.0104, + "llm_learning_rate": 2.7624337457573534e-07, + "loss": 0.6459, + "output_text_mask_loss": 0.018, + "predictor_learning_rate": 2.7624337457573534e-05, + "step": 8330 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1401, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.75864420240827e-07, + "loss": 0.7471, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 2.7586442024082697e-05, + "step": 8331 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1401, + "image_masks_loss": 0.0132, + "llm_learning_rate": 2.7548571083268303e-07, + "loss": 0.7775, + "output_text_mask_loss": 0.0664, + "predictor_learning_rate": 2.7548571083268302e-05, + "step": 8332 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1401, + "image_masks_loss": 0.0207, + "llm_learning_rate": 2.751072463930074e-07, + "loss": 0.731, + "output_text_mask_loss": 0.0538, + "predictor_learning_rate": 2.751072463930074e-05, + "step": 8333 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.14, + "image_masks_loss": 0.0126, + "llm_learning_rate": 2.747290269634775e-07, + "loss": 0.7523, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 2.7472902696347752e-05, + "step": 8334 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.14, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.7435105258574374e-07, + "loss": 0.7157, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.7435105258574373e-05, + "step": 8335 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.14, + "image_masks_loss": 0.0131, + "llm_learning_rate": 2.739733233014291e-07, + "loss": 0.7877, + "output_text_mask_loss": 0.0446, + "predictor_learning_rate": 2.739733233014291e-05, + "step": 8336 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1399, + "image_masks_loss": 0.0192, + "llm_learning_rate": 2.7359583915213067e-07, + "loss": 0.7247, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.7359583915213067e-05, + "step": 8337 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1399, + "image_masks_loss": 0.0086, + "llm_learning_rate": 2.7321860017941795e-07, + "loss": 0.594, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.732186001794179e-05, + "step": 8338 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1399, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.7284160642483266e-07, + "loss": 0.7385, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 2.7284160642483262e-05, + "step": 8339 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1398, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.724648579298902e-07, + "loss": 0.7395, + "output_text_mask_loss": 0.0767, + "predictor_learning_rate": 2.7246485792989023e-05, + "step": 8340 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1398, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.7208835473607974e-07, + "loss": 0.6879, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 2.7208835473607975e-05, + "step": 8341 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1398, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.717120968848619e-07, + "loss": 0.7226, + "output_text_mask_loss": 0.0212, + "predictor_learning_rate": 2.7171209688486187e-05, + "step": 8342 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1397, + "image_masks_loss": 0.0119, + "llm_learning_rate": 2.713360844176713e-07, + "loss": 0.7701, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.7133608441767126e-05, + "step": 8343 + }, + { + "epoch": 0.85, + "gumbel_tau": 0.1397, + "image_masks_loss": 0.0177, + "llm_learning_rate": 2.709603173759151e-07, + "loss": 0.7706, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.7096031737591507e-05, + "step": 8344 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1397, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.7058479580097384e-07, + "loss": 0.7657, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 2.7058479580097383e-05, + "step": 8345 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1396, + "image_masks_loss": 0.0203, + "llm_learning_rate": 2.7020951973420084e-07, + "loss": 0.6748, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 2.7020951973420082e-05, + "step": 8346 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1396, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.698344892169222e-07, + "loss": 0.7752, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 2.6983448921692218e-05, + "step": 8347 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1396, + "image_masks_loss": 0.0252, + "llm_learning_rate": 2.6945970429043676e-07, + "loss": 0.7051, + "output_text_mask_loss": 0.007, + "predictor_learning_rate": 2.6945970429043676e-05, + "step": 8348 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1395, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.690851649960169e-07, + "loss": 0.6804, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.6908516499601688e-05, + "step": 8349 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1395, + "image_masks_loss": 0.0099, + "llm_learning_rate": 2.687108713749081e-07, + "loss": 0.738, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 2.6871087137490814e-05, + "step": 8350 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1395, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.683368234683281e-07, + "loss": 0.715, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 2.683368234683281e-05, + "step": 8351 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1394, + "image_masks_loss": 0.0223, + "llm_learning_rate": 2.6796302131746834e-07, + "loss": 0.8032, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 2.679630213174683e-05, + "step": 8352 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1394, + "image_masks_loss": 0.0119, + "llm_learning_rate": 2.675894649634919e-07, + "loss": 0.8079, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 2.675894649634919e-05, + "step": 8353 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1394, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.6721615444753627e-07, + "loss": 0.7982, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 2.672161544475363e-05, + "step": 8354 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1393, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.6684308981071107e-07, + "loss": 0.7414, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 2.6684308981071103e-05, + "step": 8355 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1393, + "image_masks_loss": 0.0174, + "llm_learning_rate": 2.664702710940983e-07, + "loss": 0.714, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 2.6647027109409834e-05, + "step": 8356 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1393, + "image_masks_loss": 0.0165, + "llm_learning_rate": 2.6609769833875505e-07, + "loss": 0.7769, + "output_text_mask_loss": 0.0121, + "predictor_learning_rate": 2.6609769833875502e-05, + "step": 8357 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1392, + "image_masks_loss": 0.0326, + "llm_learning_rate": 2.6572537158570886e-07, + "loss": 0.8198, + "output_text_mask_loss": 0.053, + "predictor_learning_rate": 2.6572537158570888e-05, + "step": 8358 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1392, + "image_masks_loss": 0.0195, + "llm_learning_rate": 2.653532908759615e-07, + "loss": 0.8036, + "output_text_mask_loss": 0.0511, + "predictor_learning_rate": 2.653532908759615e-05, + "step": 8359 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1392, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.649814562504871e-07, + "loss": 0.7288, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 2.6498145625048708e-05, + "step": 8360 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1391, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.6460986775023315e-07, + "loss": 0.791, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 2.6460986775023317e-05, + "step": 8361 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1391, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.642385254161198e-07, + "loss": 0.7272, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 2.6423852541611977e-05, + "step": 8362 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1391, + "image_masks_loss": 0.0015, + "llm_learning_rate": 2.6386742928903976e-07, + "loss": 0.6484, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 2.638674292890397e-05, + "step": 8363 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.139, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.634965794098593e-07, + "loss": 0.7641, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.634965794098593e-05, + "step": 8364 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.139, + "image_masks_loss": 0.0032, + "llm_learning_rate": 2.6312597581941714e-07, + "loss": 0.73, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.631259758194171e-05, + "step": 8365 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.139, + "image_masks_loss": 0.0179, + "llm_learning_rate": 2.6275561855852505e-07, + "loss": 0.7442, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 2.6275561855852505e-05, + "step": 8366 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1389, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.623855076679674e-07, + "loss": 0.7484, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 2.6238550766796738e-05, + "step": 8367 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1389, + "image_masks_loss": 0.0116, + "llm_learning_rate": 2.6201564318850124e-07, + "loss": 0.6772, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 2.6201564318850125e-05, + "step": 8368 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1389, + "image_masks_loss": 0.0161, + "llm_learning_rate": 2.6164602516085796e-07, + "loss": 0.6779, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 2.6164602516085796e-05, + "step": 8369 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1388, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.6127665362573993e-07, + "loss": 0.6802, + "output_text_mask_loss": 0.0044, + "predictor_learning_rate": 2.6127665362573993e-05, + "step": 8370 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1388, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.6090752862382357e-07, + "loss": 0.7044, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.6090752862382356e-05, + "step": 8371 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1388, + "image_masks_loss": 0.02, + "llm_learning_rate": 2.605386501957577e-07, + "loss": 0.7284, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.6053865019575767e-05, + "step": 8372 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1387, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.601700183821634e-07, + "loss": 0.6729, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.6017001838216342e-05, + "step": 8373 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1387, + "image_masks_loss": 0.0194, + "llm_learning_rate": 2.598016332236361e-07, + "loss": 0.7331, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 2.598016332236361e-05, + "step": 8374 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1387, + "image_masks_loss": 0.0235, + "llm_learning_rate": 2.594334947607427e-07, + "loss": 0.7137, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 2.594334947607427e-05, + "step": 8375 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1386, + "image_masks_loss": 0.0212, + "llm_learning_rate": 2.59065603034023e-07, + "loss": 0.6997, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 2.59065603034023e-05, + "step": 8376 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1386, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.58697958083991e-07, + "loss": 0.6763, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 2.58697958083991e-05, + "step": 8377 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1386, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.583305599511321e-07, + "loss": 0.7176, + "output_text_mask_loss": 0.0028, + "predictor_learning_rate": 2.5833055995113207e-05, + "step": 8378 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1385, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.579634086759053e-07, + "loss": 0.833, + "output_text_mask_loss": 0.0378, + "predictor_learning_rate": 2.579634086759053e-05, + "step": 8379 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1385, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.5759650429874217e-07, + "loss": 0.7434, + "output_text_mask_loss": 0.0328, + "predictor_learning_rate": 2.5759650429874215e-05, + "step": 8380 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1385, + "image_masks_loss": 0.0131, + "llm_learning_rate": 2.572298468600459e-07, + "loss": 0.7955, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.572298468600459e-05, + "step": 8381 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1384, + "image_masks_loss": 0.0216, + "llm_learning_rate": 2.5686343640019486e-07, + "loss": 0.8206, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 2.5686343640019483e-05, + "step": 8382 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1384, + "image_masks_loss": 0.0118, + "llm_learning_rate": 2.5649727295953857e-07, + "loss": 0.6935, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.5649727295953857e-05, + "step": 8383 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1384, + "image_masks_loss": 0.0248, + "llm_learning_rate": 2.561313565783999e-07, + "loss": 0.72, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.5613135657839986e-05, + "step": 8384 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1384, + "image_masks_loss": 0.003, + "llm_learning_rate": 2.557656872970743e-07, + "loss": 0.563, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.5576568729707426e-05, + "step": 8385 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1383, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.554002651558299e-07, + "loss": 0.7162, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 2.5540026515582986e-05, + "step": 8386 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1383, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.550350901949078e-07, + "loss": 0.6927, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.550350901949078e-05, + "step": 8387 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1383, + "image_masks_loss": 0.0125, + "llm_learning_rate": 2.5467016245452195e-07, + "loss": 0.723, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 2.5467016245452197e-05, + "step": 8388 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1382, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.5430548197485945e-07, + "loss": 0.7384, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 2.5430548197485943e-05, + "step": 8389 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1382, + "image_masks_loss": 0.0158, + "llm_learning_rate": 2.5394104879607915e-07, + "loss": 0.7861, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 2.5394104879607916e-05, + "step": 8390 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1382, + "image_masks_loss": 0.014, + "llm_learning_rate": 2.535768629583135e-07, + "loss": 0.7187, + "output_text_mask_loss": 0.0576, + "predictor_learning_rate": 2.535768629583135e-05, + "step": 8391 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1381, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.5321292450166765e-07, + "loss": 0.7027, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.5321292450166762e-05, + "step": 8392 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1381, + "image_masks_loss": 0.0056, + "llm_learning_rate": 2.5284923346621886e-07, + "loss": 0.7962, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 2.5284923346621886e-05, + "step": 8393 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1381, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.52485789892018e-07, + "loss": 0.7086, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 2.5248578989201797e-05, + "step": 8394 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.138, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.521225938190874e-07, + "loss": 0.7808, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.5212259381908742e-05, + "step": 8395 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.138, + "image_masks_loss": 0.0063, + "llm_learning_rate": 2.5175964528742433e-07, + "loss": 0.7812, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.5175964528742434e-05, + "step": 8396 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.138, + "image_masks_loss": 0.0056, + "llm_learning_rate": 2.513969443369968e-07, + "loss": 0.816, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 2.513969443369968e-05, + "step": 8397 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1379, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.510344910077464e-07, + "loss": 0.7249, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.510344910077464e-05, + "step": 8398 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1379, + "image_masks_loss": 0.0109, + "llm_learning_rate": 2.5067228533958767e-07, + "loss": 0.7197, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 2.5067228533958765e-05, + "step": 8399 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1379, + "image_masks_loss": 0.0091, + "llm_learning_rate": 2.5031032737240666e-07, + "loss": 0.6895, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.5031032737240668e-05, + "step": 8400 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1378, + "image_masks_loss": 0.0037, + "llm_learning_rate": 2.49948617146063e-07, + "loss": 0.7456, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.49948617146063e-05, + "step": 8401 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1378, + "image_masks_loss": 0.0177, + "llm_learning_rate": 2.4958715470038973e-07, + "loss": 0.8032, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 2.4958715470038972e-05, + "step": 8402 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1378, + "image_masks_loss": 0.0053, + "llm_learning_rate": 2.4922594007519144e-07, + "loss": 0.6288, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 2.4922594007519147e-05, + "step": 8403 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1377, + "image_masks_loss": 0.0039, + "llm_learning_rate": 2.4886497331024636e-07, + "loss": 0.7678, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 2.488649733102463e-05, + "step": 8404 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1377, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.4850425444530423e-07, + "loss": 0.6992, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 2.4850425444530423e-05, + "step": 8405 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1377, + "image_masks_loss": 0.0141, + "llm_learning_rate": 2.4814378352008893e-07, + "loss": 0.7439, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 2.4814378352008892e-05, + "step": 8406 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1376, + "image_masks_loss": 0.0105, + "llm_learning_rate": 2.477835605742959e-07, + "loss": 0.6887, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 2.4778356057429592e-05, + "step": 8407 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1376, + "image_masks_loss": 0.0044, + "llm_learning_rate": 2.4742358564759344e-07, + "loss": 0.7087, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.4742358564759344e-05, + "step": 8408 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1376, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.4706385877962343e-07, + "loss": 0.7327, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 2.470638587796234e-05, + "step": 8409 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1375, + "image_masks_loss": 0.0015, + "llm_learning_rate": 2.467043800099994e-07, + "loss": 0.676, + "output_text_mask_loss": 0.0303, + "predictor_learning_rate": 2.467043800099994e-05, + "step": 8410 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1375, + "image_masks_loss": 0.0393, + "llm_learning_rate": 2.463451493783084e-07, + "loss": 0.7511, + "output_text_mask_loss": 0.0324, + "predictor_learning_rate": 2.4634514937830838e-05, + "step": 8411 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1375, + "image_masks_loss": 0.0089, + "llm_learning_rate": 2.4598616692410925e-07, + "loss": 0.7621, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 2.4598616692410924e-05, + "step": 8412 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1374, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.45627432686934e-07, + "loss": 0.7665, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 2.4562743268693398e-05, + "step": 8413 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1374, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.4526894670628716e-07, + "loss": 0.7287, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 2.4526894670628715e-05, + "step": 8414 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1374, + "image_masks_loss": 0.0156, + "llm_learning_rate": 2.4491070902164573e-07, + "loss": 0.7982, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 2.4491070902164568e-05, + "step": 8415 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1373, + "image_masks_loss": 0.018, + "llm_learning_rate": 2.4455271967246064e-07, + "loss": 0.6503, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.4455271967246062e-05, + "step": 8416 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1373, + "image_masks_loss": 0.0057, + "llm_learning_rate": 2.441949786981543e-07, + "loss": 0.6494, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.4419497869815427e-05, + "step": 8417 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1373, + "image_masks_loss": 0.0102, + "llm_learning_rate": 2.43837486138121e-07, + "loss": 0.7935, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.4383748613812095e-05, + "step": 8418 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1372, + "image_masks_loss": 0.0084, + "llm_learning_rate": 2.4348024203172894e-07, + "loss": 0.7434, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.434802420317289e-05, + "step": 8419 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1372, + "image_masks_loss": 0.0121, + "llm_learning_rate": 2.4312324641831865e-07, + "loss": 0.8018, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 2.4312324641831863e-05, + "step": 8420 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1372, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.4276649933720366e-07, + "loss": 0.6599, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.4276649933720364e-05, + "step": 8421 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1371, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.4241000082766973e-07, + "loss": 0.7002, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 2.4241000082766975e-05, + "step": 8422 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1371, + "image_masks_loss": 0.0234, + "llm_learning_rate": 2.4205375092897526e-07, + "loss": 0.7396, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.4205375092897524e-05, + "step": 8423 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1371, + "image_masks_loss": 0.0037, + "llm_learning_rate": 2.416977496803508e-07, + "loss": 0.6562, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 2.4169774968035076e-05, + "step": 8424 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1371, + "image_masks_loss": 0.0112, + "llm_learning_rate": 2.413419971210004e-07, + "loss": 0.7322, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 2.413419971210004e-05, + "step": 8425 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.137, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.4098649329010045e-07, + "loss": 0.6166, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 2.409864932901004e-05, + "step": 8426 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.137, + "image_masks_loss": 0.0175, + "llm_learning_rate": 2.406312382267989e-07, + "loss": 0.6737, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.406312382267989e-05, + "step": 8427 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.137, + "image_masks_loss": 0.0244, + "llm_learning_rate": 2.4027623197021845e-07, + "loss": 0.7106, + "output_text_mask_loss": 0.0656, + "predictor_learning_rate": 2.4027623197021842e-05, + "step": 8428 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1369, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.3992147455945276e-07, + "loss": 0.6943, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.3992147455945274e-05, + "step": 8429 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1369, + "image_masks_loss": 0.0131, + "llm_learning_rate": 2.3956696603356837e-07, + "loss": 0.6505, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 2.3956696603356838e-05, + "step": 8430 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1369, + "image_masks_loss": 0.0037, + "llm_learning_rate": 2.3921270643160464e-07, + "loss": 0.7332, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.3921270643160464e-05, + "step": 8431 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1368, + "image_masks_loss": 0.0047, + "llm_learning_rate": 2.3885869579257316e-07, + "loss": 0.7128, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 2.3885869579257314e-05, + "step": 8432 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1368, + "image_masks_loss": 0.0086, + "llm_learning_rate": 2.385049341554588e-07, + "loss": 0.7046, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.3850493415545876e-05, + "step": 8433 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1368, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.381514215592179e-07, + "loss": 0.764, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 2.3815142155921788e-05, + "step": 8434 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1367, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.37798158042781e-07, + "loss": 0.7581, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 2.37798158042781e-05, + "step": 8435 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1367, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.3744514364505033e-07, + "loss": 0.7751, + "output_text_mask_loss": 0.0904, + "predictor_learning_rate": 2.374451436450503e-05, + "step": 8436 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1367, + "image_masks_loss": 0.0225, + "llm_learning_rate": 2.3709237840489952e-07, + "loss": 0.7344, + "output_text_mask_loss": 0.007, + "predictor_learning_rate": 2.370923784048995e-05, + "step": 8437 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1366, + "image_masks_loss": 0.009, + "llm_learning_rate": 2.3673986236117635e-07, + "loss": 0.6685, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 2.3673986236117633e-05, + "step": 8438 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1366, + "image_masks_loss": 0.0184, + "llm_learning_rate": 2.363875955527009e-07, + "loss": 0.785, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 2.363875955527009e-05, + "step": 8439 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1366, + "image_masks_loss": 0.0183, + "llm_learning_rate": 2.3603557801826493e-07, + "loss": 0.6787, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.3603557801826492e-05, + "step": 8440 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1365, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.3568380979663436e-07, + "loss": 0.7392, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 2.3568380979663434e-05, + "step": 8441 + }, + { + "epoch": 0.86, + "gumbel_tau": 0.1365, + "image_masks_loss": 0.0042, + "llm_learning_rate": 2.3533229092654626e-07, + "loss": 0.626, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 2.3533229092654623e-05, + "step": 8442 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1365, + "image_masks_loss": 0.0111, + "llm_learning_rate": 2.349810214467105e-07, + "loss": 0.6442, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 2.3498102144671045e-05, + "step": 8443 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1364, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.3463000139580972e-07, + "loss": 0.7067, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 2.346300013958097e-05, + "step": 8444 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1364, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.3427923081249943e-07, + "loss": 0.7326, + "output_text_mask_loss": 0.1259, + "predictor_learning_rate": 2.342792308124994e-05, + "step": 8445 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1364, + "image_masks_loss": 0.0075, + "llm_learning_rate": 2.3392870973540648e-07, + "loss": 0.796, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.3392870973540647e-05, + "step": 8446 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1363, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.3357843820313142e-07, + "loss": 0.6544, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 2.335784382031314e-05, + "step": 8447 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1363, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.3322841625424724e-07, + "loss": 0.6141, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.3322841625424722e-05, + "step": 8448 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1363, + "image_masks_loss": 0.0174, + "llm_learning_rate": 2.3287864392729899e-07, + "loss": 0.7305, + "output_text_mask_loss": 0.0244, + "predictor_learning_rate": 2.3287864392729895e-05, + "step": 8449 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1362, + "image_masks_loss": 0.0165, + "llm_learning_rate": 2.325291212608044e-07, + "loss": 0.6984, + "output_text_mask_loss": 0.0736, + "predictor_learning_rate": 2.325291212608044e-05, + "step": 8450 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1362, + "image_masks_loss": 0.0065, + "llm_learning_rate": 2.3217984829325356e-07, + "loss": 0.7805, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 2.3217984829325357e-05, + "step": 8451 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1362, + "image_masks_loss": 0.0026, + "llm_learning_rate": 2.3183082506310904e-07, + "loss": 0.727, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 2.3183082506310904e-05, + "step": 8452 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1361, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.314820516088062e-07, + "loss": 0.7445, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 2.314820516088062e-05, + "step": 8453 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1361, + "image_masks_loss": 0.0207, + "llm_learning_rate": 2.3113352796875295e-07, + "loss": 0.7989, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 2.311335279687529e-05, + "step": 8454 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1361, + "image_masks_loss": 0.0146, + "llm_learning_rate": 2.307852541813291e-07, + "loss": 0.7883, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 2.3078525418132906e-05, + "step": 8455 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1361, + "image_masks_loss": 0.0022, + "llm_learning_rate": 2.3043723028488757e-07, + "loss": 0.7999, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.3043723028488754e-05, + "step": 8456 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.136, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.300894563177536e-07, + "loss": 0.7556, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 2.300894563177536e-05, + "step": 8457 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.136, + "image_masks_loss": 0.0171, + "llm_learning_rate": 2.2974193231822455e-07, + "loss": 0.8229, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.2974193231822454e-05, + "step": 8458 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.136, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.2939465832457043e-07, + "loss": 0.7594, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 2.2939465832457042e-05, + "step": 8459 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1359, + "image_masks_loss": 0.0053, + "llm_learning_rate": 2.2904763437503447e-07, + "loss": 0.6743, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.2904763437503444e-05, + "step": 8460 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1359, + "image_masks_loss": 0.0045, + "llm_learning_rate": 2.287008605078314e-07, + "loss": 0.7374, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 2.287008605078314e-05, + "step": 8461 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1359, + "image_masks_loss": 0.0144, + "llm_learning_rate": 2.2835433676114877e-07, + "loss": 0.7158, + "output_text_mask_loss": 0.038, + "predictor_learning_rate": 2.2835433676114873e-05, + "step": 8462 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1358, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.2800806317314627e-07, + "loss": 0.6989, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 2.2800806317314626e-05, + "step": 8463 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1358, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.2766203978195678e-07, + "loss": 0.7444, + "output_text_mask_loss": 0.132, + "predictor_learning_rate": 2.2766203978195678e-05, + "step": 8464 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1358, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.273162666256848e-07, + "loss": 0.7578, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 2.273162666256848e-05, + "step": 8465 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1357, + "image_masks_loss": 0.009, + "llm_learning_rate": 2.2697074374240735e-07, + "loss": 0.7325, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 2.2697074374240734e-05, + "step": 8466 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1357, + "image_masks_loss": 0.0211, + "llm_learning_rate": 2.2662547117017487e-07, + "loss": 0.7313, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 2.2662547117017484e-05, + "step": 8467 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1357, + "image_masks_loss": 0.0088, + "llm_learning_rate": 2.2628044894700912e-07, + "loss": 0.6161, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.262804489470091e-05, + "step": 8468 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1356, + "image_masks_loss": 0.0169, + "llm_learning_rate": 2.2593567711090532e-07, + "loss": 0.6225, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.259356771109053e-05, + "step": 8469 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1356, + "image_masks_loss": 0.0164, + "llm_learning_rate": 2.255911556998297e-07, + "loss": 0.7587, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.2559115569982968e-05, + "step": 8470 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1356, + "image_masks_loss": 0.0024, + "llm_learning_rate": 2.2524688475172195e-07, + "loss": 0.7389, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 2.2524688475172194e-05, + "step": 8471 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1355, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.2490286430449426e-07, + "loss": 0.718, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 2.2490286430449424e-05, + "step": 8472 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1355, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.2455909439603075e-07, + "loss": 0.7343, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.2455909439603074e-05, + "step": 8473 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1355, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.2421557506418807e-07, + "loss": 0.8059, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.2421557506418805e-05, + "step": 8474 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1354, + "image_masks_loss": 0.0214, + "llm_learning_rate": 2.238723063467954e-07, + "loss": 0.7368, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 2.238723063467954e-05, + "step": 8475 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1354, + "image_masks_loss": 0.0221, + "llm_learning_rate": 2.235292882816542e-07, + "loss": 0.7366, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 2.235292882816542e-05, + "step": 8476 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1354, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.2318652090653842e-07, + "loss": 0.6998, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 2.231865209065384e-05, + "step": 8477 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1353, + "image_masks_loss": 0.005, + "llm_learning_rate": 2.2284400425919422e-07, + "loss": 0.6849, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.228440042591942e-05, + "step": 8478 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1353, + "image_masks_loss": 0.0023, + "llm_learning_rate": 2.2250173837734036e-07, + "loss": 0.7176, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 2.2250173837734032e-05, + "step": 8479 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1353, + "image_masks_loss": 0.013, + "llm_learning_rate": 2.2215972329866803e-07, + "loss": 0.614, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 2.2215972329866802e-05, + "step": 8480 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1353, + "image_masks_loss": 0.0156, + "llm_learning_rate": 2.2181795906084102e-07, + "loss": 0.6191, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 2.21817959060841e-05, + "step": 8481 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1352, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.2147644570149452e-07, + "loss": 0.7104, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.214764457014945e-05, + "step": 8482 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1352, + "image_masks_loss": 0.0018, + "llm_learning_rate": 2.2113518325823703e-07, + "loss": 0.7024, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.2113518325823702e-05, + "step": 8483 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1352, + "image_masks_loss": 0.0089, + "llm_learning_rate": 2.207941717686493e-07, + "loss": 0.6967, + "output_text_mask_loss": 0.025, + "predictor_learning_rate": 2.207941717686493e-05, + "step": 8484 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1351, + "image_masks_loss": 0.0044, + "llm_learning_rate": 2.2045341127028414e-07, + "loss": 0.7536, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 2.2045341127028412e-05, + "step": 8485 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1351, + "image_masks_loss": 0.0204, + "llm_learning_rate": 2.201129018006662e-07, + "loss": 0.7796, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 2.2011290180066617e-05, + "step": 8486 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1351, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.1977264339729437e-07, + "loss": 0.7938, + "output_text_mask_loss": 0.0357, + "predictor_learning_rate": 2.1977264339729437e-05, + "step": 8487 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.135, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.1943263609763787e-07, + "loss": 0.722, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 2.1943263609763785e-05, + "step": 8488 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.135, + "image_masks_loss": 0.015, + "llm_learning_rate": 2.1909287993913952e-07, + "loss": 0.7582, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 2.190928799391395e-05, + "step": 8489 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.135, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.1875337495921355e-07, + "loss": 0.716, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.1875337495921353e-05, + "step": 8490 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1349, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.184141211952473e-07, + "loss": 0.7475, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.1841412119524728e-05, + "step": 8491 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1349, + "image_masks_loss": 0.008, + "llm_learning_rate": 2.180751186846e-07, + "loss": 0.7322, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 2.1807511868460002e-05, + "step": 8492 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1349, + "image_masks_loss": 0.0052, + "llm_learning_rate": 2.1773636746460326e-07, + "loss": 0.7481, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 2.1773636746460324e-05, + "step": 8493 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1348, + "image_masks_loss": 0.0154, + "llm_learning_rate": 2.173978675725616e-07, + "loss": 0.6935, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.173978675725616e-05, + "step": 8494 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1348, + "image_masks_loss": 0.0038, + "llm_learning_rate": 2.170596190457508e-07, + "loss": 0.6724, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.170596190457508e-05, + "step": 8495 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1348, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.1672162192141994e-07, + "loss": 0.7865, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 2.1672162192141992e-05, + "step": 8496 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1347, + "image_masks_loss": 0.0159, + "llm_learning_rate": 2.163838762367898e-07, + "loss": 0.766, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 2.163838762367898e-05, + "step": 8497 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1347, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.1604638202905343e-07, + "loss": 0.7433, + "output_text_mask_loss": 0.0595, + "predictor_learning_rate": 2.160463820290534e-05, + "step": 8498 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1347, + "image_masks_loss": 0.0123, + "llm_learning_rate": 2.1570913933537718e-07, + "loss": 0.7706, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.1570913933537718e-05, + "step": 8499 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1346, + "image_masks_loss": 0.0146, + "llm_learning_rate": 2.1537214819289858e-07, + "loss": 0.7394, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 2.1537214819289856e-05, + "step": 8500 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1346, + "image_masks_loss": 0.0049, + "llm_learning_rate": 2.1503540863872768e-07, + "loss": 0.8229, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 2.1503540863872766e-05, + "step": 8501 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1346, + "image_masks_loss": 0.0039, + "llm_learning_rate": 2.146989207099473e-07, + "loss": 0.5965, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 2.1469892070994728e-05, + "step": 8502 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1346, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.1436268444361198e-07, + "loss": 0.7453, + "output_text_mask_loss": 0.0273, + "predictor_learning_rate": 2.1436268444361195e-05, + "step": 8503 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1345, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.1402669987674902e-07, + "loss": 0.6974, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.14026699876749e-05, + "step": 8504 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1345, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.1369096704635718e-07, + "loss": 0.7681, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.1369096704635715e-05, + "step": 8505 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1345, + "image_masks_loss": 0.0079, + "llm_learning_rate": 2.1335548598940908e-07, + "loss": 0.7124, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 2.1335548598940908e-05, + "step": 8506 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1344, + "image_masks_loss": 0.0099, + "llm_learning_rate": 2.13020256742848e-07, + "loss": 0.7256, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.1302025674284798e-05, + "step": 8507 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1344, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.1268527934359045e-07, + "loss": 0.7637, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 2.1268527934359044e-05, + "step": 8508 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1344, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.1235055382852476e-07, + "loss": 0.6907, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.1235055382852475e-05, + "step": 8509 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1343, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.1201608023451199e-07, + "loss": 0.799, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 2.1201608023451197e-05, + "step": 8510 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1343, + "image_masks_loss": 0.0035, + "llm_learning_rate": 2.1168185859838374e-07, + "loss": 0.6836, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.1168185859838375e-05, + "step": 8511 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1343, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.1134788895694702e-07, + "loss": 0.6554, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 2.1134788895694702e-05, + "step": 8512 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1342, + "image_masks_loss": 0.0054, + "llm_learning_rate": 2.110141713469782e-07, + "loss": 0.8546, + "output_text_mask_loss": 0.0706, + "predictor_learning_rate": 2.110141713469782e-05, + "step": 8513 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1342, + "image_masks_loss": 0.0097, + "llm_learning_rate": 2.1068070580522766e-07, + "loss": 0.7418, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 2.1068070580522762e-05, + "step": 8514 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1342, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.1034749236841678e-07, + "loss": 0.7881, + "output_text_mask_loss": 0.0165, + "predictor_learning_rate": 2.103474923684168e-05, + "step": 8515 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1341, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.1001453107324017e-07, + "loss": 0.6378, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 2.1001453107324015e-05, + "step": 8516 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1341, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.0968182195636428e-07, + "loss": 0.6528, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.0968182195636426e-05, + "step": 8517 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1341, + "image_masks_loss": 0.0032, + "llm_learning_rate": 2.0934936505442703e-07, + "loss": 0.7801, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 2.0934936505442703e-05, + "step": 8518 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.134, + "image_masks_loss": 0.0155, + "llm_learning_rate": 2.090171604040403e-07, + "loss": 0.7642, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 2.0901716040404028e-05, + "step": 8519 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.134, + "image_masks_loss": 0.0136, + "llm_learning_rate": 2.08685208041787e-07, + "loss": 0.6122, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 2.0868520804178698e-05, + "step": 8520 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.134, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.0835350800422204e-07, + "loss": 0.6901, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 2.0835350800422205e-05, + "step": 8521 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.134, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.0802206032787347e-07, + "loss": 0.7246, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 2.0802206032787346e-05, + "step": 8522 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1339, + "image_masks_loss": 0.0138, + "llm_learning_rate": 2.076908650492407e-07, + "loss": 0.7182, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.0769086504924067e-05, + "step": 8523 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1339, + "image_masks_loss": 0.0228, + "llm_learning_rate": 2.073599222047959e-07, + "loss": 0.7618, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 2.0735992220479587e-05, + "step": 8524 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1339, + "image_masks_loss": 0.0044, + "llm_learning_rate": 2.0702923183098245e-07, + "loss": 0.7478, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 2.0702923183098244e-05, + "step": 8525 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1338, + "image_masks_loss": 0.0037, + "llm_learning_rate": 2.0669879396421787e-07, + "loss": 0.6236, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.0669879396421786e-05, + "step": 8526 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1338, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.0636860864089059e-07, + "loss": 0.7572, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.0636860864089056e-05, + "step": 8527 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1338, + "image_masks_loss": 0.0031, + "llm_learning_rate": 2.0603867589736094e-07, + "loss": 0.6509, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.0603867589736092e-05, + "step": 8528 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1337, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.0570899576996178e-07, + "loss": 0.583, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.0570899576996178e-05, + "step": 8529 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1337, + "image_masks_loss": 0.0033, + "llm_learning_rate": 2.0537956829499772e-07, + "loss": 0.7114, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 2.053795682949977e-05, + "step": 8530 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1337, + "image_masks_loss": 0.005, + "llm_learning_rate": 2.0505039350874723e-07, + "loss": 0.6625, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.0505039350874722e-05, + "step": 8531 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1336, + "image_masks_loss": 0.004, + "llm_learning_rate": 2.0472147144745906e-07, + "loss": 0.6573, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 2.0472147144745905e-05, + "step": 8532 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1336, + "image_masks_loss": 0.0096, + "llm_learning_rate": 2.0439280214735514e-07, + "loss": 0.8661, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 2.043928021473551e-05, + "step": 8533 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1336, + "image_masks_loss": 0.0162, + "llm_learning_rate": 2.0406438564462922e-07, + "loss": 0.7139, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.040643856446292e-05, + "step": 8534 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1335, + "image_masks_loss": 0.0168, + "llm_learning_rate": 2.0373622197544685e-07, + "loss": 0.7593, + "output_text_mask_loss": 0.0057, + "predictor_learning_rate": 2.0373622197544683e-05, + "step": 8535 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1335, + "image_masks_loss": 0.0154, + "llm_learning_rate": 2.0340831117594662e-07, + "loss": 0.7592, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 2.034083111759466e-05, + "step": 8536 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1335, + "image_masks_loss": 0.0273, + "llm_learning_rate": 2.0308065328223824e-07, + "loss": 0.7358, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 2.0308065328223824e-05, + "step": 8537 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1334, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.0275324833040483e-07, + "loss": 0.7774, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.0275324833040483e-05, + "step": 8538 + }, + { + "epoch": 0.87, + "gumbel_tau": 0.1334, + "image_masks_loss": 0.0095, + "llm_learning_rate": 2.0242609635650057e-07, + "loss": 0.7225, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 2.0242609635650054e-05, + "step": 8539 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1334, + "image_masks_loss": 0.0041, + "llm_learning_rate": 2.020991973965522e-07, + "loss": 0.7968, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.020991973965522e-05, + "step": 8540 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1334, + "image_masks_loss": 0.0127, + "llm_learning_rate": 2.0177255148655873e-07, + "loss": 0.6866, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 2.0177255148655872e-05, + "step": 8541 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1333, + "image_masks_loss": 0.0106, + "llm_learning_rate": 2.0144615866249108e-07, + "loss": 0.7187, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 2.0144615866249104e-05, + "step": 8542 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1333, + "image_masks_loss": 0.015, + "llm_learning_rate": 2.0112001896029216e-07, + "loss": 0.679, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 2.0112001896029215e-05, + "step": 8543 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1333, + "image_masks_loss": 0.0153, + "llm_learning_rate": 2.0079413241587687e-07, + "loss": 0.7067, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 2.0079413241587686e-05, + "step": 8544 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1332, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.0046849906513316e-07, + "loss": 0.6873, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 2.0046849906513315e-05, + "step": 8545 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1332, + "image_masks_loss": 0.0144, + "llm_learning_rate": 2.0014311894392068e-07, + "loss": 0.7429, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.0014311894392066e-05, + "step": 8546 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1332, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.9981799208807107e-07, + "loss": 0.7092, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.9981799208807106e-05, + "step": 8547 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1331, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.9949311853338705e-07, + "loss": 0.744, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 1.9949311853338703e-05, + "step": 8548 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1331, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.9916849831564527e-07, + "loss": 0.7438, + "output_text_mask_loss": 0.0125, + "predictor_learning_rate": 1.9916849831564526e-05, + "step": 8549 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1331, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.9884413147059271e-07, + "loss": 0.6834, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 1.9884413147059273e-05, + "step": 8550 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.133, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.9852001803395026e-07, + "loss": 0.7851, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 1.9852001803395025e-05, + "step": 8551 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.133, + "image_masks_loss": 0.005, + "llm_learning_rate": 1.9819615804141017e-07, + "loss": 0.7318, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.981961580414102e-05, + "step": 8552 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.133, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.9787255152863588e-07, + "loss": 0.6814, + "output_text_mask_loss": 0.0299, + "predictor_learning_rate": 1.978725515286359e-05, + "step": 8553 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1329, + "image_masks_loss": 0.0182, + "llm_learning_rate": 1.9754919853126414e-07, + "loss": 0.7319, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 1.9754919853126412e-05, + "step": 8554 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1329, + "image_masks_loss": 0.0182, + "llm_learning_rate": 1.9722609908490342e-07, + "loss": 0.7643, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.9722609908490342e-05, + "step": 8555 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1329, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.969032532251336e-07, + "loss": 0.7494, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 1.969032532251336e-05, + "step": 8556 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1328, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.9658066098750705e-07, + "loss": 0.7298, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.9658066098750704e-05, + "step": 8557 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1328, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.9625832240754928e-07, + "loss": 0.8302, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 1.9625832240754928e-05, + "step": 8558 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1328, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.9593623752075657e-07, + "loss": 0.7927, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 1.9593623752075656e-05, + "step": 8559 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1328, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.9561440636259727e-07, + "loss": 0.7245, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 1.9561440636259727e-05, + "step": 8560 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1327, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.9529282896851244e-07, + "loss": 0.7726, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 1.952928289685124e-05, + "step": 8561 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1327, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.9497150537391484e-07, + "loss": 0.744, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.9497150537391485e-05, + "step": 8562 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1327, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.9465043561418928e-07, + "loss": 0.7201, + "output_text_mask_loss": 0.0224, + "predictor_learning_rate": 1.9465043561418928e-05, + "step": 8563 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1326, + "image_masks_loss": 0.0224, + "llm_learning_rate": 1.9432961972469216e-07, + "loss": 0.7388, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 1.9432961972469215e-05, + "step": 8564 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1326, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.9400905774075357e-07, + "loss": 0.6332, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.9400905774075356e-05, + "step": 8565 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1326, + "image_masks_loss": 0.0089, + "llm_learning_rate": 1.9368874969767443e-07, + "loss": 0.7918, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 1.936887496976744e-05, + "step": 8566 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1325, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.933686956307268e-07, + "loss": 0.7099, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 1.933686956307268e-05, + "step": 8567 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1325, + "image_masks_loss": 0.0232, + "llm_learning_rate": 1.930488955751561e-07, + "loss": 0.7268, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 1.930488955751561e-05, + "step": 8568 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1325, + "image_masks_loss": 0.0043, + "llm_learning_rate": 1.9272934956617916e-07, + "loss": 0.7639, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 1.9272934956617916e-05, + "step": 8569 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1324, + "image_masks_loss": 0.0114, + "llm_learning_rate": 1.9241005763898618e-07, + "loss": 0.745, + "output_text_mask_loss": 0.0212, + "predictor_learning_rate": 1.9241005763898616e-05, + "step": 8570 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1324, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.9209101982873734e-07, + "loss": 0.686, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 1.9209101982873733e-05, + "step": 8571 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1324, + "image_masks_loss": 0.005, + "llm_learning_rate": 1.9177223617056622e-07, + "loss": 0.7101, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.917722361705662e-05, + "step": 8572 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1323, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.9145370669957751e-07, + "loss": 0.713, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.914537066995775e-05, + "step": 8573 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1323, + "image_masks_loss": 0.0137, + "llm_learning_rate": 1.91135431450849e-07, + "loss": 0.8207, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 1.91135431450849e-05, + "step": 8574 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1323, + "image_masks_loss": 0.0298, + "llm_learning_rate": 1.908174104594296e-07, + "loss": 0.786, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 1.9081741045942958e-05, + "step": 8575 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1323, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.9049964376033986e-07, + "loss": 0.7594, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 1.9049964376033986e-05, + "step": 8576 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1322, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.9018213138857406e-07, + "loss": 0.7243, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.9018213138857403e-05, + "step": 8577 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1322, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.8986487337909698e-07, + "loss": 0.71, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 1.8986487337909696e-05, + "step": 8578 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1322, + "image_masks_loss": 0.0144, + "llm_learning_rate": 1.8954786976684542e-07, + "loss": 0.8143, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 1.8954786976684542e-05, + "step": 8579 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1321, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.8923112058672894e-07, + "loss": 0.7759, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 1.8923112058672893e-05, + "step": 8580 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1321, + "image_masks_loss": 0.018, + "llm_learning_rate": 1.8891462587362852e-07, + "loss": 0.7473, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 1.889146258736285e-05, + "step": 8581 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1321, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.8859838566239713e-07, + "loss": 0.7227, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.8859838566239712e-05, + "step": 8582 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.132, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.8828239998785968e-07, + "loss": 0.6544, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.8828239998785968e-05, + "step": 8583 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.132, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.8796666888481418e-07, + "loss": 0.736, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.8796666888481416e-05, + "step": 8584 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.132, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.8765119238802865e-07, + "loss": 0.7563, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 1.8765119238802863e-05, + "step": 8585 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1319, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.8733597053224446e-07, + "loss": 0.7376, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 1.8733597053224445e-05, + "step": 8586 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1319, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.8702100335217438e-07, + "loss": 0.6779, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 1.8702100335217438e-05, + "step": 8587 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1319, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.867062908825032e-07, + "loss": 0.7532, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 1.867062908825032e-05, + "step": 8588 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1319, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.8639183315788763e-07, + "loss": 0.7513, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 1.863918331578876e-05, + "step": 8589 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1318, + "image_masks_loss": 0.0153, + "llm_learning_rate": 1.8607763021295717e-07, + "loss": 0.7924, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 1.8607763021295717e-05, + "step": 8590 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1318, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.857636820823122e-07, + "loss": 0.8279, + "output_text_mask_loss": 0.0087, + "predictor_learning_rate": 1.8576368208231222e-05, + "step": 8591 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1318, + "image_masks_loss": 0.004, + "llm_learning_rate": 1.8544998880052511e-07, + "loss": 0.7852, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 1.854499888005251e-05, + "step": 8592 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1317, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.8513655040214102e-07, + "loss": 0.7266, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.85136550402141e-05, + "step": 8593 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1317, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.8482336692167592e-07, + "loss": 0.7231, + "output_text_mask_loss": 0.0452, + "predictor_learning_rate": 1.848233669216759e-05, + "step": 8594 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1317, + "image_masks_loss": 0.0271, + "llm_learning_rate": 1.845104383936186e-07, + "loss": 0.6305, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.845104383936186e-05, + "step": 8595 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1316, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.8419776485242874e-07, + "loss": 0.7223, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.8419776485242872e-05, + "step": 8596 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1316, + "image_masks_loss": 0.0079, + "llm_learning_rate": 1.8388534633253991e-07, + "loss": 0.7008, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 1.838853463325399e-05, + "step": 8597 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1316, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.8357318286835568e-07, + "loss": 0.6963, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 1.835731828683557e-05, + "step": 8598 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1315, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.8326127449425217e-07, + "loss": 0.7262, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 1.8326127449425217e-05, + "step": 8599 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1315, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.8294962124457723e-07, + "loss": 0.6761, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 1.8294962124457722e-05, + "step": 8600 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1315, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.8263822315365137e-07, + "loss": 0.616, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.8263822315365135e-05, + "step": 8601 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1314, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.8232708025576585e-07, + "loss": 0.7916, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.8232708025576584e-05, + "step": 8602 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1314, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.820161925851846e-07, + "loss": 0.7119, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.8201619258518458e-05, + "step": 8603 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1314, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.817055601761436e-07, + "loss": 0.7758, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.817055601761436e-05, + "step": 8604 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1314, + "image_masks_loss": 0.0037, + "llm_learning_rate": 1.8139518306284992e-07, + "loss": 0.7329, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 1.813951830628499e-05, + "step": 8605 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1313, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.8108506127948317e-07, + "loss": 0.7362, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 1.8108506127948315e-05, + "step": 8606 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1313, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.807751948601949e-07, + "loss": 0.7545, + "output_text_mask_loss": 0.0047, + "predictor_learning_rate": 1.807751948601949e-05, + "step": 8607 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1313, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.8046558383910783e-07, + "loss": 0.7725, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.804655838391078e-05, + "step": 8608 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1312, + "image_masks_loss": 0.0207, + "llm_learning_rate": 1.8015622825031692e-07, + "loss": 0.7217, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.801562282503169e-05, + "step": 8609 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1312, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.7984712812788964e-07, + "loss": 0.6658, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 1.798471281278896e-05, + "step": 8610 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1312, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.7953828350586456e-07, + "loss": 0.7332, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 1.7953828350586453e-05, + "step": 8611 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1311, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.7922969441825256e-07, + "loss": 0.7214, + "output_text_mask_loss": 0.0096, + "predictor_learning_rate": 1.7922969441825253e-05, + "step": 8612 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1311, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.789213608990359e-07, + "loss": 0.6941, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.789213608990359e-05, + "step": 8613 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1311, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.786132829821688e-07, + "loss": 0.6639, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 1.7861328298216876e-05, + "step": 8614 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.131, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.7830546070157773e-07, + "loss": 0.7239, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.783054607015777e-05, + "step": 8615 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.131, + "image_masks_loss": 0.024, + "llm_learning_rate": 1.779978940911603e-07, + "loss": 0.6968, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.779978940911603e-05, + "step": 8616 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.131, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.7769058318478722e-07, + "loss": 0.7611, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 1.776905831847872e-05, + "step": 8617 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.131, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.7738352801629972e-07, + "loss": 0.7129, + "output_text_mask_loss": 0.0515, + "predictor_learning_rate": 1.773835280162997e-05, + "step": 8618 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1309, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.7707672861951186e-07, + "loss": 0.7524, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.7707672861951186e-05, + "step": 8619 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1309, + "image_masks_loss": 0.0212, + "llm_learning_rate": 1.7677018502820858e-07, + "loss": 0.787, + "output_text_mask_loss": 0.0698, + "predictor_learning_rate": 1.7677018502820858e-05, + "step": 8620 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1309, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.7646389727614733e-07, + "loss": 0.7239, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.764638972761473e-05, + "step": 8621 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1308, + "image_masks_loss": 0.003, + "llm_learning_rate": 1.7615786539705693e-07, + "loss": 0.7097, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 1.761578653970569e-05, + "step": 8622 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1308, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.7585208942463878e-07, + "loss": 0.7309, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 1.7585208942463878e-05, + "step": 8623 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1308, + "image_masks_loss": 0.0045, + "llm_learning_rate": 1.7554656939256565e-07, + "loss": 0.5805, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.7554656939256562e-05, + "step": 8624 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1307, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.7524130533448147e-07, + "loss": 0.6891, + "output_text_mask_loss": 0.0338, + "predictor_learning_rate": 1.7524130533448146e-05, + "step": 8625 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1307, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.7493629728400296e-07, + "loss": 0.7804, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 1.7493629728400295e-05, + "step": 8626 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1307, + "image_masks_loss": 0.0179, + "llm_learning_rate": 1.7463154527471826e-07, + "loss": 0.7631, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.7463154527471824e-05, + "step": 8627 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1306, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.743270493401872e-07, + "loss": 0.6993, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 1.7432704934018715e-05, + "step": 8628 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1306, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.7402280951394178e-07, + "loss": 0.7011, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.740228095139418e-05, + "step": 8629 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1306, + "image_masks_loss": 0.0068, + "llm_learning_rate": 1.7371882582948557e-07, + "loss": 0.6949, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 1.7371882582948555e-05, + "step": 8630 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1306, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.7341509832029397e-07, + "loss": 0.7075, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.7341509832029396e-05, + "step": 8631 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1305, + "image_masks_loss": 0.0127, + "llm_learning_rate": 1.731116270198141e-07, + "loss": 0.7315, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.731116270198141e-05, + "step": 8632 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1305, + "image_masks_loss": 0.0024, + "llm_learning_rate": 1.7280841196146453e-07, + "loss": 0.7044, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.7280841196146452e-05, + "step": 8633 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1305, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.725054531786363e-07, + "loss": 0.7468, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 1.725054531786363e-05, + "step": 8634 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1304, + "image_masks_loss": 0.0175, + "llm_learning_rate": 1.7220275070469167e-07, + "loss": 0.6988, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 1.7220275070469167e-05, + "step": 8635 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1304, + "image_masks_loss": 0.0202, + "llm_learning_rate": 1.7190030457296531e-07, + "loss": 0.741, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 1.719003045729653e-05, + "step": 8636 + }, + { + "epoch": 0.88, + "gumbel_tau": 0.1304, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.7159811481676308e-07, + "loss": 0.6932, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 1.715981148167631e-05, + "step": 8637 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1303, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.7129618146936284e-07, + "loss": 0.8359, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.712961814693628e-05, + "step": 8638 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1303, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.7099450456401407e-07, + "loss": 0.7899, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 1.7099450456401404e-05, + "step": 8639 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1303, + "image_masks_loss": 0.011, + "llm_learning_rate": 1.7069308413393853e-07, + "loss": 0.6801, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 1.7069308413393852e-05, + "step": 8640 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1302, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.7039192021232803e-07, + "loss": 0.7484, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.70391920212328e-05, + "step": 8641 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1302, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.7009101283234852e-07, + "loss": 0.8015, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.700910128323485e-05, + "step": 8642 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1302, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.6979036202713656e-07, + "loss": 0.7536, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 1.6979036202713656e-05, + "step": 8643 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1302, + "image_masks_loss": 0.0043, + "llm_learning_rate": 1.694899678298001e-07, + "loss": 0.7102, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 1.694899678298001e-05, + "step": 8644 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1301, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.6918983027341963e-07, + "loss": 0.7751, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 1.691898302734196e-05, + "step": 8645 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1301, + "image_masks_loss": 0.0131, + "llm_learning_rate": 1.6888994939104647e-07, + "loss": 0.7023, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.6888994939104647e-05, + "step": 8646 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1301, + "image_masks_loss": 0.0037, + "llm_learning_rate": 1.685903252157045e-07, + "loss": 0.7675, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 1.685903252157045e-05, + "step": 8647 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.13, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.6829095778038872e-07, + "loss": 0.6863, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 1.6829095778038872e-05, + "step": 8648 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.13, + "image_masks_loss": 0.0038, + "llm_learning_rate": 1.6799184711806667e-07, + "loss": 0.6964, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 1.6799184711806664e-05, + "step": 8649 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.13, + "image_masks_loss": 0.0201, + "llm_learning_rate": 1.6769299326167666e-07, + "loss": 0.7222, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.6769299326167664e-05, + "step": 8650 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1299, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.6739439624412935e-07, + "loss": 0.7474, + "output_text_mask_loss": 0.0418, + "predictor_learning_rate": 1.6739439624412934e-05, + "step": 8651 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1299, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.6709605609830677e-07, + "loss": 0.7276, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.6709605609830675e-05, + "step": 8652 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1299, + "image_masks_loss": 0.0047, + "llm_learning_rate": 1.6679797285706262e-07, + "loss": 0.7684, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 1.667979728570626e-05, + "step": 8653 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1298, + "image_masks_loss": 0.0196, + "llm_learning_rate": 1.6650014655322288e-07, + "loss": 0.728, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 1.6650014655322286e-05, + "step": 8654 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1298, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.662025772195844e-07, + "loss": 0.7038, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 1.6620257721958436e-05, + "step": 8655 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1298, + "image_masks_loss": 0.003, + "llm_learning_rate": 1.6590526488891672e-07, + "loss": 0.8053, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.6590526488891673e-05, + "step": 8656 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1298, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.6560820959396012e-07, + "loss": 0.7448, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.6560820959396012e-05, + "step": 8657 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1297, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.6531141136742783e-07, + "loss": 0.7074, + "output_text_mask_loss": 0.0206, + "predictor_learning_rate": 1.6531141136742784e-05, + "step": 8658 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1297, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.6501487024200264e-07, + "loss": 0.7162, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.6501487024200262e-05, + "step": 8659 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1297, + "image_masks_loss": 0.0252, + "llm_learning_rate": 1.647185862503406e-07, + "loss": 0.8028, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 1.6471858625034057e-05, + "step": 8660 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1296, + "image_masks_loss": 0.0027, + "llm_learning_rate": 1.644225594250698e-07, + "loss": 0.7483, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.6442255942506978e-05, + "step": 8661 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1296, + "image_masks_loss": 0.0206, + "llm_learning_rate": 1.641267897987889e-07, + "loss": 0.702, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 1.6412678979878887e-05, + "step": 8662 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1296, + "image_masks_loss": 0.0416, + "llm_learning_rate": 1.63831277404069e-07, + "loss": 0.7653, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.63831277404069e-05, + "step": 8663 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1295, + "image_masks_loss": 0.0278, + "llm_learning_rate": 1.635360222734525e-07, + "loss": 0.6712, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 1.635360222734525e-05, + "step": 8664 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1295, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.6324102443945362e-07, + "loss": 0.7047, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.632410244394536e-05, + "step": 8665 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1295, + "image_masks_loss": 0.0116, + "llm_learning_rate": 1.6294628393455803e-07, + "loss": 0.7628, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.62946283934558e-05, + "step": 8666 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1294, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.6265180079122256e-07, + "loss": 0.7209, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 1.6265180079122256e-05, + "step": 8667 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1294, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.6235757504187765e-07, + "loss": 0.7208, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.6235757504187763e-05, + "step": 8668 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1294, + "image_masks_loss": 0.0229, + "llm_learning_rate": 1.6206360671892352e-07, + "loss": 0.7646, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 1.620636067189235e-05, + "step": 8669 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1294, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.617698958547323e-07, + "loss": 0.8163, + "output_text_mask_loss": 0.0315, + "predictor_learning_rate": 1.617698958547323e-05, + "step": 8670 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1293, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.614764424816484e-07, + "loss": 0.7462, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.614764424816484e-05, + "step": 8671 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1293, + "image_masks_loss": 0.0244, + "llm_learning_rate": 1.6118324663198765e-07, + "loss": 0.7463, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.6118324663198764e-05, + "step": 8672 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1293, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.6089030833803726e-07, + "loss": 0.7468, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 1.6089030833803726e-05, + "step": 8673 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1292, + "image_masks_loss": 0.0199, + "llm_learning_rate": 1.605976276320559e-07, + "loss": 0.7127, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.6059762763205587e-05, + "step": 8674 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1292, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.6030520454627495e-07, + "loss": 0.7348, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.6030520454627496e-05, + "step": 8675 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1292, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.600130391128965e-07, + "loss": 0.6031, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.600130391128965e-05, + "step": 8676 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1291, + "image_masks_loss": 0.0192, + "llm_learning_rate": 1.597211313640945e-07, + "loss": 0.6933, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 1.597211313640945e-05, + "step": 8677 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1291, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.594294813320141e-07, + "loss": 0.723, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 1.5942948133201407e-05, + "step": 8678 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1291, + "image_masks_loss": 0.0182, + "llm_learning_rate": 1.5913808904877233e-07, + "loss": 0.7199, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.591380890487723e-05, + "step": 8679 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1291, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.58846954546458e-07, + "loss": 0.755, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 1.58846954546458e-05, + "step": 8680 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.129, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.5855607785713212e-07, + "loss": 0.7197, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 1.585560778571321e-05, + "step": 8681 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.129, + "image_masks_loss": 0.0343, + "llm_learning_rate": 1.5826545901282659e-07, + "loss": 0.6895, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.5826545901282658e-05, + "step": 8682 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.129, + "image_masks_loss": 0.0083, + "llm_learning_rate": 1.579750980455444e-07, + "loss": 0.7814, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 1.579750980455444e-05, + "step": 8683 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1289, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.5768499498726108e-07, + "loss": 0.7366, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 1.576849949872611e-05, + "step": 8684 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1289, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.5739514986992361e-07, + "loss": 0.707, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 1.573951498699236e-05, + "step": 8685 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1289, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.5710556272545003e-07, + "loss": 0.7113, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.5710556272545002e-05, + "step": 8686 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1288, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.568162335857301e-07, + "loss": 0.8055, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 1.568162335857301e-05, + "step": 8687 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1288, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.5652716248262618e-07, + "loss": 0.828, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 1.5652716248262616e-05, + "step": 8688 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1288, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.5623834944797105e-07, + "loss": 0.7554, + "output_text_mask_loss": 0.0089, + "predictor_learning_rate": 1.5623834944797105e-05, + "step": 8689 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1287, + "image_masks_loss": 0.0259, + "llm_learning_rate": 1.5594979451356935e-07, + "loss": 0.7625, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 1.5594979451356934e-05, + "step": 8690 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1287, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.5566149771119727e-07, + "loss": 0.7291, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 1.5566149771119724e-05, + "step": 8691 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1287, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.5537345907260303e-07, + "loss": 0.7401, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 1.5537345907260304e-05, + "step": 8692 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1287, + "image_masks_loss": 0.0173, + "llm_learning_rate": 1.5508567862950603e-07, + "loss": 0.6947, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 1.55085678629506e-05, + "step": 8693 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1286, + "image_masks_loss": 0.0267, + "llm_learning_rate": 1.5479815641359668e-07, + "loss": 0.7143, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.5479815641359668e-05, + "step": 8694 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1286, + "image_masks_loss": 0.0097, + "llm_learning_rate": 1.5451089245653888e-07, + "loss": 0.6541, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 1.5451089245653888e-05, + "step": 8695 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1286, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.5422388678996564e-07, + "loss": 0.6938, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.5422388678996563e-05, + "step": 8696 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1285, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.5393713944548304e-07, + "loss": 0.7884, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 1.5393713944548303e-05, + "step": 8697 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1285, + "image_masks_loss": 0.0269, + "llm_learning_rate": 1.536506504546681e-07, + "loss": 0.7801, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 1.5365065045466808e-05, + "step": 8698 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1285, + "image_masks_loss": 0.0116, + "llm_learning_rate": 1.533644198490697e-07, + "loss": 0.7326, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.5336441984906967e-05, + "step": 8699 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1284, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.5307844766020847e-07, + "loss": 0.8424, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.5307844766020846e-05, + "step": 8700 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1284, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.5279273391957615e-07, + "loss": 0.6903, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 1.5279273391957616e-05, + "step": 8701 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1284, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.5250727865863623e-07, + "loss": 0.733, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 1.5250727865863623e-05, + "step": 8702 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1284, + "image_masks_loss": 0.0089, + "llm_learning_rate": 1.5222208190882382e-07, + "loss": 0.7415, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 1.5222208190882381e-05, + "step": 8703 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1283, + "image_masks_loss": 0.0265, + "llm_learning_rate": 1.5193714370154488e-07, + "loss": 0.7415, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.5193714370154487e-05, + "step": 8704 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1283, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.5165246406817798e-07, + "loss": 0.7059, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.5165246406817796e-05, + "step": 8705 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1283, + "image_masks_loss": 0.0178, + "llm_learning_rate": 1.5136804304007213e-07, + "loss": 0.7722, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 1.5136804304007212e-05, + "step": 8706 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1282, + "image_masks_loss": 0.0214, + "llm_learning_rate": 1.51083880648549e-07, + "loss": 0.6699, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 1.51083880648549e-05, + "step": 8707 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1282, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.5079997692490105e-07, + "loss": 0.7648, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 1.5079997692490105e-05, + "step": 8708 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1282, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.5051633190039216e-07, + "loss": 0.6681, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 1.5051633190039216e-05, + "step": 8709 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1281, + "image_masks_loss": 0.0022, + "llm_learning_rate": 1.5023294560625817e-07, + "loss": 0.7461, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.5023294560625816e-05, + "step": 8710 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1281, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.4994981807370607e-07, + "loss": 0.7997, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 1.4994981807370605e-05, + "step": 8711 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1281, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.496669493339145e-07, + "loss": 0.6333, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.496669493339145e-05, + "step": 8712 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.128, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.493843394180333e-07, + "loss": 0.697, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.4938433941803331e-05, + "step": 8713 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.128, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.491019883571851e-07, + "loss": 0.718, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 1.4910198835718508e-05, + "step": 8714 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.128, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.4881989618246217e-07, + "loss": 0.6742, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.4881989618246217e-05, + "step": 8715 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.128, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.4853806292492918e-07, + "loss": 0.7287, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.4853806292492916e-05, + "step": 8716 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1279, + "image_masks_loss": 0.0125, + "llm_learning_rate": 1.4825648861562235e-07, + "loss": 0.8626, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.4825648861562236e-05, + "step": 8717 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1279, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.4797517328554944e-07, + "loss": 0.7269, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.4797517328554944e-05, + "step": 8718 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1279, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.4769411696568897e-07, + "loss": 0.7244, + "output_text_mask_loss": 0.0168, + "predictor_learning_rate": 1.4769411696568896e-05, + "step": 8719 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1278, + "image_masks_loss": 0.0123, + "llm_learning_rate": 1.4741331968699228e-07, + "loss": 0.6273, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 1.4741331968699229e-05, + "step": 8720 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1278, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.4713278148038078e-07, + "loss": 0.7203, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.4713278148038078e-05, + "step": 8721 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1278, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.4685250237674836e-07, + "loss": 0.7757, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 1.4685250237674836e-05, + "step": 8722 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1277, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.4657248240696005e-07, + "loss": 0.739, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.4657248240696003e-05, + "step": 8723 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1277, + "image_masks_loss": 0.0136, + "llm_learning_rate": 1.462927216018517e-07, + "loss": 0.7695, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.462927216018517e-05, + "step": 8724 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1277, + "image_masks_loss": 0.0267, + "llm_learning_rate": 1.460132199922315e-07, + "loss": 0.7247, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.460132199922315e-05, + "step": 8725 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1277, + "image_masks_loss": 0.023, + "llm_learning_rate": 1.457339776088787e-07, + "loss": 0.6929, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 1.4573397760887869e-05, + "step": 8726 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1276, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.4545499448254425e-07, + "loss": 0.8211, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 1.4545499448254424e-05, + "step": 8727 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1276, + "image_masks_loss": 0.0204, + "llm_learning_rate": 1.451762706439508e-07, + "loss": 0.7512, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.4517627064395079e-05, + "step": 8728 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1276, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.448978061237913e-07, + "loss": 0.7652, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.4489780612379127e-05, + "step": 8729 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1275, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.446196009527312e-07, + "loss": 0.7684, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 1.446196009527312e-05, + "step": 8730 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1275, + "image_masks_loss": 0.0161, + "llm_learning_rate": 1.4434165516140687e-07, + "loss": 0.7412, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.4434165516140684e-05, + "step": 8731 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1275, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.4406396878042683e-07, + "loss": 0.7533, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.4406396878042682e-05, + "step": 8732 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1274, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.4378654184037e-07, + "loss": 0.7272, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.4378654184036999e-05, + "step": 8733 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1274, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.4350937437178748e-07, + "loss": 0.7562, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 1.4350937437178745e-05, + "step": 8734 + }, + { + "epoch": 0.89, + "gumbel_tau": 0.1274, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.432324664052015e-07, + "loss": 0.7231, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 1.432324664052015e-05, + "step": 8735 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1274, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.4295581797110575e-07, + "loss": 0.7295, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.4295581797110574e-05, + "step": 8736 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1273, + "image_masks_loss": 0.0033, + "llm_learning_rate": 1.4267942909996558e-07, + "loss": 0.6541, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.4267942909996556e-05, + "step": 8737 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1273, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.4240329982221717e-07, + "loss": 0.7271, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 1.4240329982221717e-05, + "step": 8738 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1273, + "image_masks_loss": 0.0152, + "llm_learning_rate": 1.4212743016826874e-07, + "loss": 0.7323, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 1.4212743016826873e-05, + "step": 8739 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1272, + "image_masks_loss": 0.0218, + "llm_learning_rate": 1.4185182016849984e-07, + "loss": 0.7726, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.4185182016849984e-05, + "step": 8740 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1272, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.4157646985326124e-07, + "loss": 0.764, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 1.4157646985326123e-05, + "step": 8741 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1272, + "image_masks_loss": 0.0153, + "llm_learning_rate": 1.4130137925287473e-07, + "loss": 0.6991, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 1.4130137925287474e-05, + "step": 8742 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1271, + "image_masks_loss": 0.0209, + "llm_learning_rate": 1.410265483976342e-07, + "loss": 0.7799, + "output_text_mask_loss": 0.015, + "predictor_learning_rate": 1.4102654839763418e-05, + "step": 8743 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1271, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.4075197731780455e-07, + "loss": 0.6548, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.4075197731780454e-05, + "step": 8744 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1271, + "image_masks_loss": 0.0047, + "llm_learning_rate": 1.4047766604362162e-07, + "loss": 0.6861, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 1.4047766604362162e-05, + "step": 8745 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1271, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.402036146052943e-07, + "loss": 0.7025, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 1.4020361460529429e-05, + "step": 8746 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.127, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.3992982303300123e-07, + "loss": 0.6575, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.3992982303300122e-05, + "step": 8747 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.127, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.39656291356893e-07, + "loss": 0.697, + "output_text_mask_loss": 0.0389, + "predictor_learning_rate": 1.3965629135689301e-05, + "step": 8748 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.127, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.393830196070911e-07, + "loss": 0.7585, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.393830196070911e-05, + "step": 8749 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1269, + "image_masks_loss": 0.0038, + "llm_learning_rate": 1.391100078136895e-07, + "loss": 0.7458, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 1.3911000781368949e-05, + "step": 8750 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1269, + "image_masks_loss": 0.0035, + "llm_learning_rate": 1.388372560067522e-07, + "loss": 0.7449, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.3883725600675218e-05, + "step": 8751 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1269, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.385647642163157e-07, + "loss": 0.7873, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 1.3856476421631569e-05, + "step": 8752 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1268, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.3829253247238715e-07, + "loss": 0.8259, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 1.3829253247238715e-05, + "step": 8753 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1268, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.3802056080494563e-07, + "loss": 0.625, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 1.3802056080494563e-05, + "step": 8754 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1268, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.3774884924394078e-07, + "loss": 0.7088, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 1.3774884924394076e-05, + "step": 8755 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1268, + "image_masks_loss": 0.0145, + "llm_learning_rate": 1.3747739781929475e-07, + "loss": 0.7452, + "output_text_mask_loss": 0.0214, + "predictor_learning_rate": 1.3747739781929475e-05, + "step": 8756 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1267, + "image_masks_loss": 0.018, + "llm_learning_rate": 1.372062065608998e-07, + "loss": 0.6263, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.3720620656089978e-05, + "step": 8757 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1267, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.3693527549861973e-07, + "loss": 0.6954, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.3693527549861972e-05, + "step": 8758 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1267, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.366646046622913e-07, + "loss": 0.7323, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 1.366646046622913e-05, + "step": 8759 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1266, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.3639419408172037e-07, + "loss": 0.794, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 1.3639419408172039e-05, + "step": 8760 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1266, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.3612404378668593e-07, + "loss": 0.709, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.3612404378668591e-05, + "step": 8761 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1266, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.358541538069369e-07, + "loss": 0.6829, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.358541538069369e-05, + "step": 8762 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1265, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.3558452417219425e-07, + "loss": 0.6852, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 1.3558452417219425e-05, + "step": 8763 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1265, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.3531515491215064e-07, + "loss": 0.8323, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 1.3531515491215064e-05, + "step": 8764 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1265, + "image_masks_loss": 0.0061, + "llm_learning_rate": 1.350460460564687e-07, + "loss": 0.7113, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 1.3504604605646869e-05, + "step": 8765 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1265, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.3477719763478448e-07, + "loss": 0.7826, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 1.3477719763478446e-05, + "step": 8766 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1264, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.3450860967670342e-07, + "loss": 0.7245, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.3450860967670342e-05, + "step": 8767 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1264, + "image_masks_loss": 0.0174, + "llm_learning_rate": 1.3424028221180303e-07, + "loss": 0.7493, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 1.3424028221180301e-05, + "step": 8768 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1264, + "image_masks_loss": 0.0271, + "llm_learning_rate": 1.3397221526963294e-07, + "loss": 0.6928, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.3397221526963293e-05, + "step": 8769 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1263, + "image_masks_loss": 0.0186, + "llm_learning_rate": 1.337044088797118e-07, + "loss": 0.6984, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.337044088797118e-05, + "step": 8770 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1263, + "image_masks_loss": 0.0213, + "llm_learning_rate": 1.3343686307153213e-07, + "loss": 0.7227, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.3343686307153214e-05, + "step": 8771 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1263, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.3316957787455648e-07, + "loss": 0.7726, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 1.3316957787455647e-05, + "step": 8772 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1262, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.3290255331821906e-07, + "loss": 0.7847, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.3290255331821905e-05, + "step": 8773 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1262, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.3263578943192468e-07, + "loss": 0.6793, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 1.3263578943192467e-05, + "step": 8774 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1262, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.323692862450504e-07, + "loss": 0.7829, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 1.323692862450504e-05, + "step": 8775 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1262, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.3210304378694387e-07, + "loss": 0.7368, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 1.3210304378694387e-05, + "step": 8776 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1261, + "image_masks_loss": 0.0229, + "llm_learning_rate": 1.3183706208692414e-07, + "loss": 0.608, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.3183706208692415e-05, + "step": 8777 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1261, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.315713411742825e-07, + "loss": 0.771, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 1.315713411742825e-05, + "step": 8778 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1261, + "image_masks_loss": 0.0151, + "llm_learning_rate": 1.3130588107827997e-07, + "loss": 0.7973, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.3130588107827995e-05, + "step": 8779 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.126, + "image_masks_loss": 0.0235, + "llm_learning_rate": 1.3104068182815012e-07, + "loss": 0.7273, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.310406818281501e-05, + "step": 8780 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.126, + "image_masks_loss": 0.0161, + "llm_learning_rate": 1.307757434530968e-07, + "loss": 0.706, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.3077574345309678e-05, + "step": 8781 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.126, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.3051106598229614e-07, + "loss": 0.7958, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 1.3051106598229612e-05, + "step": 8782 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.126, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.302466494448945e-07, + "loss": 0.6995, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 1.302466494448945e-05, + "step": 8783 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1259, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.2998249387001006e-07, + "loss": 0.684, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 1.2998249387001005e-05, + "step": 8784 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1259, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.2971859928673254e-07, + "loss": 0.7203, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.2971859928673252e-05, + "step": 8785 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1259, + "image_masks_loss": 0.028, + "llm_learning_rate": 1.2945496572412259e-07, + "loss": 0.7403, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 1.2945496572412258e-05, + "step": 8786 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1258, + "image_masks_loss": 0.0019, + "llm_learning_rate": 1.29191593211212e-07, + "loss": 0.7358, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 1.29191593211212e-05, + "step": 8787 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1258, + "image_masks_loss": 0.0053, + "llm_learning_rate": 1.289284817770045e-07, + "loss": 0.6563, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.289284817770045e-05, + "step": 8788 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1258, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.2866563145047333e-07, + "loss": 0.7862, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.2866563145047333e-05, + "step": 8789 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1257, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.284030422605645e-07, + "loss": 0.7871, + "output_text_mask_loss": 0.0246, + "predictor_learning_rate": 1.2840304226056448e-05, + "step": 8790 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1257, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.281407142361954e-07, + "loss": 0.7705, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.281407142361954e-05, + "step": 8791 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1257, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.2787864740625404e-07, + "loss": 0.6448, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 1.2787864740625404e-05, + "step": 8792 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1257, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.2761684179959987e-07, + "loss": 0.6793, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.2761684179959988e-05, + "step": 8793 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1256, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.2735529744506315e-07, + "loss": 0.7669, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.2735529744506314e-05, + "step": 8794 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1256, + "image_masks_loss": 0.0359, + "llm_learning_rate": 1.270940143714461e-07, + "loss": 0.7493, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 1.2709401437144607e-05, + "step": 8795 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1256, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.2683299260752152e-07, + "loss": 0.8104, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.2683299260752152e-05, + "step": 8796 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1255, + "image_masks_loss": 0.0123, + "llm_learning_rate": 1.2657223218203341e-07, + "loss": 0.7517, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 1.265722321820334e-05, + "step": 8797 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1255, + "image_masks_loss": 0.0029, + "llm_learning_rate": 1.2631173312369821e-07, + "loss": 0.6626, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.263117331236982e-05, + "step": 8798 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1255, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.2605149546120189e-07, + "loss": 0.8359, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 1.2605149546120188e-05, + "step": 8799 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1254, + "image_masks_loss": 0.0195, + "llm_learning_rate": 1.2579151922320286e-07, + "loss": 0.7824, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.2579151922320287e-05, + "step": 8800 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1254, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.2553180443832996e-07, + "loss": 0.6469, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 1.2553180443832995e-05, + "step": 8801 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1254, + "image_masks_loss": 0.0115, + "llm_learning_rate": 1.2527235113518388e-07, + "loss": 0.7615, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 1.2527235113518387e-05, + "step": 8802 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1254, + "image_masks_loss": 0.0059, + "llm_learning_rate": 1.2501315934233565e-07, + "loss": 0.7876, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.2501315934233564e-05, + "step": 8803 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1253, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.247542290883283e-07, + "loss": 0.8305, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 1.2475422908832828e-05, + "step": 8804 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1253, + "image_masks_loss": 0.0051, + "llm_learning_rate": 1.244955604016762e-07, + "loss": 0.725, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 1.244955604016762e-05, + "step": 8805 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1253, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.2423715331086467e-07, + "loss": 0.6641, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.2423715331086465e-05, + "step": 8806 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1252, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.2397900784434896e-07, + "loss": 0.6688, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.2397900784434895e-05, + "step": 8807 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1252, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.2372112403055748e-07, + "loss": 0.7677, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.2372112403055747e-05, + "step": 8808 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1252, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.2346350189788858e-07, + "loss": 0.7361, + "output_text_mask_loss": 0.0113, + "predictor_learning_rate": 1.2346350189788858e-05, + "step": 8809 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1252, + "image_masks_loss": 0.0066, + "llm_learning_rate": 1.2320614147471266e-07, + "loss": 0.7454, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 1.2320614147471265e-05, + "step": 8810 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1251, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.2294904278937038e-07, + "loss": 0.6886, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 1.2294904278937036e-05, + "step": 8811 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1251, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.2269220587017434e-07, + "loss": 0.8163, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.2269220587017432e-05, + "step": 8812 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1251, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.2243563074540777e-07, + "loss": 0.7488, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.2243563074540776e-05, + "step": 8813 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.125, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.2217931744332528e-07, + "loss": 0.7743, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 1.2217931744332529e-05, + "step": 8814 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.125, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.2192326599215265e-07, + "loss": 0.7145, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.2192326599215264e-05, + "step": 8815 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.125, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.2166747642008674e-07, + "loss": 0.7092, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 1.2166747642008675e-05, + "step": 8816 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1249, + "image_masks_loss": 0.0098, + "llm_learning_rate": 1.2141194875529615e-07, + "loss": 0.8035, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 1.2141194875529616e-05, + "step": 8817 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1249, + "image_masks_loss": 0.0224, + "llm_learning_rate": 1.2115668302592005e-07, + "loss": 0.6984, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 1.2115668302592003e-05, + "step": 8818 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1249, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.209016792600684e-07, + "loss": 0.8197, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.209016792600684e-05, + "step": 8819 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1249, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.2064693748582323e-07, + "loss": 0.7434, + "output_text_mask_loss": 0.01, + "predictor_learning_rate": 1.2064693748582322e-05, + "step": 8820 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1248, + "image_masks_loss": 0.0173, + "llm_learning_rate": 1.2039245773123709e-07, + "loss": 0.7838, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 1.2039245773123708e-05, + "step": 8821 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1248, + "image_masks_loss": 0.0159, + "llm_learning_rate": 1.2013824002433394e-07, + "loss": 0.8008, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.2013824002433393e-05, + "step": 8822 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1248, + "image_masks_loss": 0.0066, + "llm_learning_rate": 1.1988428439310835e-07, + "loss": 0.6374, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.1988428439310833e-05, + "step": 8823 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1247, + "image_masks_loss": 0.0155, + "llm_learning_rate": 1.1963059086552708e-07, + "loss": 0.7304, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.1963059086552708e-05, + "step": 8824 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1247, + "image_masks_loss": 0.0151, + "llm_learning_rate": 1.1937715946952756e-07, + "loss": 0.7899, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 1.1937715946952754e-05, + "step": 8825 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1247, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.1912399023301768e-07, + "loss": 0.5995, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1912399023301768e-05, + "step": 8826 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1247, + "image_masks_loss": 0.0275, + "llm_learning_rate": 1.1887108318387713e-07, + "loss": 0.7276, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.1887108318387713e-05, + "step": 8827 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1246, + "image_masks_loss": 0.004, + "llm_learning_rate": 1.1861843834995668e-07, + "loss": 0.6798, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1861843834995667e-05, + "step": 8828 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1246, + "image_masks_loss": 0.0114, + "llm_learning_rate": 1.1836605575907767e-07, + "loss": 0.6673, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 1.1836605575907767e-05, + "step": 8829 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1246, + "image_masks_loss": 0.0094, + "llm_learning_rate": 1.18113935439034e-07, + "loss": 0.7793, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.1811393543903399e-05, + "step": 8830 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1245, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.1786207741758898e-07, + "loss": 0.7336, + "output_text_mask_loss": 0.0128, + "predictor_learning_rate": 1.1786207741758898e-05, + "step": 8831 + }, + { + "epoch": 0.9, + "gumbel_tau": 0.1245, + "image_masks_loss": 0.0104, + "llm_learning_rate": 1.1761048172247824e-07, + "loss": 0.6731, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.1761048172247824e-05, + "step": 8832 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1245, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.1735914838140767e-07, + "loss": 0.7194, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 1.1735914838140765e-05, + "step": 8833 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1244, + "image_masks_loss": 0.0033, + "llm_learning_rate": 1.1710807742205482e-07, + "loss": 0.7293, + "output_text_mask_loss": 0.0091, + "predictor_learning_rate": 1.1710807742205482e-05, + "step": 8834 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1244, + "image_masks_loss": 0.0145, + "llm_learning_rate": 1.1685726887206816e-07, + "loss": 0.7541, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 1.1685726887206815e-05, + "step": 8835 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1244, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.1660672275906698e-07, + "loss": 0.7322, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 1.1660672275906697e-05, + "step": 8836 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1244, + "image_masks_loss": 0.0155, + "llm_learning_rate": 1.1635643911064254e-07, + "loss": 0.7554, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 1.1635643911064252e-05, + "step": 8837 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1243, + "image_masks_loss": 0.0136, + "llm_learning_rate": 1.161064179543564e-07, + "loss": 0.8314, + "output_text_mask_loss": 0.0018, + "predictor_learning_rate": 1.161064179543564e-05, + "step": 8838 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1243, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.1585665931774154e-07, + "loss": 0.754, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1585665931774153e-05, + "step": 8839 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1243, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.1560716322830179e-07, + "loss": 0.7534, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.1560716322830179e-05, + "step": 8840 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1242, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.1535792971351239e-07, + "loss": 0.7852, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 1.1535792971351238e-05, + "step": 8841 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1242, + "image_masks_loss": 0.005, + "llm_learning_rate": 1.1510895880081918e-07, + "loss": 0.8167, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 1.1510895880081918e-05, + "step": 8842 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1242, + "image_masks_loss": 0.0199, + "llm_learning_rate": 1.1486025051763939e-07, + "loss": 0.6352, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 1.1486025051763938e-05, + "step": 8843 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1242, + "image_masks_loss": 0.0182, + "llm_learning_rate": 1.1461180489136192e-07, + "loss": 0.7702, + "output_text_mask_loss": 0.0196, + "predictor_learning_rate": 1.1461180489136192e-05, + "step": 8844 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1241, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.1436362194934547e-07, + "loss": 0.7511, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1436362194934547e-05, + "step": 8845 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1241, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.1411570171892067e-07, + "loss": 0.7522, + "output_text_mask_loss": 0.0591, + "predictor_learning_rate": 1.1411570171892066e-05, + "step": 8846 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1241, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.1386804422738929e-07, + "loss": 0.6745, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 1.1386804422738928e-05, + "step": 8847 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.124, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.1362064950202395e-07, + "loss": 0.8084, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 1.1362064950202394e-05, + "step": 8848 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.124, + "image_masks_loss": 0.0236, + "llm_learning_rate": 1.1337351757006759e-07, + "loss": 0.6657, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.1337351757006759e-05, + "step": 8849 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.124, + "image_masks_loss": 0.0026, + "llm_learning_rate": 1.1312664845873566e-07, + "loss": 0.6803, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 1.1312664845873567e-05, + "step": 8850 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1239, + "image_masks_loss": 0.0123, + "llm_learning_rate": 1.128800421952142e-07, + "loss": 0.792, + "output_text_mask_loss": 0.0341, + "predictor_learning_rate": 1.128800421952142e-05, + "step": 8851 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1239, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.1263369880665926e-07, + "loss": 0.6635, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1263369880665925e-05, + "step": 8852 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1239, + "image_masks_loss": 0.0231, + "llm_learning_rate": 1.1238761832019912e-07, + "loss": 0.8101, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.1238761832019912e-05, + "step": 8853 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1239, + "image_masks_loss": 0.0125, + "llm_learning_rate": 1.1214180076293269e-07, + "loss": 0.7402, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1214180076293268e-05, + "step": 8854 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1238, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.1189624616192995e-07, + "loss": 0.7664, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 1.1189624616192996e-05, + "step": 8855 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1238, + "image_masks_loss": 0.0124, + "llm_learning_rate": 1.1165095454423152e-07, + "loss": 0.6622, + "output_text_mask_loss": 0.0018, + "predictor_learning_rate": 1.1165095454423151e-05, + "step": 8856 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1238, + "image_masks_loss": 0.0144, + "llm_learning_rate": 1.1140592593684996e-07, + "loss": 0.6798, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1140592593684996e-05, + "step": 8857 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1237, + "image_masks_loss": 0.004, + "llm_learning_rate": 1.1116116036676839e-07, + "loss": 0.6305, + "output_text_mask_loss": 0.0465, + "predictor_learning_rate": 1.1116116036676837e-05, + "step": 8858 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1237, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.109166578609408e-07, + "loss": 0.7412, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.109166578609408e-05, + "step": 8859 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1237, + "image_masks_loss": 0.0194, + "llm_learning_rate": 1.1067241844629234e-07, + "loss": 0.8228, + "output_text_mask_loss": 0.0134, + "predictor_learning_rate": 1.1067241844629234e-05, + "step": 8860 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1237, + "image_masks_loss": 0.0045, + "llm_learning_rate": 1.1042844214971898e-07, + "loss": 0.7109, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1042844214971898e-05, + "step": 8861 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1236, + "image_masks_loss": 0.0154, + "llm_learning_rate": 1.101847289980884e-07, + "loss": 0.6877, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 1.101847289980884e-05, + "step": 8862 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1236, + "image_masks_loss": 0.0168, + "llm_learning_rate": 1.0994127901823831e-07, + "loss": 0.7084, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.099412790182383e-05, + "step": 8863 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1236, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.0969809223697836e-07, + "loss": 0.7857, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.0969809223697835e-05, + "step": 8864 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1235, + "image_masks_loss": 0.0192, + "llm_learning_rate": 1.0945516868108852e-07, + "loss": 0.8212, + "output_text_mask_loss": 0.004, + "predictor_learning_rate": 1.094551686810885e-05, + "step": 8865 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1235, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.0921250837732045e-07, + "loss": 0.7489, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.0921250837732044e-05, + "step": 8866 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1235, + "image_masks_loss": 0.0125, + "llm_learning_rate": 1.089701113523961e-07, + "loss": 0.655, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 1.0897011135239609e-05, + "step": 8867 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1235, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.0872797763300885e-07, + "loss": 0.7689, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 1.0872797763300885e-05, + "step": 8868 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1234, + "image_masks_loss": 0.0028, + "llm_learning_rate": 1.0848610724582292e-07, + "loss": 0.7694, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 1.0848610724582292e-05, + "step": 8869 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1234, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.0824450021747424e-07, + "loss": 0.7944, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 1.0824450021747423e-05, + "step": 8870 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1234, + "image_masks_loss": 0.0123, + "llm_learning_rate": 1.0800315657456816e-07, + "loss": 0.7491, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.0800315657456816e-05, + "step": 8871 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1233, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.0776207634368263e-07, + "loss": 0.6967, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 1.0776207634368263e-05, + "step": 8872 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1233, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.0752125955136583e-07, + "loss": 0.7, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.0752125955136583e-05, + "step": 8873 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1233, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.0728070622413683e-07, + "loss": 0.7304, + "output_text_mask_loss": 0.0128, + "predictor_learning_rate": 1.0728070622413682e-05, + "step": 8874 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1232, + "image_masks_loss": 0.0151, + "llm_learning_rate": 1.0704041638848584e-07, + "loss": 0.7377, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.0704041638848583e-05, + "step": 8875 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1232, + "image_masks_loss": 0.0025, + "llm_learning_rate": 1.0680039007087473e-07, + "loss": 0.7401, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 1.0680039007087471e-05, + "step": 8876 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1232, + "image_masks_loss": 0.0188, + "llm_learning_rate": 1.0656062729773515e-07, + "loss": 0.7327, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 1.0656062729773513e-05, + "step": 8877 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1232, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.0632112809547041e-07, + "loss": 0.7228, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.063211280954704e-05, + "step": 8878 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1231, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.060818924904547e-07, + "loss": 0.6908, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 1.0608189249045469e-05, + "step": 8879 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1231, + "image_masks_loss": 0.0095, + "llm_learning_rate": 1.0584292050903389e-07, + "loss": 0.6903, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.0584292050903388e-05, + "step": 8880 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1231, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.0560421217752247e-07, + "loss": 0.7345, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 1.0560421217752247e-05, + "step": 8881 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.123, + "image_masks_loss": 0.0389, + "llm_learning_rate": 1.053657675222089e-07, + "loss": 0.65, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 1.053657675222089e-05, + "step": 8882 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.123, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.0512758656935102e-07, + "loss": 0.7619, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 1.05127586569351e-05, + "step": 8883 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.123, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.0488966934517758e-07, + "loss": 0.7478, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 1.0488966934517757e-05, + "step": 8884 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.123, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.0465201587588847e-07, + "loss": 0.7197, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 1.0465201587588846e-05, + "step": 8885 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1229, + "image_masks_loss": 0.0133, + "llm_learning_rate": 1.0441462618765497e-07, + "loss": 0.8277, + "output_text_mask_loss": 0.0227, + "predictor_learning_rate": 1.0441462618765495e-05, + "step": 8886 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1229, + "image_masks_loss": 0.0156, + "llm_learning_rate": 1.0417750030661893e-07, + "loss": 0.7288, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 1.0417750030661893e-05, + "step": 8887 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1229, + "image_masks_loss": 0.009, + "llm_learning_rate": 1.0394063825889255e-07, + "loss": 0.6973, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 1.0394063825889255e-05, + "step": 8888 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1228, + "image_masks_loss": 0.0153, + "llm_learning_rate": 1.037040400705605e-07, + "loss": 0.7645, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 1.037040400705605e-05, + "step": 8889 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1228, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.0346770576767696e-07, + "loss": 0.7367, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 1.0346770576767694e-05, + "step": 8890 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1228, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.0323163537626774e-07, + "loss": 0.7317, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 1.0323163537626774e-05, + "step": 8891 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1228, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.0299582892232956e-07, + "loss": 0.6251, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.0299582892232956e-05, + "step": 8892 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1227, + "image_masks_loss": 0.0069, + "llm_learning_rate": 1.0276028643182972e-07, + "loss": 0.6347, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.0276028643182972e-05, + "step": 8893 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1227, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.0252500793070691e-07, + "loss": 0.7969, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 1.0252500793070691e-05, + "step": 8894 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1227, + "image_masks_loss": 0.0118, + "llm_learning_rate": 1.0228999344486984e-07, + "loss": 0.7038, + "output_text_mask_loss": 0.0189, + "predictor_learning_rate": 1.0228999344486984e-05, + "step": 8895 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1226, + "image_masks_loss": 0.0148, + "llm_learning_rate": 1.0205524300020004e-07, + "loss": 0.7351, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.0205524300020003e-05, + "step": 8896 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1226, + "image_masks_loss": 0.0122, + "llm_learning_rate": 1.0182075662254792e-07, + "loss": 0.7054, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 1.0182075662254792e-05, + "step": 8897 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1226, + "image_masks_loss": 0.0165, + "llm_learning_rate": 1.0158653433773563e-07, + "loss": 0.7168, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 1.0158653433773563e-05, + "step": 8898 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1226, + "image_masks_loss": 0.013, + "llm_learning_rate": 1.0135257617155697e-07, + "loss": 0.7135, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.0135257617155697e-05, + "step": 8899 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1225, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.0111888214977467e-07, + "loss": 0.7436, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 1.0111888214977466e-05, + "step": 8900 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1225, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.0088545229812479e-07, + "loss": 0.6568, + "output_text_mask_loss": 0.0061, + "predictor_learning_rate": 1.008854522981248e-05, + "step": 8901 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1225, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.0065228664231264e-07, + "loss": 0.7334, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 1.0065228664231263e-05, + "step": 8902 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1224, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.0041938520801487e-07, + "loss": 0.7626, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.0041938520801486e-05, + "step": 8903 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1224, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.0018674802087958e-07, + "loss": 0.665, + "output_text_mask_loss": 0.0094, + "predictor_learning_rate": 1.0018674802087958e-05, + "step": 8904 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1224, + "image_masks_loss": 0.007, + "llm_learning_rate": 9.995437510652462e-08, + "loss": 0.7879, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 9.995437510652461e-06, + "step": 8905 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1223, + "image_masks_loss": 0.0219, + "llm_learning_rate": 9.972226649054007e-08, + "loss": 0.7162, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 9.972226649054006e-06, + "step": 8906 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1223, + "image_masks_loss": 0.0179, + "llm_learning_rate": 9.949042219848548e-08, + "loss": 0.7404, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.949042219848547e-06, + "step": 8907 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1223, + "image_masks_loss": 0.0114, + "llm_learning_rate": 9.925884225589294e-08, + "loss": 0.6469, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.925884225589293e-06, + "step": 8908 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1223, + "image_masks_loss": 0.0341, + "llm_learning_rate": 9.902752668826399e-08, + "loss": 0.7651, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 9.902752668826398e-06, + "step": 8909 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1222, + "image_masks_loss": 0.0103, + "llm_learning_rate": 9.879647552107186e-08, + "loss": 0.6822, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 9.879647552107184e-06, + "step": 8910 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1222, + "image_masks_loss": 0.0025, + "llm_learning_rate": 9.856568877976036e-08, + "loss": 0.7859, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 9.856568877976036e-06, + "step": 8911 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1222, + "image_masks_loss": 0.0052, + "llm_learning_rate": 9.833516648974417e-08, + "loss": 0.7856, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 9.833516648974417e-06, + "step": 8912 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1221, + "image_masks_loss": 0.0146, + "llm_learning_rate": 9.810490867640882e-08, + "loss": 0.7129, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 9.81049086764088e-06, + "step": 8913 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1221, + "image_masks_loss": 0.0106, + "llm_learning_rate": 9.787491536511068e-08, + "loss": 0.736, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.787491536511068e-06, + "step": 8914 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1221, + "image_masks_loss": 0.0068, + "llm_learning_rate": 9.764518658117782e-08, + "loss": 0.7149, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 9.764518658117783e-06, + "step": 8915 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1221, + "image_masks_loss": 0.0046, + "llm_learning_rate": 9.741572234990782e-08, + "loss": 0.7737, + "output_text_mask_loss": 0.0145, + "predictor_learning_rate": 9.74157223499078e-06, + "step": 8916 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.122, + "image_masks_loss": 0.0085, + "llm_learning_rate": 9.718652269657014e-08, + "loss": 0.7025, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 9.718652269657014e-06, + "step": 8917 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.122, + "image_masks_loss": 0.0122, + "llm_learning_rate": 9.695758764640517e-08, + "loss": 0.7019, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 9.695758764640517e-06, + "step": 8918 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.122, + "image_masks_loss": 0.0039, + "llm_learning_rate": 9.672891722462247e-08, + "loss": 0.7207, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 9.672891722462247e-06, + "step": 8919 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1219, + "image_masks_loss": 0.0018, + "llm_learning_rate": 9.650051145640438e-08, + "loss": 0.7198, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 9.650051145640438e-06, + "step": 8920 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1219, + "image_masks_loss": 0.0055, + "llm_learning_rate": 9.627237036690412e-08, + "loss": 0.7294, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 9.627237036690412e-06, + "step": 8921 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1219, + "image_masks_loss": 0.0071, + "llm_learning_rate": 9.60444939812441e-08, + "loss": 0.7548, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 9.60444939812441e-06, + "step": 8922 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1219, + "image_masks_loss": 0.0049, + "llm_learning_rate": 9.581688232451925e-08, + "loss": 0.7704, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 9.581688232451923e-06, + "step": 8923 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1218, + "image_masks_loss": 0.0165, + "llm_learning_rate": 9.55895354217945e-08, + "loss": 0.7533, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 9.55895354217945e-06, + "step": 8924 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1218, + "image_masks_loss": 0.0072, + "llm_learning_rate": 9.536245329810539e-08, + "loss": 0.8152, + "output_text_mask_loss": 0.0463, + "predictor_learning_rate": 9.536245329810538e-06, + "step": 8925 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1218, + "image_masks_loss": 0.0035, + "llm_learning_rate": 9.513563597845942e-08, + "loss": 0.5717, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 9.513563597845943e-06, + "step": 8926 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1217, + "image_masks_loss": 0.0042, + "llm_learning_rate": 9.490908348783329e-08, + "loss": 0.7056, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 9.490908348783329e-06, + "step": 8927 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1217, + "image_masks_loss": 0.0109, + "llm_learning_rate": 9.468279585117646e-08, + "loss": 0.6814, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 9.468279585117646e-06, + "step": 8928 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1217, + "image_masks_loss": 0.0041, + "llm_learning_rate": 9.445677309340761e-08, + "loss": 0.6793, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 9.445677309340761e-06, + "step": 8929 + }, + { + "epoch": 0.91, + "gumbel_tau": 0.1217, + "image_masks_loss": 0.0048, + "llm_learning_rate": 9.423101523941714e-08, + "loss": 0.7147, + "output_text_mask_loss": 0.006, + "predictor_learning_rate": 9.423101523941714e-06, + "step": 8930 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1216, + "image_masks_loss": 0.0152, + "llm_learning_rate": 9.400552231406596e-08, + "loss": 0.736, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 9.400552231406596e-06, + "step": 8931 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1216, + "image_masks_loss": 0.0027, + "llm_learning_rate": 9.378029434218588e-08, + "loss": 0.7747, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 9.378029434218588e-06, + "step": 8932 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1216, + "image_masks_loss": 0.0059, + "llm_learning_rate": 9.35553313485793e-08, + "loss": 0.7278, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 9.35553313485793e-06, + "step": 8933 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1215, + "image_masks_loss": 0.0078, + "llm_learning_rate": 9.333063335801945e-08, + "loss": 0.7435, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 9.333063335801944e-06, + "step": 8934 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1215, + "image_masks_loss": 0.0032, + "llm_learning_rate": 9.310620039525126e-08, + "loss": 0.6406, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 9.310620039525125e-06, + "step": 8935 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1215, + "image_masks_loss": 0.0105, + "llm_learning_rate": 9.288203248498967e-08, + "loss": 0.7356, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 9.288203248498967e-06, + "step": 8936 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1215, + "image_masks_loss": 0.0029, + "llm_learning_rate": 9.265812965192023e-08, + "loss": 0.7792, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 9.265812965192022e-06, + "step": 8937 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1214, + "image_masks_loss": 0.0082, + "llm_learning_rate": 9.243449192069931e-08, + "loss": 0.6704, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.243449192069931e-06, + "step": 8938 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1214, + "image_masks_loss": 0.0202, + "llm_learning_rate": 9.221111931595472e-08, + "loss": 0.7393, + "output_text_mask_loss": 0.0215, + "predictor_learning_rate": 9.221111931595473e-06, + "step": 8939 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1214, + "image_masks_loss": 0.0095, + "llm_learning_rate": 9.198801186228512e-08, + "loss": 0.7763, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 9.198801186228511e-06, + "step": 8940 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1213, + "image_masks_loss": 0.0124, + "llm_learning_rate": 9.176516958425918e-08, + "loss": 0.8728, + "output_text_mask_loss": 0.0042, + "predictor_learning_rate": 9.176516958425918e-06, + "step": 8941 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1213, + "image_masks_loss": 0.0127, + "llm_learning_rate": 9.154259250641701e-08, + "loss": 0.6856, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 9.1542592506417e-06, + "step": 8942 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1213, + "image_masks_loss": 0.0118, + "llm_learning_rate": 9.132028065326926e-08, + "loss": 0.6553, + "output_text_mask_loss": 0.0204, + "predictor_learning_rate": 9.132028065326925e-06, + "step": 8943 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1213, + "image_masks_loss": 0.0153, + "llm_learning_rate": 9.109823404929719e-08, + "loss": 0.6804, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 9.109823404929718e-06, + "step": 8944 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1212, + "image_masks_loss": 0.0198, + "llm_learning_rate": 9.087645271895317e-08, + "loss": 0.7546, + "output_text_mask_loss": 0.0366, + "predictor_learning_rate": 9.087645271895317e-06, + "step": 8945 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1212, + "image_masks_loss": 0.0074, + "llm_learning_rate": 9.06549366866602e-08, + "loss": 0.6267, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 9.065493668666019e-06, + "step": 8946 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1212, + "image_masks_loss": 0.0284, + "llm_learning_rate": 9.043368597681234e-08, + "loss": 0.6369, + "output_text_mask_loss": 0.0024, + "predictor_learning_rate": 9.043368597681235e-06, + "step": 8947 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1211, + "image_masks_loss": 0.0072, + "llm_learning_rate": 9.021270061377402e-08, + "loss": 0.7434, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 9.0212700613774e-06, + "step": 8948 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1211, + "image_masks_loss": 0.0109, + "llm_learning_rate": 8.999198062188075e-08, + "loss": 0.8181, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 8.999198062188074e-06, + "step": 8949 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1211, + "image_masks_loss": 0.0099, + "llm_learning_rate": 8.977152602543866e-08, + "loss": 0.688, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 8.977152602543865e-06, + "step": 8950 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1211, + "image_masks_loss": 0.0192, + "llm_learning_rate": 8.955133684872469e-08, + "loss": 0.7099, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 8.95513368487247e-06, + "step": 8951 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.121, + "image_masks_loss": 0.006, + "llm_learning_rate": 8.933141311598669e-08, + "loss": 0.7287, + "output_text_mask_loss": 0.0282, + "predictor_learning_rate": 8.933141311598669e-06, + "step": 8952 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.121, + "image_masks_loss": 0.0044, + "llm_learning_rate": 8.91117548514428e-08, + "loss": 0.6992, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.911175485144279e-06, + "step": 8953 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.121, + "image_masks_loss": 0.0079, + "llm_learning_rate": 8.889236207928254e-08, + "loss": 0.7302, + "output_text_mask_loss": 0.0068, + "predictor_learning_rate": 8.889236207928253e-06, + "step": 8954 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1209, + "image_masks_loss": 0.0073, + "llm_learning_rate": 8.86732348236663e-08, + "loss": 0.7484, + "output_text_mask_loss": 0.0408, + "predictor_learning_rate": 8.867323482366629e-06, + "step": 8955 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1209, + "image_masks_loss": 0.0209, + "llm_learning_rate": 8.845437310872423e-08, + "loss": 0.6774, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.845437310872423e-06, + "step": 8956 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1209, + "image_masks_loss": 0.0067, + "llm_learning_rate": 8.823577695855789e-08, + "loss": 0.706, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 8.82357769585579e-06, + "step": 8957 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1209, + "image_masks_loss": 0.0063, + "llm_learning_rate": 8.801744639723997e-08, + "loss": 0.6792, + "output_text_mask_loss": 0.0391, + "predictor_learning_rate": 8.801744639723996e-06, + "step": 8958 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1208, + "image_masks_loss": 0.0109, + "llm_learning_rate": 8.779938144881318e-08, + "loss": 0.7813, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 8.779938144881316e-06, + "step": 8959 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1208, + "image_masks_loss": 0.0158, + "llm_learning_rate": 8.758158213729162e-08, + "loss": 0.6738, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 8.758158213729161e-06, + "step": 8960 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1208, + "image_masks_loss": 0.0238, + "llm_learning_rate": 8.736404848665947e-08, + "loss": 0.7324, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 8.736404848665946e-06, + "step": 8961 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1207, + "image_masks_loss": 0.0071, + "llm_learning_rate": 8.714678052087255e-08, + "loss": 0.7033, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 8.714678052087256e-06, + "step": 8962 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1207, + "image_masks_loss": 0.0107, + "llm_learning_rate": 8.692977826385645e-08, + "loss": 0.619, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.692977826385645e-06, + "step": 8963 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1207, + "image_masks_loss": 0.0051, + "llm_learning_rate": 8.671304173950789e-08, + "loss": 0.7645, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 8.67130417395079e-06, + "step": 8964 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1207, + "image_masks_loss": 0.0034, + "llm_learning_rate": 8.649657097169501e-08, + "loss": 0.7193, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.6496570971695e-06, + "step": 8965 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1206, + "image_masks_loss": 0.0146, + "llm_learning_rate": 8.628036598425482e-08, + "loss": 0.6927, + "output_text_mask_loss": 0.0311, + "predictor_learning_rate": 8.628036598425482e-06, + "step": 8966 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1206, + "image_masks_loss": 0.0073, + "llm_learning_rate": 8.606442680099774e-08, + "loss": 0.6276, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 8.606442680099775e-06, + "step": 8967 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1206, + "image_masks_loss": 0.0033, + "llm_learning_rate": 8.584875344570254e-08, + "loss": 0.6767, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 8.584875344570253e-06, + "step": 8968 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1205, + "image_masks_loss": 0.0085, + "llm_learning_rate": 8.56333459421202e-08, + "loss": 0.7027, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 8.56333459421202e-06, + "step": 8969 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1205, + "image_masks_loss": 0.0075, + "llm_learning_rate": 8.541820431397147e-08, + "loss": 0.6984, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 8.541820431397146e-06, + "step": 8970 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1205, + "image_masks_loss": 0.0115, + "llm_learning_rate": 8.520332858494823e-08, + "loss": 0.7153, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 8.520332858494822e-06, + "step": 8971 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1205, + "image_masks_loss": 0.0044, + "llm_learning_rate": 8.498871877871323e-08, + "loss": 0.7227, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 8.498871877871323e-06, + "step": 8972 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1204, + "image_masks_loss": 0.0116, + "llm_learning_rate": 8.477437491889978e-08, + "loss": 0.7245, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 8.477437491889978e-06, + "step": 8973 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1204, + "image_masks_loss": 0.003, + "llm_learning_rate": 8.456029702911206e-08, + "loss": 0.7336, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 8.456029702911205e-06, + "step": 8974 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1204, + "image_masks_loss": 0.0041, + "llm_learning_rate": 8.434648513292454e-08, + "loss": 0.7696, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 8.434648513292454e-06, + "step": 8975 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1203, + "image_masks_loss": 0.0087, + "llm_learning_rate": 8.413293925388256e-08, + "loss": 0.817, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 8.413293925388254e-06, + "step": 8976 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1203, + "image_masks_loss": 0.0149, + "llm_learning_rate": 8.391965941550257e-08, + "loss": 0.7361, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 8.391965941550257e-06, + "step": 8977 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1203, + "image_masks_loss": 0.0078, + "llm_learning_rate": 8.370664564127078e-08, + "loss": 0.7476, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 8.370664564127078e-06, + "step": 8978 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1203, + "image_masks_loss": 0.0153, + "llm_learning_rate": 8.349389795464568e-08, + "loss": 0.6909, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 8.349389795464568e-06, + "step": 8979 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1202, + "image_masks_loss": 0.0103, + "llm_learning_rate": 8.328141637905517e-08, + "loss": 0.7545, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.328141637905518e-06, + "step": 8980 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1202, + "image_masks_loss": 0.0064, + "llm_learning_rate": 8.306920093789805e-08, + "loss": 0.7408, + "output_text_mask_loss": 0.0357, + "predictor_learning_rate": 8.306920093789805e-06, + "step": 8981 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1202, + "image_masks_loss": 0.0144, + "llm_learning_rate": 8.285725165454423e-08, + "loss": 0.7107, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 8.285725165454421e-06, + "step": 8982 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1201, + "image_masks_loss": 0.0066, + "llm_learning_rate": 8.264556855233363e-08, + "loss": 0.7417, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 8.264556855233363e-06, + "step": 8983 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1201, + "image_masks_loss": 0.0094, + "llm_learning_rate": 8.243415165457736e-08, + "loss": 0.6777, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 8.243415165457734e-06, + "step": 8984 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1201, + "image_masks_loss": 0.0197, + "llm_learning_rate": 8.222300098455732e-08, + "loss": 0.7587, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 8.222300098455731e-06, + "step": 8985 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1201, + "image_masks_loss": 0.0079, + "llm_learning_rate": 8.201211656552577e-08, + "loss": 0.638, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 8.201211656552577e-06, + "step": 8986 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.12, + "image_masks_loss": 0.0064, + "llm_learning_rate": 8.180149842070578e-08, + "loss": 0.7858, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 8.180149842070578e-06, + "step": 8987 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.12, + "image_masks_loss": 0.0036, + "llm_learning_rate": 8.159114657329131e-08, + "loss": 0.7422, + "output_text_mask_loss": 0.1007, + "predictor_learning_rate": 8.15911465732913e-06, + "step": 8988 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.12, + "image_masks_loss": 0.0216, + "llm_learning_rate": 8.138106104644633e-08, + "loss": 0.6936, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.138106104644633e-06, + "step": 8989 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1199, + "image_masks_loss": 0.01, + "llm_learning_rate": 8.11712418633065e-08, + "loss": 0.7955, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 8.11712418633065e-06, + "step": 8990 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1199, + "image_masks_loss": 0.0108, + "llm_learning_rate": 8.096168904697693e-08, + "loss": 0.6553, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.096168904697693e-06, + "step": 8991 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1199, + "image_masks_loss": 0.0056, + "llm_learning_rate": 8.075240262053447e-08, + "loss": 0.7626, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 8.075240262053445e-06, + "step": 8992 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1199, + "image_masks_loss": 0.0062, + "llm_learning_rate": 8.05433826070262e-08, + "loss": 0.6469, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 8.05433826070262e-06, + "step": 8993 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1198, + "image_masks_loss": 0.0054, + "llm_learning_rate": 8.033462902946959e-08, + "loss": 0.6988, + "output_text_mask_loss": 0.0748, + "predictor_learning_rate": 8.033462902946959e-06, + "step": 8994 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1198, + "image_masks_loss": 0.0066, + "llm_learning_rate": 8.012614191085343e-08, + "loss": 0.7601, + "output_text_mask_loss": 0.0126, + "predictor_learning_rate": 8.012614191085344e-06, + "step": 8995 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1198, + "image_masks_loss": 0.0073, + "llm_learning_rate": 7.991792127413633e-08, + "loss": 0.673, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.991792127413634e-06, + "step": 8996 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1198, + "image_masks_loss": 0.0163, + "llm_learning_rate": 7.970996714224855e-08, + "loss": 0.7627, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 7.970996714224854e-06, + "step": 8997 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1197, + "image_masks_loss": 0.0236, + "llm_learning_rate": 7.950227953808954e-08, + "loss": 0.7048, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 7.950227953808953e-06, + "step": 8998 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1197, + "image_masks_loss": 0.0069, + "llm_learning_rate": 7.929485848453155e-08, + "loss": 0.6516, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 7.929485848453155e-06, + "step": 8999 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1197, + "image_masks_loss": 0.0165, + "llm_learning_rate": 7.908770400441551e-08, + "loss": 0.8558, + "output_text_mask_loss": 0.0278, + "predictor_learning_rate": 7.90877040044155e-06, + "step": 9000 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1196, + "image_masks_loss": 0.0089, + "llm_learning_rate": 7.888081612055399e-08, + "loss": 0.7101, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.888081612055397e-06, + "step": 9001 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1196, + "image_masks_loss": 0.0189, + "llm_learning_rate": 7.867419485572958e-08, + "loss": 0.7136, + "output_text_mask_loss": 0.0192, + "predictor_learning_rate": 7.867419485572957e-06, + "step": 9002 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1196, + "image_masks_loss": 0.0037, + "llm_learning_rate": 7.846784023269632e-08, + "loss": 0.7195, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.846784023269632e-06, + "step": 9003 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1196, + "image_masks_loss": 0.0094, + "llm_learning_rate": 7.826175227417826e-08, + "loss": 0.6997, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.826175227417826e-06, + "step": 9004 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1195, + "image_masks_loss": 0.0101, + "llm_learning_rate": 7.805593100287001e-08, + "loss": 0.6835, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 7.805593100287001e-06, + "step": 9005 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1195, + "image_masks_loss": 0.0132, + "llm_learning_rate": 7.785037644143733e-08, + "loss": 0.7326, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.785037644143734e-06, + "step": 9006 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1195, + "image_masks_loss": 0.0074, + "llm_learning_rate": 7.764508861251657e-08, + "loss": 0.6782, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.764508861251658e-06, + "step": 9007 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1194, + "image_masks_loss": 0.0233, + "llm_learning_rate": 7.744006753871408e-08, + "loss": 0.7508, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 7.744006753871407e-06, + "step": 9008 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1194, + "image_masks_loss": 0.0094, + "llm_learning_rate": 7.723531324260764e-08, + "loss": 0.74, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.723531324260763e-06, + "step": 9009 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1194, + "image_masks_loss": 0.0081, + "llm_learning_rate": 7.703082574674475e-08, + "loss": 0.6463, + "output_text_mask_loss": 0.0786, + "predictor_learning_rate": 7.703082574674475e-06, + "step": 9010 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1194, + "image_masks_loss": 0.0092, + "llm_learning_rate": 7.682660507364437e-08, + "loss": 0.8038, + "output_text_mask_loss": 0.028, + "predictor_learning_rate": 7.682660507364437e-06, + "step": 9011 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1193, + "image_masks_loss": 0.0175, + "llm_learning_rate": 7.662265124579572e-08, + "loss": 0.6606, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 7.66226512457957e-06, + "step": 9012 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1193, + "image_masks_loss": 0.0161, + "llm_learning_rate": 7.641896428565832e-08, + "loss": 0.6672, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 7.641896428565831e-06, + "step": 9013 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1193, + "image_masks_loss": 0.0068, + "llm_learning_rate": 7.621554421566312e-08, + "loss": 0.6858, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.621554421566312e-06, + "step": 9014 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1192, + "image_masks_loss": 0.0074, + "llm_learning_rate": 7.601239105821084e-08, + "loss": 0.8153, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 7.601239105821084e-06, + "step": 9015 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1192, + "image_masks_loss": 0.0143, + "llm_learning_rate": 7.580950483567329e-08, + "loss": 0.7663, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 7.580950483567328e-06, + "step": 9016 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1192, + "image_masks_loss": 0.0056, + "llm_learning_rate": 7.560688557039259e-08, + "loss": 0.7882, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 7.5606885570392585e-06, + "step": 9017 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1192, + "image_masks_loss": 0.023, + "llm_learning_rate": 7.5404533284682e-08, + "loss": 0.6748, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 7.5404533284682e-06, + "step": 9018 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1191, + "image_masks_loss": 0.0051, + "llm_learning_rate": 7.520244800082482e-08, + "loss": 0.7816, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 7.520244800082482e-06, + "step": 9019 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1191, + "image_masks_loss": 0.0077, + "llm_learning_rate": 7.500062974107491e-08, + "loss": 0.6479, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.500062974107491e-06, + "step": 9020 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1191, + "image_masks_loss": 0.0115, + "llm_learning_rate": 7.479907852765728e-08, + "loss": 0.7706, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 7.479907852765727e-06, + "step": 9021 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.119, + "image_masks_loss": 0.0049, + "llm_learning_rate": 7.459779438276693e-08, + "loss": 0.7145, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 7.459779438276693e-06, + "step": 9022 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.119, + "image_masks_loss": 0.0134, + "llm_learning_rate": 7.439677732857004e-08, + "loss": 0.8398, + "output_text_mask_loss": 0.0014, + "predictor_learning_rate": 7.439677732857003e-06, + "step": 9023 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.119, + "image_masks_loss": 0.0117, + "llm_learning_rate": 7.41960273872025e-08, + "loss": 0.7062, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 7.4196027387202495e-06, + "step": 9024 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.119, + "image_masks_loss": 0.0094, + "llm_learning_rate": 7.399554458077191e-08, + "loss": 0.6988, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 7.39955445807719e-06, + "step": 9025 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1189, + "image_masks_loss": 0.008, + "llm_learning_rate": 7.37953289313556e-08, + "loss": 0.7492, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 7.37953289313556e-06, + "step": 9026 + }, + { + "epoch": 0.92, + "gumbel_tau": 0.1189, + "image_masks_loss": 0.0063, + "llm_learning_rate": 7.359538046100207e-08, + "loss": 0.6897, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.359538046100206e-06, + "step": 9027 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1189, + "image_masks_loss": 0.0181, + "llm_learning_rate": 7.33956991917295e-08, + "loss": 0.7751, + "output_text_mask_loss": 0.0319, + "predictor_learning_rate": 7.33956991917295e-06, + "step": 9028 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1189, + "image_masks_loss": 0.0028, + "llm_learning_rate": 7.319628514552813e-08, + "loss": 0.7067, + "output_text_mask_loss": 0.0338, + "predictor_learning_rate": 7.319628514552812e-06, + "step": 9029 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1188, + "image_masks_loss": 0.0135, + "llm_learning_rate": 7.299713834435673e-08, + "loss": 0.7289, + "output_text_mask_loss": 0.0402, + "predictor_learning_rate": 7.299713834435673e-06, + "step": 9030 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1188, + "image_masks_loss": 0.0073, + "llm_learning_rate": 7.279825881014641e-08, + "loss": 0.6982, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 7.2798258810146405e-06, + "step": 9031 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1188, + "image_masks_loss": 0.0055, + "llm_learning_rate": 7.259964656479851e-08, + "loss": 0.7108, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 7.259964656479851e-06, + "step": 9032 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1187, + "image_masks_loss": 0.0172, + "llm_learning_rate": 7.240130163018443e-08, + "loss": 0.6489, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 7.240130163018444e-06, + "step": 9033 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1187, + "image_masks_loss": 0.0096, + "llm_learning_rate": 7.220322402814589e-08, + "loss": 0.7793, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 7.220322402814589e-06, + "step": 9034 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1187, + "image_masks_loss": 0.0034, + "llm_learning_rate": 7.200541378049625e-08, + "loss": 0.7443, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 7.200541378049624e-06, + "step": 9035 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1187, + "image_masks_loss": 0.0147, + "llm_learning_rate": 7.180787090901864e-08, + "loss": 0.6049, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 7.180787090901863e-06, + "step": 9036 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1186, + "image_masks_loss": 0.0127, + "llm_learning_rate": 7.16105954354665e-08, + "loss": 0.7075, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 7.16105954354665e-06, + "step": 9037 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1186, + "image_masks_loss": 0.0091, + "llm_learning_rate": 7.141358738156495e-08, + "loss": 0.7168, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 7.141358738156495e-06, + "step": 9038 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1186, + "image_masks_loss": 0.0046, + "llm_learning_rate": 7.121684676900858e-08, + "loss": 0.6822, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 7.121684676900858e-06, + "step": 9039 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1185, + "image_masks_loss": 0.0023, + "llm_learning_rate": 7.102037361946312e-08, + "loss": 0.7399, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 7.102037361946312e-06, + "step": 9040 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1185, + "image_masks_loss": 0.0096, + "llm_learning_rate": 7.082416795456432e-08, + "loss": 0.6829, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 7.0824167954564315e-06, + "step": 9041 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1185, + "image_masks_loss": 0.0036, + "llm_learning_rate": 7.062822979591905e-08, + "loss": 0.7059, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 7.062822979591904e-06, + "step": 9042 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1185, + "image_masks_loss": 0.0165, + "llm_learning_rate": 7.043255916510395e-08, + "loss": 0.7864, + "output_text_mask_loss": 0.0077, + "predictor_learning_rate": 7.043255916510394e-06, + "step": 9043 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1184, + "image_masks_loss": 0.0025, + "llm_learning_rate": 7.023715608366705e-08, + "loss": 0.6694, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.023715608366704e-06, + "step": 9044 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1184, + "image_masks_loss": 0.0057, + "llm_learning_rate": 7.004202057312698e-08, + "loss": 0.7028, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 7.004202057312697e-06, + "step": 9045 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1184, + "image_masks_loss": 0.0096, + "llm_learning_rate": 6.98471526549721e-08, + "loss": 0.7193, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 6.98471526549721e-06, + "step": 9046 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1183, + "image_masks_loss": 0.008, + "llm_learning_rate": 6.965255235066192e-08, + "loss": 0.7558, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.965255235066193e-06, + "step": 9047 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1183, + "image_masks_loss": 0.0083, + "llm_learning_rate": 6.945821968162569e-08, + "loss": 0.7642, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 6.945821968162569e-06, + "step": 9048 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1183, + "image_masks_loss": 0.0081, + "llm_learning_rate": 6.926415466926406e-08, + "loss": 0.6933, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 6.926415466926405e-06, + "step": 9049 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1183, + "image_masks_loss": 0.0062, + "llm_learning_rate": 6.907035733494799e-08, + "loss": 0.7925, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.9070357334947986e-06, + "step": 9050 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1182, + "image_masks_loss": 0.0134, + "llm_learning_rate": 6.887682770001902e-08, + "loss": 0.7723, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.887682770001902e-06, + "step": 9051 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1182, + "image_masks_loss": 0.0124, + "llm_learning_rate": 6.8683565785789e-08, + "loss": 0.8105, + "output_text_mask_loss": 0.0694, + "predictor_learning_rate": 6.8683565785789e-06, + "step": 9052 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1182, + "image_masks_loss": 0.0112, + "llm_learning_rate": 6.849057161354006e-08, + "loss": 0.7389, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.849057161354006e-06, + "step": 9053 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1182, + "image_masks_loss": 0.0078, + "llm_learning_rate": 6.829784520452548e-08, + "loss": 0.7765, + "output_text_mask_loss": 0.0216, + "predictor_learning_rate": 6.829784520452548e-06, + "step": 9054 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1181, + "image_masks_loss": 0.0103, + "llm_learning_rate": 6.810538657996857e-08, + "loss": 0.739, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.810538657996856e-06, + "step": 9055 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1181, + "image_masks_loss": 0.0078, + "llm_learning_rate": 6.791319576106293e-08, + "loss": 0.7238, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.791319576106292e-06, + "step": 9056 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1181, + "image_masks_loss": 0.0139, + "llm_learning_rate": 6.772127276897383e-08, + "loss": 0.7779, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 6.772127276897383e-06, + "step": 9057 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.118, + "image_masks_loss": 0.0105, + "llm_learning_rate": 6.752961762483606e-08, + "loss": 0.7557, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 6.752961762483606e-06, + "step": 9058 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.118, + "image_masks_loss": 0.0106, + "llm_learning_rate": 6.733823034975468e-08, + "loss": 0.6367, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 6.733823034975467e-06, + "step": 9059 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.118, + "image_masks_loss": 0.0079, + "llm_learning_rate": 6.714711096480587e-08, + "loss": 0.7314, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.714711096480586e-06, + "step": 9060 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.118, + "image_masks_loss": 0.0089, + "llm_learning_rate": 6.695625949103612e-08, + "loss": 0.6151, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.695625949103612e-06, + "step": 9061 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1179, + "image_masks_loss": 0.0134, + "llm_learning_rate": 6.676567594946254e-08, + "loss": 0.6742, + "output_text_mask_loss": 0.0151, + "predictor_learning_rate": 6.676567594946253e-06, + "step": 9062 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1179, + "image_masks_loss": 0.0103, + "llm_learning_rate": 6.657536036107248e-08, + "loss": 0.6795, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 6.657536036107248e-06, + "step": 9063 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1179, + "image_masks_loss": 0.0149, + "llm_learning_rate": 6.638531274682391e-08, + "loss": 0.7219, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 6.6385312746823915e-06, + "step": 9064 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1178, + "image_masks_loss": 0.0047, + "llm_learning_rate": 6.619553312764565e-08, + "loss": 0.7312, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.619553312764564e-06, + "step": 9065 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1178, + "image_masks_loss": 0.0062, + "llm_learning_rate": 6.600602152443653e-08, + "loss": 0.6687, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 6.600602152443652e-06, + "step": 9066 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1178, + "image_masks_loss": 0.0213, + "llm_learning_rate": 6.58167779580654e-08, + "loss": 0.7372, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 6.58167779580654e-06, + "step": 9067 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1178, + "image_masks_loss": 0.0213, + "llm_learning_rate": 6.562780244937284e-08, + "loss": 0.6767, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.5627802449372835e-06, + "step": 9068 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1177, + "image_masks_loss": 0.0131, + "llm_learning_rate": 6.543909501916856e-08, + "loss": 0.6717, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 6.543909501916856e-06, + "step": 9069 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1177, + "image_masks_loss": 0.0115, + "llm_learning_rate": 6.525065568823429e-08, + "loss": 0.8235, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 6.5250655688234296e-06, + "step": 9070 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1177, + "image_masks_loss": 0.0129, + "llm_learning_rate": 6.506248447732095e-08, + "loss": 0.7845, + "output_text_mask_loss": 0.0229, + "predictor_learning_rate": 6.506248447732094e-06, + "step": 9071 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1177, + "image_masks_loss": 0.0124, + "llm_learning_rate": 6.487458140715053e-08, + "loss": 0.7427, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 6.487458140715052e-06, + "step": 9072 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1176, + "image_masks_loss": 0.0075, + "llm_learning_rate": 6.468694649841539e-08, + "loss": 0.6919, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 6.468694649841539e-06, + "step": 9073 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1176, + "image_masks_loss": 0.0101, + "llm_learning_rate": 6.44995797717779e-08, + "loss": 0.7036, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.449957977177789e-06, + "step": 9074 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1176, + "image_masks_loss": 0.0049, + "llm_learning_rate": 6.431248124787154e-08, + "loss": 0.6943, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 6.431248124787154e-06, + "step": 9075 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1175, + "image_masks_loss": 0.0195, + "llm_learning_rate": 6.412565094729984e-08, + "loss": 0.7596, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 6.412565094729983e-06, + "step": 9076 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1175, + "image_masks_loss": 0.0204, + "llm_learning_rate": 6.393908889063744e-08, + "loss": 0.5465, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 6.393908889063743e-06, + "step": 9077 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1175, + "image_masks_loss": 0.0307, + "llm_learning_rate": 6.375279509842847e-08, + "loss": 0.7336, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 6.375279509842846e-06, + "step": 9078 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1175, + "image_masks_loss": 0.0045, + "llm_learning_rate": 6.356676959118846e-08, + "loss": 0.7602, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 6.356676959118846e-06, + "step": 9079 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1174, + "image_masks_loss": 0.0041, + "llm_learning_rate": 6.338101238940269e-08, + "loss": 0.7329, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 6.338101238940269e-06, + "step": 9080 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1174, + "image_masks_loss": 0.0072, + "llm_learning_rate": 6.319552351352731e-08, + "loss": 0.7585, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 6.319552351352731e-06, + "step": 9081 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1174, + "image_masks_loss": 0.028, + "llm_learning_rate": 6.301030298398846e-08, + "loss": 0.7045, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 6.3010302983988465e-06, + "step": 9082 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1173, + "image_masks_loss": 0.0076, + "llm_learning_rate": 6.282535082118291e-08, + "loss": 0.7759, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 6.28253508211829e-06, + "step": 9083 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1173, + "image_masks_loss": 0.0103, + "llm_learning_rate": 6.264066704547906e-08, + "loss": 0.7967, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 6.264066704547905e-06, + "step": 9084 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1173, + "image_masks_loss": 0.0107, + "llm_learning_rate": 6.245625167721341e-08, + "loss": 0.678, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.245625167721341e-06, + "step": 9085 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1173, + "image_masks_loss": 0.0199, + "llm_learning_rate": 6.227210473669504e-08, + "loss": 0.7975, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 6.227210473669504e-06, + "step": 9086 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1172, + "image_masks_loss": 0.0099, + "llm_learning_rate": 6.208822624420213e-08, + "loss": 0.7218, + "output_text_mask_loss": 0.0016, + "predictor_learning_rate": 6.208822624420213e-06, + "step": 9087 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1172, + "image_masks_loss": 0.0057, + "llm_learning_rate": 6.190461621998406e-08, + "loss": 0.6292, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.190461621998405e-06, + "step": 9088 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1172, + "image_masks_loss": 0.0092, + "llm_learning_rate": 6.17212746842602e-08, + "loss": 0.7026, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 6.17212746842602e-06, + "step": 9089 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1172, + "image_masks_loss": 0.0181, + "llm_learning_rate": 6.153820165722052e-08, + "loss": 0.7175, + "output_text_mask_loss": 0.0294, + "predictor_learning_rate": 6.153820165722052e-06, + "step": 9090 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1171, + "image_masks_loss": 0.0292, + "llm_learning_rate": 6.135539715902584e-08, + "loss": 0.7378, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 6.135539715902582e-06, + "step": 9091 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1171, + "image_masks_loss": 0.0075, + "llm_learning_rate": 6.11728612098067e-08, + "loss": 0.8, + "output_text_mask_loss": 0.0243, + "predictor_learning_rate": 6.11728612098067e-06, + "step": 9092 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1171, + "image_masks_loss": 0.0036, + "llm_learning_rate": 6.099059382966399e-08, + "loss": 0.7141, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.0990593829663985e-06, + "step": 9093 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.117, + "image_masks_loss": 0.0185, + "llm_learning_rate": 6.080859503867026e-08, + "loss": 0.7929, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.080859503867025e-06, + "step": 9094 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.117, + "image_masks_loss": 0.0052, + "llm_learning_rate": 6.06268648568667e-08, + "loss": 0.6395, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 6.062686485686669e-06, + "step": 9095 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.117, + "image_masks_loss": 0.0128, + "llm_learning_rate": 6.044540330426617e-08, + "loss": 0.7277, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 6.044540330426618e-06, + "step": 9096 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.117, + "image_masks_loss": 0.0074, + "llm_learning_rate": 6.026421040085217e-08, + "loss": 0.7535, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 6.026421040085217e-06, + "step": 9097 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1169, + "image_masks_loss": 0.0052, + "llm_learning_rate": 6.008328616657733e-08, + "loss": 0.7106, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 6.008328616657732e-06, + "step": 9098 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1169, + "image_masks_loss": 0.0114, + "llm_learning_rate": 5.990263062136598e-08, + "loss": 0.6319, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.990263062136597e-06, + "step": 9099 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1169, + "image_masks_loss": 0.0046, + "llm_learning_rate": 5.972224378511193e-08, + "loss": 0.6641, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 5.972224378511193e-06, + "step": 9100 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1168, + "image_masks_loss": 0.0029, + "llm_learning_rate": 5.954212567767986e-08, + "loss": 0.7259, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 5.954212567767986e-06, + "step": 9101 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1168, + "image_masks_loss": 0.0036, + "llm_learning_rate": 5.936227631890501e-08, + "loss": 0.7935, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 5.936227631890501e-06, + "step": 9102 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1168, + "image_masks_loss": 0.0069, + "llm_learning_rate": 5.9182695728592363e-08, + "loss": 0.7564, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 5.918269572859236e-06, + "step": 9103 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1168, + "image_masks_loss": 0.0139, + "llm_learning_rate": 5.900338392651805e-08, + "loss": 0.8102, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 5.900338392651805e-06, + "step": 9104 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1167, + "image_masks_loss": 0.0047, + "llm_learning_rate": 5.882434093242823e-08, + "loss": 0.8192, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.882434093242822e-06, + "step": 9105 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1167, + "image_masks_loss": 0.0071, + "llm_learning_rate": 5.864556676603933e-08, + "loss": 0.7277, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 5.8645566766039325e-06, + "step": 9106 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1167, + "image_masks_loss": 0.0043, + "llm_learning_rate": 5.846706144703868e-08, + "loss": 0.7496, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 5.846706144703867e-06, + "step": 9107 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1167, + "image_masks_loss": 0.0228, + "llm_learning_rate": 5.828882499508332e-08, + "loss": 0.6322, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.828882499508331e-06, + "step": 9108 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1166, + "image_masks_loss": 0.0141, + "llm_learning_rate": 5.811085742980144e-08, + "loss": 0.7483, + "output_text_mask_loss": 0.0633, + "predictor_learning_rate": 5.811085742980144e-06, + "step": 9109 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1166, + "image_masks_loss": 0.0092, + "llm_learning_rate": 5.793315877079097e-08, + "loss": 0.7024, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.793315877079098e-06, + "step": 9110 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1166, + "image_masks_loss": 0.0185, + "llm_learning_rate": 5.7755729037620724e-08, + "loss": 0.6963, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 5.775572903762072e-06, + "step": 9111 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1165, + "image_masks_loss": 0.0078, + "llm_learning_rate": 5.7578568249829215e-08, + "loss": 0.7156, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 5.757856824982921e-06, + "step": 9112 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1165, + "image_masks_loss": 0.009, + "llm_learning_rate": 5.74016764269264e-08, + "loss": 0.7459, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.74016764269264e-06, + "step": 9113 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1165, + "image_masks_loss": 0.0094, + "llm_learning_rate": 5.722505358839142e-08, + "loss": 0.7653, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 5.722505358839142e-06, + "step": 9114 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1165, + "image_masks_loss": 0.0132, + "llm_learning_rate": 5.7048699753674276e-08, + "loss": 0.6556, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 5.704869975367427e-06, + "step": 9115 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1164, + "image_masks_loss": 0.0109, + "llm_learning_rate": 5.687261494219609e-08, + "loss": 0.7194, + "output_text_mask_loss": 0.0284, + "predictor_learning_rate": 5.6872614942196086e-06, + "step": 9116 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1164, + "image_masks_loss": 0.0263, + "llm_learning_rate": 5.669679917334747e-08, + "loss": 0.6553, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 5.6696799173347466e-06, + "step": 9117 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1164, + "image_masks_loss": 0.0054, + "llm_learning_rate": 5.652125246648959e-08, + "loss": 0.7914, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 5.652125246648959e-06, + "step": 9118 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1164, + "image_masks_loss": 0.012, + "llm_learning_rate": 5.634597484095394e-08, + "loss": 0.7317, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 5.634597484095394e-06, + "step": 9119 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1163, + "image_masks_loss": 0.0046, + "llm_learning_rate": 5.6170966316042574e-08, + "loss": 0.7306, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 5.617096631604257e-06, + "step": 9120 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1163, + "image_masks_loss": 0.0042, + "llm_learning_rate": 5.599622691102785e-08, + "loss": 0.7813, + "output_text_mask_loss": 0.0109, + "predictor_learning_rate": 5.599622691102785e-06, + "step": 9121 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1163, + "image_masks_loss": 0.0142, + "llm_learning_rate": 5.5821756645152435e-08, + "loss": 0.7552, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 5.582175664515243e-06, + "step": 9122 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1162, + "image_masks_loss": 0.0021, + "llm_learning_rate": 5.564755553762929e-08, + "loss": 0.6987, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 5.564755553762929e-06, + "step": 9123 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1162, + "image_masks_loss": 0.0057, + "llm_learning_rate": 5.547362360764197e-08, + "loss": 0.775, + "output_text_mask_loss": 0.0016, + "predictor_learning_rate": 5.547362360764197e-06, + "step": 9124 + }, + { + "epoch": 0.93, + "gumbel_tau": 0.1162, + "image_masks_loss": 0.0055, + "llm_learning_rate": 5.5299960874344306e-08, + "loss": 0.7563, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.52999608743443e-06, + "step": 9125 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1162, + "image_masks_loss": 0.0083, + "llm_learning_rate": 5.5126567356860174e-08, + "loss": 0.7359, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.512656735686017e-06, + "step": 9126 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1161, + "image_masks_loss": 0.0113, + "llm_learning_rate": 5.4953443074284295e-08, + "loss": 0.7028, + "output_text_mask_loss": 0.0599, + "predictor_learning_rate": 5.495344307428429e-06, + "step": 9127 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1161, + "image_masks_loss": 0.0023, + "llm_learning_rate": 5.478058804568115e-08, + "loss": 0.7715, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 5.4780588045681145e-06, + "step": 9128 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1161, + "image_masks_loss": 0.019, + "llm_learning_rate": 5.460800229008634e-08, + "loss": 0.7815, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 5.460800229008633e-06, + "step": 9129 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1161, + "image_masks_loss": 0.0171, + "llm_learning_rate": 5.443568582650494e-08, + "loss": 0.6459, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.443568582650493e-06, + "step": 9130 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.116, + "image_masks_loss": 0.0144, + "llm_learning_rate": 5.4263638673913434e-08, + "loss": 0.703, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.4263638673913425e-06, + "step": 9131 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.116, + "image_masks_loss": 0.0175, + "llm_learning_rate": 5.409186085125778e-08, + "loss": 0.7991, + "output_text_mask_loss": 0.0504, + "predictor_learning_rate": 5.409186085125778e-06, + "step": 9132 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.116, + "image_masks_loss": 0.02, + "llm_learning_rate": 5.392035237745424e-08, + "loss": 0.645, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 5.392035237745424e-06, + "step": 9133 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1159, + "image_masks_loss": 0.0056, + "llm_learning_rate": 5.374911327138993e-08, + "loss": 0.6562, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.374911327138992e-06, + "step": 9134 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1159, + "image_masks_loss": 0.0102, + "llm_learning_rate": 5.357814355192198e-08, + "loss": 0.6904, + "output_text_mask_loss": 0.032, + "predictor_learning_rate": 5.357814355192198e-06, + "step": 9135 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1159, + "image_masks_loss": 0.006, + "llm_learning_rate": 5.340744323787811e-08, + "loss": 0.6827, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 5.3407443237878104e-06, + "step": 9136 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1159, + "image_masks_loss": 0.0198, + "llm_learning_rate": 5.323701234805634e-08, + "loss": 0.7637, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.323701234805634e-06, + "step": 9137 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1158, + "image_masks_loss": 0.0071, + "llm_learning_rate": 5.3066850901224434e-08, + "loss": 0.6789, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 5.306685090122443e-06, + "step": 9138 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1158, + "image_masks_loss": 0.0041, + "llm_learning_rate": 5.289695891612157e-08, + "loss": 0.7185, + "output_text_mask_loss": 0.024, + "predictor_learning_rate": 5.289695891612157e-06, + "step": 9139 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1158, + "image_masks_loss": 0.0079, + "llm_learning_rate": 5.272733641145611e-08, + "loss": 0.6208, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.272733641145611e-06, + "step": 9140 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1158, + "image_masks_loss": 0.0034, + "llm_learning_rate": 5.255798340590756e-08, + "loss": 0.7268, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 5.255798340590756e-06, + "step": 9141 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1157, + "image_masks_loss": 0.0074, + "llm_learning_rate": 5.238889991812518e-08, + "loss": 0.7307, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 5.238889991812518e-06, + "step": 9142 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1157, + "image_masks_loss": 0.0078, + "llm_learning_rate": 5.2220085966729063e-08, + "loss": 0.7547, + "output_text_mask_loss": 0.0072, + "predictor_learning_rate": 5.222008596672906e-06, + "step": 9143 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1157, + "image_masks_loss": 0.0103, + "llm_learning_rate": 5.205154157030934e-08, + "loss": 0.671, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.205154157030934e-06, + "step": 9144 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1156, + "image_masks_loss": 0.0089, + "llm_learning_rate": 5.188326674742644e-08, + "loss": 0.5764, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.188326674742644e-06, + "step": 9145 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1156, + "image_masks_loss": 0.0055, + "llm_learning_rate": 5.17152615166111e-08, + "loss": 0.6811, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 5.17152615166111e-06, + "step": 9146 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1156, + "image_masks_loss": 0.0068, + "llm_learning_rate": 5.154752589636436e-08, + "loss": 0.7599, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 5.154752589636435e-06, + "step": 9147 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1156, + "image_masks_loss": 0.0112, + "llm_learning_rate": 5.13800599051581e-08, + "loss": 0.7866, + "output_text_mask_loss": 0.0797, + "predictor_learning_rate": 5.13800599051581e-06, + "step": 9148 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1155, + "image_masks_loss": 0.0105, + "llm_learning_rate": 5.1212863561433676e-08, + "loss": 0.7738, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 5.121286356143368e-06, + "step": 9149 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1155, + "image_masks_loss": 0.0182, + "llm_learning_rate": 5.1045936883603044e-08, + "loss": 0.6689, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 5.104593688360304e-06, + "step": 9150 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1155, + "image_masks_loss": 0.0133, + "llm_learning_rate": 5.08792798900487e-08, + "loss": 0.7828, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 5.08792798900487e-06, + "step": 9151 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1155, + "image_masks_loss": 0.0074, + "llm_learning_rate": 5.071289259912349e-08, + "loss": 0.6977, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 5.071289259912349e-06, + "step": 9152 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1154, + "image_masks_loss": 0.0022, + "llm_learning_rate": 5.054677502914995e-08, + "loss": 0.814, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.054677502914995e-06, + "step": 9153 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1154, + "image_masks_loss": 0.0099, + "llm_learning_rate": 5.0380927198421246e-08, + "loss": 0.6556, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.038092719842125e-06, + "step": 9154 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1154, + "image_masks_loss": 0.0074, + "llm_learning_rate": 5.0215349125201375e-08, + "loss": 0.7181, + "output_text_mask_loss": 0.0035, + "predictor_learning_rate": 5.021534912520137e-06, + "step": 9155 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1153, + "image_masks_loss": 0.0108, + "llm_learning_rate": 5.00500408277238e-08, + "loss": 0.7284, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.005004082772379e-06, + "step": 9156 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1153, + "image_masks_loss": 0.0044, + "llm_learning_rate": 4.9885002324192855e-08, + "loss": 0.7449, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 4.988500232419285e-06, + "step": 9157 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1153, + "image_masks_loss": 0.0161, + "llm_learning_rate": 4.9720233632783166e-08, + "loss": 0.8, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 4.9720233632783165e-06, + "step": 9158 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1153, + "image_masks_loss": 0.0013, + "llm_learning_rate": 4.9555734771639106e-08, + "loss": 0.7011, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.955573477163911e-06, + "step": 9159 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1152, + "image_masks_loss": 0.0068, + "llm_learning_rate": 4.939150575887508e-08, + "loss": 0.7411, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 4.939150575887508e-06, + "step": 9160 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1152, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.922754661257745e-08, + "loss": 0.7153, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 4.922754661257745e-06, + "step": 9161 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1152, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.9063857350801216e-08, + "loss": 0.7399, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 4.9063857350801215e-06, + "step": 9162 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1152, + "image_masks_loss": 0.0239, + "llm_learning_rate": 4.8900437991572236e-08, + "loss": 0.7195, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 4.890043799157224e-06, + "step": 9163 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1151, + "image_masks_loss": 0.0167, + "llm_learning_rate": 4.87372885528864e-08, + "loss": 0.7062, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 4.87372885528864e-06, + "step": 9164 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1151, + "image_masks_loss": 0.0046, + "llm_learning_rate": 4.857440905271071e-08, + "loss": 0.7286, + "output_text_mask_loss": 0.058, + "predictor_learning_rate": 4.857440905271071e-06, + "step": 9165 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1151, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.8411799508981114e-08, + "loss": 0.7918, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 4.841179950898111e-06, + "step": 9166 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.115, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.824945993960495e-08, + "loss": 0.7014, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 4.824945993960494e-06, + "step": 9167 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.115, + "image_masks_loss": 0.0183, + "llm_learning_rate": 4.80873903624593e-08, + "loss": 0.7425, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 4.808739036245929e-06, + "step": 9168 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.115, + "image_masks_loss": 0.0056, + "llm_learning_rate": 4.792559079539183e-08, + "loss": 0.7945, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 4.792559079539183e-06, + "step": 9169 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.115, + "image_masks_loss": 0.0101, + "llm_learning_rate": 4.776406125621996e-08, + "loss": 0.697, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.7764061256219956e-06, + "step": 9170 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1149, + "image_masks_loss": 0.0178, + "llm_learning_rate": 4.760280176273169e-08, + "loss": 0.6644, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 4.760280176273168e-06, + "step": 9171 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1149, + "image_masks_loss": 0.0074, + "llm_learning_rate": 4.744181233268558e-08, + "loss": 0.6734, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 4.744181233268557e-06, + "step": 9172 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1149, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.728109298380995e-08, + "loss": 0.7534, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.728109298380995e-06, + "step": 9173 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1149, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.712064373380315e-08, + "loss": 0.7953, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 4.712064373380315e-06, + "step": 9174 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1148, + "image_masks_loss": 0.0065, + "llm_learning_rate": 4.696046460033521e-08, + "loss": 0.8046, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 4.6960464600335205e-06, + "step": 9175 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1148, + "image_masks_loss": 0.0096, + "llm_learning_rate": 4.680055560104452e-08, + "loss": 0.731, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 4.680055560104451e-06, + "step": 9176 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1148, + "image_masks_loss": 0.0034, + "llm_learning_rate": 4.6640916753541157e-08, + "loss": 0.5997, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.664091675354115e-06, + "step": 9177 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1147, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.648154807540467e-08, + "loss": 0.7627, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 4.648154807540467e-06, + "step": 9178 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1147, + "image_masks_loss": 0.0127, + "llm_learning_rate": 4.6322449584184916e-08, + "loss": 0.7173, + "output_text_mask_loss": 0.0011, + "predictor_learning_rate": 4.632244958418491e-06, + "step": 9179 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1147, + "image_masks_loss": 0.0169, + "llm_learning_rate": 4.616362129740232e-08, + "loss": 0.6581, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 4.616362129740231e-06, + "step": 9180 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1147, + "image_masks_loss": 0.0122, + "llm_learning_rate": 4.6005063232547606e-08, + "loss": 0.6704, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.600506323254761e-06, + "step": 9181 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1146, + "image_masks_loss": 0.0095, + "llm_learning_rate": 4.5846775407081266e-08, + "loss": 0.7582, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 4.584677540708126e-06, + "step": 9182 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1146, + "image_masks_loss": 0.0103, + "llm_learning_rate": 4.568875783843463e-08, + "loss": 0.7659, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 4.568875783843462e-06, + "step": 9183 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1146, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.553101054400877e-08, + "loss": 0.6877, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 4.5531010544008764e-06, + "step": 9184 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1146, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.5373533541175074e-08, + "loss": 0.7045, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 4.537353354117507e-06, + "step": 9185 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1145, + "image_masks_loss": 0.0046, + "llm_learning_rate": 4.521632684727495e-08, + "loss": 0.689, + "output_text_mask_loss": 0.0435, + "predictor_learning_rate": 4.521632684727495e-06, + "step": 9186 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1145, + "image_masks_loss": 0.0082, + "llm_learning_rate": 4.5059390479621215e-08, + "loss": 0.7226, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 4.505939047962121e-06, + "step": 9187 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1145, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.4902724455495316e-08, + "loss": 0.7118, + "output_text_mask_loss": 0.0603, + "predictor_learning_rate": 4.490272445549531e-06, + "step": 9188 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1144, + "image_masks_loss": 0.0086, + "llm_learning_rate": 4.474632879215013e-08, + "loss": 0.7427, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.474632879215012e-06, + "step": 9189 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1144, + "image_masks_loss": 0.0021, + "llm_learning_rate": 4.459020350680798e-08, + "loss": 0.8011, + "output_text_mask_loss": 0.0288, + "predictor_learning_rate": 4.459020350680798e-06, + "step": 9190 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1144, + "image_masks_loss": 0.014, + "llm_learning_rate": 4.4434348616661795e-08, + "loss": 0.7352, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 4.443434861666179e-06, + "step": 9191 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1144, + "image_masks_loss": 0.0113, + "llm_learning_rate": 4.42787641388745e-08, + "loss": 0.6892, + "output_text_mask_loss": 0.0053, + "predictor_learning_rate": 4.427876413887449e-06, + "step": 9192 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1143, + "image_masks_loss": 0.003, + "llm_learning_rate": 4.4123450090579614e-08, + "loss": 0.7628, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 4.412345009057961e-06, + "step": 9193 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1143, + "image_masks_loss": 0.005, + "llm_learning_rate": 4.396840648888096e-08, + "loss": 0.7262, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.3968406488880955e-06, + "step": 9194 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1143, + "image_masks_loss": 0.0084, + "llm_learning_rate": 4.3813633350852095e-08, + "loss": 0.7094, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 4.381363335085209e-06, + "step": 9195 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1143, + "image_masks_loss": 0.0074, + "llm_learning_rate": 4.365913069353661e-08, + "loss": 0.7799, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.365913069353661e-06, + "step": 9196 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1142, + "image_masks_loss": 0.007, + "llm_learning_rate": 4.3504898533948955e-08, + "loss": 0.7956, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 4.350489853394896e-06, + "step": 9197 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1142, + "image_masks_loss": 0.0042, + "llm_learning_rate": 4.33509368890736e-08, + "loss": 0.7174, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 4.33509368890736e-06, + "step": 9198 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1142, + "image_masks_loss": 0.0037, + "llm_learning_rate": 4.319724577586476e-08, + "loss": 0.6995, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 4.3197245775864755e-06, + "step": 9199 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1142, + "image_masks_loss": 0.0054, + "llm_learning_rate": 4.304382521124778e-08, + "loss": 0.7269, + "output_text_mask_loss": 0.0206, + "predictor_learning_rate": 4.3043825211247775e-06, + "step": 9200 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1141, + "image_masks_loss": 0.0032, + "llm_learning_rate": 4.289067521211748e-08, + "loss": 0.7516, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 4.289067521211748e-06, + "step": 9201 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1141, + "image_masks_loss": 0.0085, + "llm_learning_rate": 4.273779579533899e-08, + "loss": 0.752, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.273779579533899e-06, + "step": 9202 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1141, + "image_masks_loss": 0.0073, + "llm_learning_rate": 4.2585186977747714e-08, + "loss": 0.7331, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.258518697774771e-06, + "step": 9203 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.114, + "image_masks_loss": 0.012, + "llm_learning_rate": 4.243284877614912e-08, + "loss": 0.7395, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 4.243284877614911e-06, + "step": 9204 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.114, + "image_masks_loss": 0.0118, + "llm_learning_rate": 4.2280781207319486e-08, + "loss": 0.7384, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 4.228078120731949e-06, + "step": 9205 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.114, + "image_masks_loss": 0.0132, + "llm_learning_rate": 4.2128984288004326e-08, + "loss": 0.652, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.212898428800432e-06, + "step": 9206 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.114, + "image_masks_loss": 0.0046, + "llm_learning_rate": 4.197745803492026e-08, + "loss": 0.7241, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 4.197745803492026e-06, + "step": 9207 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1139, + "image_masks_loss": 0.0108, + "llm_learning_rate": 4.182620246475366e-08, + "loss": 0.6875, + "output_text_mask_loss": 0.0131, + "predictor_learning_rate": 4.182620246475366e-06, + "step": 9208 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1139, + "image_masks_loss": 0.0059, + "llm_learning_rate": 4.167521759416093e-08, + "loss": 0.8266, + "output_text_mask_loss": 0.0066, + "predictor_learning_rate": 4.167521759416093e-06, + "step": 9209 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1139, + "image_masks_loss": 0.0112, + "llm_learning_rate": 4.152450343976905e-08, + "loss": 0.6968, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 4.152450343976905e-06, + "step": 9210 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1139, + "image_masks_loss": 0.0141, + "llm_learning_rate": 4.1374060018174736e-08, + "loss": 0.6443, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.137406001817473e-06, + "step": 9211 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1138, + "image_masks_loss": 0.0144, + "llm_learning_rate": 4.122388734594529e-08, + "loss": 0.7677, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.122388734594529e-06, + "step": 9212 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1138, + "image_masks_loss": 0.0156, + "llm_learning_rate": 4.107398543961805e-08, + "loss": 0.7184, + "output_text_mask_loss": 0.042, + "predictor_learning_rate": 4.107398543961804e-06, + "step": 9213 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1138, + "image_masks_loss": 0.0053, + "llm_learning_rate": 4.092435431570091e-08, + "loss": 0.7447, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 4.092435431570091e-06, + "step": 9214 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1137, + "image_masks_loss": 0.0034, + "llm_learning_rate": 4.0774993990670975e-08, + "loss": 0.6533, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 4.077499399067097e-06, + "step": 9215 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1137, + "image_masks_loss": 0.0154, + "llm_learning_rate": 4.0625904480976474e-08, + "loss": 0.6642, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 4.062590448097648e-06, + "step": 9216 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1137, + "image_masks_loss": 0.0165, + "llm_learning_rate": 4.0477085803035386e-08, + "loss": 0.6589, + "output_text_mask_loss": 0.0614, + "predictor_learning_rate": 4.047708580303538e-06, + "step": 9217 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1137, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.032853797323599e-08, + "loss": 0.7854, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 4.032853797323599e-06, + "step": 9218 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1136, + "image_masks_loss": 0.0116, + "llm_learning_rate": 4.018026100793659e-08, + "loss": 0.7586, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.018026100793659e-06, + "step": 9219 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1136, + "image_masks_loss": 0.0174, + "llm_learning_rate": 4.0032254923466076e-08, + "loss": 0.7301, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.003225492346607e-06, + "step": 9220 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1136, + "image_masks_loss": 0.0072, + "llm_learning_rate": 3.988451973612278e-08, + "loss": 0.7134, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.988451973612278e-06, + "step": 9221 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1136, + "image_masks_loss": 0.0195, + "llm_learning_rate": 3.973705546217621e-08, + "loss": 0.793, + "output_text_mask_loss": 0.0165, + "predictor_learning_rate": 3.973705546217621e-06, + "step": 9222 + }, + { + "epoch": 0.94, + "gumbel_tau": 0.1135, + "image_masks_loss": 0.0064, + "llm_learning_rate": 3.9589862117864755e-08, + "loss": 0.7362, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.958986211786475e-06, + "step": 9223 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1135, + "image_masks_loss": 0.0126, + "llm_learning_rate": 3.944293971939822e-08, + "loss": 0.6837, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.944293971939822e-06, + "step": 9224 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1135, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.929628828295562e-08, + "loss": 0.6991, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.9296288282955616e-06, + "step": 9225 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1135, + "image_masks_loss": 0.005, + "llm_learning_rate": 3.914990782468708e-08, + "loss": 0.6528, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 3.914990782468708e-06, + "step": 9226 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1134, + "image_masks_loss": 0.0054, + "llm_learning_rate": 3.9003798360711943e-08, + "loss": 0.7761, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 3.900379836071194e-06, + "step": 9227 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1134, + "image_masks_loss": 0.0142, + "llm_learning_rate": 3.88579599071201e-08, + "loss": 0.8142, + "output_text_mask_loss": 0.0345, + "predictor_learning_rate": 3.88579599071201e-06, + "step": 9228 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1134, + "image_masks_loss": 0.0085, + "llm_learning_rate": 3.871239247997177e-08, + "loss": 0.7546, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 3.871239247997177e-06, + "step": 9229 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1133, + "image_masks_loss": 0.0191, + "llm_learning_rate": 3.8567096095296906e-08, + "loss": 0.672, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 3.856709609529691e-06, + "step": 9230 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1133, + "image_masks_loss": 0.0045, + "llm_learning_rate": 3.842207076909632e-08, + "loss": 0.8356, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 3.842207076909632e-06, + "step": 9231 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1133, + "image_masks_loss": 0.006, + "llm_learning_rate": 3.827731651734001e-08, + "loss": 0.8006, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 3.827731651734001e-06, + "step": 9232 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1133, + "image_masks_loss": 0.0131, + "llm_learning_rate": 3.813283335596884e-08, + "loss": 0.6808, + "output_text_mask_loss": 0.061, + "predictor_learning_rate": 3.8132833355968844e-06, + "step": 9233 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1132, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.798862130089398e-08, + "loss": 0.726, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 3.7988621300893978e-06, + "step": 9234 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1132, + "image_masks_loss": 0.0201, + "llm_learning_rate": 3.784468036799577e-08, + "loss": 0.7942, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.7844680367995767e-06, + "step": 9235 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1132, + "image_masks_loss": 0.0094, + "llm_learning_rate": 3.77010105731257e-08, + "loss": 0.6348, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.770101057312569e-06, + "step": 9236 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1132, + "image_masks_loss": 0.0082, + "llm_learning_rate": 3.7557611932104995e-08, + "loss": 0.7441, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 3.7557611932104994e-06, + "step": 9237 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1131, + "image_masks_loss": 0.0075, + "llm_learning_rate": 3.7414484460724655e-08, + "loss": 0.7021, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 3.741448446072465e-06, + "step": 9238 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1131, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.72716281747465e-08, + "loss": 0.6841, + "output_text_mask_loss": 0.036, + "predictor_learning_rate": 3.7271628174746495e-06, + "step": 9239 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1131, + "image_masks_loss": 0.0033, + "llm_learning_rate": 3.7129043089902394e-08, + "loss": 0.6753, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.712904308990239e-06, + "step": 9240 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1131, + "image_masks_loss": 0.0144, + "llm_learning_rate": 3.698672922189395e-08, + "loss": 0.6463, + "output_text_mask_loss": 0.0105, + "predictor_learning_rate": 3.6986729221893945e-06, + "step": 9241 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.113, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.6844686586392786e-08, + "loss": 0.7187, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 3.6844686586392783e-06, + "step": 9242 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.113, + "image_masks_loss": 0.0066, + "llm_learning_rate": 3.67029151990414e-08, + "loss": 0.7131, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 3.67029151990414e-06, + "step": 9243 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.113, + "image_masks_loss": 0.0049, + "llm_learning_rate": 3.6561415075451745e-08, + "loss": 0.7338, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 3.6561415075451745e-06, + "step": 9244 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1129, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.64201862312058e-08, + "loss": 0.7604, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 3.6420186231205797e-06, + "step": 9245 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1129, + "image_masks_loss": 0.0032, + "llm_learning_rate": 3.627922868185668e-08, + "loss": 0.7467, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 3.627922868185668e-06, + "step": 9246 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1129, + "image_masks_loss": 0.0036, + "llm_learning_rate": 3.6138542442926693e-08, + "loss": 0.654, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 3.6138542442926692e-06, + "step": 9247 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1129, + "image_masks_loss": 0.0046, + "llm_learning_rate": 3.5998127529908444e-08, + "loss": 0.709, + "output_text_mask_loss": 0.0174, + "predictor_learning_rate": 3.5998127529908442e-06, + "step": 9248 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1128, + "image_masks_loss": 0.0079, + "llm_learning_rate": 3.5857983958264567e-08, + "loss": 0.8009, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 3.5857983958264562e-06, + "step": 9249 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1128, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.5718111743428266e-08, + "loss": 0.6475, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.571811174342826e-06, + "step": 9250 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1128, + "image_masks_loss": 0.0174, + "llm_learning_rate": 3.55785109008025e-08, + "loss": 0.7709, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.5578510900802497e-06, + "step": 9251 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1128, + "image_masks_loss": 0.0167, + "llm_learning_rate": 3.543918144576025e-08, + "loss": 0.6402, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 3.543918144576025e-06, + "step": 9252 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1127, + "image_masks_loss": 0.0067, + "llm_learning_rate": 3.530012339364508e-08, + "loss": 0.7753, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 3.530012339364508e-06, + "step": 9253 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1127, + "image_masks_loss": 0.0079, + "llm_learning_rate": 3.516133675977001e-08, + "loss": 0.7701, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.5161336759770012e-06, + "step": 9254 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1127, + "image_masks_loss": 0.0252, + "llm_learning_rate": 3.502282155941894e-08, + "loss": 0.775, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.502282155941894e-06, + "step": 9255 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1127, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.488457780784549e-08, + "loss": 0.7248, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.4884577807845485e-06, + "step": 9256 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1126, + "image_masks_loss": 0.0102, + "llm_learning_rate": 3.474660552027276e-08, + "loss": 0.7281, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.4746605520272757e-06, + "step": 9257 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1126, + "image_masks_loss": 0.0046, + "llm_learning_rate": 3.4608904711894986e-08, + "loss": 0.7213, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 3.4608904711894982e-06, + "step": 9258 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1126, + "image_masks_loss": 0.0168, + "llm_learning_rate": 3.447147539787615e-08, + "loss": 0.6934, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 3.447147539787615e-06, + "step": 9259 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1125, + "image_masks_loss": 0.0091, + "llm_learning_rate": 3.4334317593350266e-08, + "loss": 0.7063, + "output_text_mask_loss": 0.028, + "predictor_learning_rate": 3.4334317593350262e-06, + "step": 9260 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1125, + "image_masks_loss": 0.0177, + "llm_learning_rate": 3.419743131342135e-08, + "loss": 0.7366, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 3.4197431313421347e-06, + "step": 9261 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1125, + "image_masks_loss": 0.0054, + "llm_learning_rate": 3.4060816573163734e-08, + "loss": 0.7166, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.4060816573163733e-06, + "step": 9262 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1125, + "image_masks_loss": 0.0066, + "llm_learning_rate": 3.392447338762178e-08, + "loss": 0.6101, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.392447338762178e-06, + "step": 9263 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1124, + "image_masks_loss": 0.0123, + "llm_learning_rate": 3.378840177180931e-08, + "loss": 0.6623, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 3.3788401771809306e-06, + "step": 9264 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1124, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.365260174071156e-08, + "loss": 0.6788, + "output_text_mask_loss": 0.0488, + "predictor_learning_rate": 3.3652601740711554e-06, + "step": 9265 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1124, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.3517073309282945e-08, + "loss": 0.6759, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 3.351707330928294e-06, + "step": 9266 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1124, + "image_masks_loss": 0.0075, + "llm_learning_rate": 3.33818164924482e-08, + "loss": 0.7616, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 3.3381816492448204e-06, + "step": 9267 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1123, + "image_masks_loss": 0.0079, + "llm_learning_rate": 3.324683130510181e-08, + "loss": 0.7529, + "output_text_mask_loss": 0.0469, + "predictor_learning_rate": 3.324683130510181e-06, + "step": 9268 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1123, + "image_masks_loss": 0.0043, + "llm_learning_rate": 3.31121177621091e-08, + "loss": 0.7752, + "output_text_mask_loss": 0.0286, + "predictor_learning_rate": 3.3112117762109096e-06, + "step": 9269 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1123, + "image_masks_loss": 0.0093, + "llm_learning_rate": 3.297767587830458e-08, + "loss": 0.6491, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.297767587830458e-06, + "step": 9270 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1123, + "image_masks_loss": 0.017, + "llm_learning_rate": 3.284350566849337e-08, + "loss": 0.7362, + "output_text_mask_loss": 0.0015, + "predictor_learning_rate": 3.2843505668493368e-06, + "step": 9271 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1122, + "image_masks_loss": 0.0054, + "llm_learning_rate": 3.2709607147450864e-08, + "loss": 0.5654, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 3.270960714745086e-06, + "step": 9272 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1122, + "image_masks_loss": 0.028, + "llm_learning_rate": 3.257598032992193e-08, + "loss": 0.7225, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.257598032992193e-06, + "step": 9273 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1122, + "image_masks_loss": 0.0132, + "llm_learning_rate": 3.244262523062203e-08, + "loss": 0.7675, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 3.2442625230622024e-06, + "step": 9274 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1121, + "image_masks_loss": 0.0072, + "llm_learning_rate": 3.230954186423663e-08, + "loss": 0.7971, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.2309541864236623e-06, + "step": 9275 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1121, + "image_masks_loss": 0.0097, + "llm_learning_rate": 3.217673024542095e-08, + "loss": 0.6964, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 3.217673024542095e-06, + "step": 9276 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1121, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.2044190388800524e-08, + "loss": 0.7619, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 3.204419038880052e-06, + "step": 9277 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1121, + "image_masks_loss": 0.009, + "llm_learning_rate": 3.1911922308971166e-08, + "loss": 0.6821, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.1911922308971163e-06, + "step": 9278 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.112, + "image_masks_loss": 0.0102, + "llm_learning_rate": 3.177992602049818e-08, + "loss": 0.7318, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 3.1779926020498173e-06, + "step": 9279 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.112, + "image_masks_loss": 0.0033, + "llm_learning_rate": 3.16482015379177e-08, + "loss": 0.6955, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.16482015379177e-06, + "step": 9280 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.112, + "image_masks_loss": 0.005, + "llm_learning_rate": 3.151674887573508e-08, + "loss": 0.8466, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 3.151674887573508e-06, + "step": 9281 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.112, + "image_masks_loss": 0.0084, + "llm_learning_rate": 3.1385568048426795e-08, + "loss": 0.6969, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 3.1385568048426795e-06, + "step": 9282 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1119, + "image_masks_loss": 0.0051, + "llm_learning_rate": 3.1254659070437954e-08, + "loss": 0.78, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 3.125465907043795e-06, + "step": 9283 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1119, + "image_masks_loss": 0.0099, + "llm_learning_rate": 3.112402195618508e-08, + "loss": 0.7681, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 3.1124021956185078e-06, + "step": 9284 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1119, + "image_masks_loss": 0.0069, + "llm_learning_rate": 3.099365672005417e-08, + "loss": 0.7317, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 3.0993656720054163e-06, + "step": 9285 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1119, + "image_masks_loss": 0.0125, + "llm_learning_rate": 3.0863563376401226e-08, + "loss": 0.7279, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 3.086356337640123e-06, + "step": 9286 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1118, + "image_masks_loss": 0.0134, + "llm_learning_rate": 3.0733741939552585e-08, + "loss": 0.6944, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.0733741939552583e-06, + "step": 9287 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1118, + "image_masks_loss": 0.0044, + "llm_learning_rate": 3.060419242380458e-08, + "loss": 0.8289, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 3.0604192423804576e-06, + "step": 9288 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1118, + "image_masks_loss": 0.0125, + "llm_learning_rate": 3.047491484342274e-08, + "loss": 0.6693, + "output_text_mask_loss": 0.0086, + "predictor_learning_rate": 3.0474914843422742e-06, + "step": 9289 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1118, + "image_masks_loss": 0.0107, + "llm_learning_rate": 3.03459092126443e-08, + "loss": 0.6774, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.0345909212644294e-06, + "step": 9290 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1117, + "image_masks_loss": 0.0169, + "llm_learning_rate": 3.021717554567511e-08, + "loss": 0.7396, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 3.02171755456751e-06, + "step": 9291 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1117, + "image_masks_loss": 0.0045, + "llm_learning_rate": 3.008871385669187e-08, + "loss": 0.7483, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 3.0088713856691873e-06, + "step": 9292 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1117, + "image_masks_loss": 0.0195, + "llm_learning_rate": 2.996052415984107e-08, + "loss": 0.7091, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.9960524159841073e-06, + "step": 9293 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1116, + "image_masks_loss": 0.0073, + "llm_learning_rate": 2.983260646923891e-08, + "loss": 0.6942, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.983260646923891e-06, + "step": 9294 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1116, + "image_masks_loss": 0.0049, + "llm_learning_rate": 2.970496079897217e-08, + "loss": 0.7359, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.9704960798972167e-06, + "step": 9295 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1116, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.9577587163097666e-08, + "loss": 0.694, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 2.9577587163097663e-06, + "step": 9296 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1116, + "image_masks_loss": 0.0065, + "llm_learning_rate": 2.945048557564195e-08, + "loss": 0.6785, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 2.945048557564195e-06, + "step": 9297 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1115, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.932365605060161e-08, + "loss": 0.7359, + "output_text_mask_loss": 0.0328, + "predictor_learning_rate": 2.932365605060161e-06, + "step": 9298 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1115, + "image_masks_loss": 0.0145, + "llm_learning_rate": 2.919709860194353e-08, + "loss": 0.8141, + "output_text_mask_loss": 0.0082, + "predictor_learning_rate": 2.919709860194353e-06, + "step": 9299 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1115, + "image_masks_loss": 0.0094, + "llm_learning_rate": 2.9070813243604336e-08, + "loss": 0.6539, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 2.9070813243604334e-06, + "step": 9300 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1115, + "image_masks_loss": 0.0065, + "llm_learning_rate": 2.8944799989491234e-08, + "loss": 0.7214, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.8944799989491233e-06, + "step": 9301 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1114, + "image_masks_loss": 0.0053, + "llm_learning_rate": 2.8819058853480632e-08, + "loss": 0.6894, + "output_text_mask_loss": 0.0076, + "predictor_learning_rate": 2.8819058853480627e-06, + "step": 9302 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1114, + "image_masks_loss": 0.0157, + "llm_learning_rate": 2.8693589849419502e-08, + "loss": 0.8297, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 2.86935898494195e-06, + "step": 9303 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1114, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.856839299112485e-08, + "loss": 0.6982, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 2.8568392991124845e-06, + "step": 9304 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1114, + "image_masks_loss": 0.0172, + "llm_learning_rate": 2.844346829238398e-08, + "loss": 0.6867, + "output_text_mask_loss": 0.0755, + "predictor_learning_rate": 2.844346829238398e-06, + "step": 9305 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1113, + "image_masks_loss": 0.0054, + "llm_learning_rate": 2.8318815766953388e-08, + "loss": 0.741, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.8318815766953387e-06, + "step": 9306 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1113, + "image_masks_loss": 0.0085, + "llm_learning_rate": 2.819443542856043e-08, + "loss": 0.7235, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.8194435428560427e-06, + "step": 9307 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1113, + "image_masks_loss": 0.0234, + "llm_learning_rate": 2.8070327290901645e-08, + "loss": 0.6736, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 2.807032729090164e-06, + "step": 9308 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1113, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.7946491367644435e-08, + "loss": 0.7525, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 2.7946491367644433e-06, + "step": 9309 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1112, + "image_masks_loss": 0.0114, + "llm_learning_rate": 2.7822927672426224e-08, + "loss": 0.6763, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.7822927672426226e-06, + "step": 9310 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1112, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.7699636218853355e-08, + "loss": 0.6985, + "output_text_mask_loss": 0.0177, + "predictor_learning_rate": 2.769963621885335e-06, + "step": 9311 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1112, + "image_masks_loss": 0.0081, + "llm_learning_rate": 2.7576617020503848e-08, + "loss": 0.7253, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 2.7576617020503846e-06, + "step": 9312 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1111, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.7453870090924094e-08, + "loss": 0.7319, + "output_text_mask_loss": 0.0486, + "predictor_learning_rate": 2.7453870090924094e-06, + "step": 9313 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1111, + "image_masks_loss": 0.0094, + "llm_learning_rate": 2.7331395443631615e-08, + "loss": 0.6934, + "output_text_mask_loss": 0.0128, + "predictor_learning_rate": 2.733139544363161e-06, + "step": 9314 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1111, + "image_masks_loss": 0.0045, + "llm_learning_rate": 2.7209193092113395e-08, + "loss": 0.7229, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 2.7209193092113395e-06, + "step": 9315 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.1111, + "image_masks_loss": 0.0107, + "llm_learning_rate": 2.7087263049826727e-08, + "loss": 0.6906, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.7087263049826727e-06, + "step": 9316 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.111, + "image_masks_loss": 0.0292, + "llm_learning_rate": 2.696560533019893e-08, + "loss": 0.724, + "output_text_mask_loss": 0.0355, + "predictor_learning_rate": 2.6965605330198926e-06, + "step": 9317 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.111, + "image_masks_loss": 0.0136, + "llm_learning_rate": 2.6844219946627338e-08, + "loss": 0.7647, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.6844219946627334e-06, + "step": 9318 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.111, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.6723106912478757e-08, + "loss": 0.7075, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 2.6723106912478755e-06, + "step": 9319 + }, + { + "epoch": 0.95, + "gumbel_tau": 0.111, + "image_masks_loss": 0.0055, + "llm_learning_rate": 2.6602266241090856e-08, + "loss": 0.6443, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.6602266241090855e-06, + "step": 9320 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1109, + "image_masks_loss": 0.008, + "llm_learning_rate": 2.648169794577049e-08, + "loss": 0.7559, + "output_text_mask_loss": 0.0305, + "predictor_learning_rate": 2.648169794577049e-06, + "step": 9321 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1109, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.6361402039795093e-08, + "loss": 0.7436, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.636140203979509e-06, + "step": 9322 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1109, + "image_masks_loss": 0.005, + "llm_learning_rate": 2.6241378536411842e-08, + "loss": 0.7197, + "output_text_mask_loss": 0.0143, + "predictor_learning_rate": 2.624137853641184e-06, + "step": 9323 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1109, + "image_masks_loss": 0.017, + "llm_learning_rate": 2.6121627448838227e-08, + "loss": 0.716, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 2.6121627448838227e-06, + "step": 9324 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1108, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.600214879026147e-08, + "loss": 0.678, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.600214879026147e-06, + "step": 9325 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1108, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.5882942573838553e-08, + "loss": 0.8117, + "output_text_mask_loss": 0.0051, + "predictor_learning_rate": 2.588294257383855e-06, + "step": 9326 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1108, + "image_masks_loss": 0.0177, + "llm_learning_rate": 2.5764008812697028e-08, + "loss": 0.6775, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 2.5764008812697026e-06, + "step": 9327 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1108, + "image_masks_loss": 0.0076, + "llm_learning_rate": 2.5645347519933917e-08, + "loss": 0.7695, + "output_text_mask_loss": 0.009, + "predictor_learning_rate": 2.5645347519933915e-06, + "step": 9328 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1107, + "image_masks_loss": 0.007, + "llm_learning_rate": 2.5526958708616267e-08, + "loss": 0.7787, + "output_text_mask_loss": 0.0374, + "predictor_learning_rate": 2.5526958708616267e-06, + "step": 9329 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1107, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.5408842391781708e-08, + "loss": 0.8236, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 2.5408842391781706e-06, + "step": 9330 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1107, + "image_masks_loss": 0.012, + "llm_learning_rate": 2.529099858243761e-08, + "loss": 0.7529, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.5290998582437606e-06, + "step": 9331 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1107, + "image_masks_loss": 0.0099, + "llm_learning_rate": 2.5173427293560815e-08, + "loss": 0.7744, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 2.517342729356081e-06, + "step": 9332 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1106, + "image_masks_loss": 0.0222, + "llm_learning_rate": 2.505612853809847e-08, + "loss": 0.7398, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.5056128538098466e-06, + "step": 9333 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1106, + "image_masks_loss": 0.0186, + "llm_learning_rate": 2.493910232896829e-08, + "loss": 0.7469, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.493910232896829e-06, + "step": 9334 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1106, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.4822348679056918e-08, + "loss": 0.6775, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.4822348679056918e-06, + "step": 9335 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1105, + "image_masks_loss": 0.0151, + "llm_learning_rate": 2.4705867601221846e-08, + "loss": 0.6834, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.4705867601221844e-06, + "step": 9336 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1105, + "image_masks_loss": 0.017, + "llm_learning_rate": 2.458965910829003e-08, + "loss": 0.8305, + "output_text_mask_loss": 0.0012, + "predictor_learning_rate": 2.458965910829003e-06, + "step": 9337 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1105, + "image_masks_loss": 0.018, + "llm_learning_rate": 2.4473723213058742e-08, + "loss": 0.7064, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.447372321305874e-06, + "step": 9338 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1105, + "image_masks_loss": 0.0142, + "llm_learning_rate": 2.4358059928295264e-08, + "loss": 0.7345, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 2.4358059928295262e-06, + "step": 9339 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1104, + "image_masks_loss": 0.022, + "llm_learning_rate": 2.424266926673635e-08, + "loss": 0.7119, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.424266926673635e-06, + "step": 9340 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1104, + "image_masks_loss": 0.0062, + "llm_learning_rate": 2.4127551241089342e-08, + "loss": 0.6749, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 2.4127551241089342e-06, + "step": 9341 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1104, + "image_masks_loss": 0.0156, + "llm_learning_rate": 2.401270586403104e-08, + "loss": 0.7816, + "output_text_mask_loss": 0.0492, + "predictor_learning_rate": 2.401270586403104e-06, + "step": 9342 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1104, + "image_masks_loss": 0.0056, + "llm_learning_rate": 2.3898133148208547e-08, + "loss": 0.7731, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 2.3898133148208543e-06, + "step": 9343 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1103, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.3783833106239272e-08, + "loss": 0.7472, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 2.378383310623927e-06, + "step": 9344 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1103, + "image_masks_loss": 0.0052, + "llm_learning_rate": 2.366980575070954e-08, + "loss": 0.8269, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 2.3669805750709537e-06, + "step": 9345 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1103, + "image_masks_loss": 0.0083, + "llm_learning_rate": 2.35560510941768e-08, + "loss": 0.81, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 2.35560510941768e-06, + "step": 9346 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1103, + "image_masks_loss": 0.0056, + "llm_learning_rate": 2.3442569149167705e-08, + "loss": 0.6957, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.3442569149167704e-06, + "step": 9347 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1102, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.3329359928179195e-08, + "loss": 0.8074, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 2.3329359928179194e-06, + "step": 9348 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1102, + "image_masks_loss": 0.0152, + "llm_learning_rate": 2.3216423443677972e-08, + "loss": 0.7769, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 2.321642344367797e-06, + "step": 9349 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1102, + "image_masks_loss": 0.0129, + "llm_learning_rate": 2.3103759708101302e-08, + "loss": 0.6976, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.31037597081013e-06, + "step": 9350 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1102, + "image_masks_loss": 0.0058, + "llm_learning_rate": 2.2991368733855656e-08, + "loss": 0.8144, + "output_text_mask_loss": 0.0687, + "predictor_learning_rate": 2.2991368733855656e-06, + "step": 9351 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1101, + "image_masks_loss": 0.0071, + "llm_learning_rate": 2.287925053331752e-08, + "loss": 0.7324, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 2.2879250533317518e-06, + "step": 9352 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1101, + "image_masks_loss": 0.0143, + "llm_learning_rate": 2.276740511883424e-08, + "loss": 0.7862, + "output_text_mask_loss": 0.0057, + "predictor_learning_rate": 2.276740511883424e-06, + "step": 9353 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1101, + "image_masks_loss": 0.0176, + "llm_learning_rate": 2.2655832502721797e-08, + "loss": 0.7499, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 2.2655832502721796e-06, + "step": 9354 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1101, + "image_masks_loss": 0.0129, + "llm_learning_rate": 2.254453269726731e-08, + "loss": 0.7074, + "output_text_mask_loss": 0.0721, + "predictor_learning_rate": 2.2544532697267305e-06, + "step": 9355 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.11, + "image_masks_loss": 0.0061, + "llm_learning_rate": 2.2433505714727078e-08, + "loss": 0.6822, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 2.2433505714727074e-06, + "step": 9356 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.11, + "image_masks_loss": 0.0155, + "llm_learning_rate": 2.2322751567327995e-08, + "loss": 0.7618, + "output_text_mask_loss": 0.033, + "predictor_learning_rate": 2.2322751567327994e-06, + "step": 9357 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.11, + "image_masks_loss": 0.015, + "llm_learning_rate": 2.2212270267266133e-08, + "loss": 0.7429, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.2212270267266132e-06, + "step": 9358 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1099, + "image_masks_loss": 0.0093, + "llm_learning_rate": 2.2102061826708155e-08, + "loss": 0.7254, + "output_text_mask_loss": 0.0433, + "predictor_learning_rate": 2.210206182670815e-06, + "step": 9359 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1099, + "image_masks_loss": 0.0194, + "llm_learning_rate": 2.1992126257790735e-08, + "loss": 0.7416, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 2.1992126257790734e-06, + "step": 9360 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1099, + "image_masks_loss": 0.0136, + "llm_learning_rate": 2.1882463572619474e-08, + "loss": 0.7242, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 2.188246357261947e-06, + "step": 9361 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1099, + "image_masks_loss": 0.0134, + "llm_learning_rate": 2.177307378327137e-08, + "loss": 0.7381, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.177307378327137e-06, + "step": 9362 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1098, + "image_masks_loss": 0.004, + "llm_learning_rate": 2.166395690179235e-08, + "loss": 0.6858, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 2.1663956901792347e-06, + "step": 9363 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1098, + "image_masks_loss": 0.003, + "llm_learning_rate": 2.155511294019863e-08, + "loss": 0.7571, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 2.155511294019863e-06, + "step": 9364 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1098, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.144654191047646e-08, + "loss": 0.6995, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.144654191047646e-06, + "step": 9365 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1098, + "image_masks_loss": 0.0103, + "llm_learning_rate": 2.1338243824581826e-08, + "loss": 0.7045, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.1338243824581825e-06, + "step": 9366 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1097, + "image_masks_loss": 0.0066, + "llm_learning_rate": 2.123021869444075e-08, + "loss": 0.6752, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 2.123021869444075e-06, + "step": 9367 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1097, + "image_masks_loss": 0.0142, + "llm_learning_rate": 2.112246653194927e-08, + "loss": 0.7147, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.112246653194927e-06, + "step": 9368 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1097, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.101498734897317e-08, + "loss": 0.5722, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 2.101498734897317e-06, + "step": 9369 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1097, + "image_masks_loss": 0.0064, + "llm_learning_rate": 2.090778115734854e-08, + "loss": 0.7527, + "output_text_mask_loss": 0.0422, + "predictor_learning_rate": 2.090778115734854e-06, + "step": 9370 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1096, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.0800847968880933e-08, + "loss": 0.6953, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 2.080084796888093e-06, + "step": 9371 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1096, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.069418779534621e-08, + "loss": 0.7176, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 2.069418779534621e-06, + "step": 9372 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1096, + "image_masks_loss": 0.0113, + "llm_learning_rate": 2.0587800648489974e-08, + "loss": 0.8225, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.0587800648489974e-06, + "step": 9373 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1096, + "image_masks_loss": 0.0092, + "llm_learning_rate": 2.048168654002758e-08, + "loss": 0.7476, + "output_text_mask_loss": 0.0032, + "predictor_learning_rate": 2.0481686540027577e-06, + "step": 9374 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1095, + "image_masks_loss": 0.0047, + "llm_learning_rate": 2.037584548164495e-08, + "loss": 0.7852, + "output_text_mask_loss": 0.0519, + "predictor_learning_rate": 2.037584548164495e-06, + "step": 9375 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1095, + "image_masks_loss": 0.0139, + "llm_learning_rate": 2.0270277484997213e-08, + "loss": 0.7065, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 2.027027748499721e-06, + "step": 9376 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1095, + "image_masks_loss": 0.0091, + "llm_learning_rate": 2.0164982561710068e-08, + "loss": 0.8018, + "output_text_mask_loss": 0.025, + "predictor_learning_rate": 2.0164982561710068e-06, + "step": 9377 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1095, + "image_masks_loss": 0.0034, + "llm_learning_rate": 2.005996072337868e-08, + "loss": 0.7467, + "output_text_mask_loss": 0.0062, + "predictor_learning_rate": 2.005996072337868e-06, + "step": 9378 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1094, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.9955211981568246e-08, + "loss": 0.7256, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.9955211981568246e-06, + "step": 9379 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1094, + "image_masks_loss": 0.0128, + "llm_learning_rate": 1.985073634781398e-08, + "loss": 0.6944, + "output_text_mask_loss": 0.0052, + "predictor_learning_rate": 1.985073634781398e-06, + "step": 9380 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1094, + "image_masks_loss": 0.0216, + "llm_learning_rate": 1.974653383362085e-08, + "loss": 0.7329, + "output_text_mask_loss": 0.0214, + "predictor_learning_rate": 1.974653383362085e-06, + "step": 9381 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1094, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.964260445046412e-08, + "loss": 0.7502, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.964260445046412e-06, + "step": 9382 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1093, + "image_masks_loss": 0.0102, + "llm_learning_rate": 1.9538948209788522e-08, + "loss": 0.7288, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.953894820978852e-06, + "step": 9383 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1093, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.9435565123008814e-08, + "loss": 0.8246, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 1.9435565123008815e-06, + "step": 9384 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1093, + "image_masks_loss": 0.0077, + "llm_learning_rate": 1.9332455201510058e-08, + "loss": 0.7314, + "output_text_mask_loss": 0.0254, + "predictor_learning_rate": 1.9332455201510056e-06, + "step": 9385 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1093, + "image_masks_loss": 0.0164, + "llm_learning_rate": 1.9229618456646505e-08, + "loss": 0.7343, + "output_text_mask_loss": 0.0024, + "predictor_learning_rate": 1.9229618456646503e-06, + "step": 9386 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1092, + "image_masks_loss": 0.0125, + "llm_learning_rate": 1.912705489974326e-08, + "loss": 0.6863, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 1.9127054899743257e-06, + "step": 9387 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1092, + "image_masks_loss": 0.007, + "llm_learning_rate": 1.9024764542094622e-08, + "loss": 0.7832, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.9024764542094619e-06, + "step": 9388 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1092, + "image_masks_loss": 0.0119, + "llm_learning_rate": 1.8922747394965192e-08, + "loss": 0.7445, + "output_text_mask_loss": 0.0301, + "predictor_learning_rate": 1.8922747394965189e-06, + "step": 9389 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1091, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.8821003469589317e-08, + "loss": 0.7245, + "output_text_mask_loss": 0.0084, + "predictor_learning_rate": 1.8821003469589314e-06, + "step": 9390 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1091, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.8719532777170813e-08, + "loss": 0.8129, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 1.8719532777170812e-06, + "step": 9391 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1091, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.861833532888435e-08, + "loss": 0.7495, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 1.861833532888435e-06, + "step": 9392 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1091, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.8517411135874075e-08, + "loss": 0.7142, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 1.8517411135874073e-06, + "step": 9393 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.109, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.8416760209253314e-08, + "loss": 0.7389, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 1.8416760209253313e-06, + "step": 9394 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.109, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.8316382560106816e-08, + "loss": 0.7524, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.8316382560106815e-06, + "step": 9395 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.109, + "image_masks_loss": 0.0031, + "llm_learning_rate": 1.821627819948768e-08, + "loss": 0.7219, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.8216278199487678e-06, + "step": 9396 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.109, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.8116447138420422e-08, + "loss": 0.552, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.8116447138420423e-06, + "step": 9397 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1089, + "image_masks_loss": 0.0115, + "llm_learning_rate": 1.801688938789764e-08, + "loss": 0.7849, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 1.801688938789764e-06, + "step": 9398 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1089, + "image_masks_loss": 0.0116, + "llm_learning_rate": 1.7917604958883894e-08, + "loss": 0.676, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.7917604958883892e-06, + "step": 9399 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1089, + "image_masks_loss": 0.0026, + "llm_learning_rate": 1.781859386231183e-08, + "loss": 0.7555, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.7818593862311827e-06, + "step": 9400 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1089, + "image_masks_loss": 0.0165, + "llm_learning_rate": 1.7719856109084943e-08, + "loss": 0.7484, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 1.7719856109084941e-06, + "step": 9401 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1088, + "image_masks_loss": 0.0025, + "llm_learning_rate": 1.7621391710076764e-08, + "loss": 0.7133, + "output_text_mask_loss": 0.0618, + "predictor_learning_rate": 1.7621391710076762e-06, + "step": 9402 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1088, + "image_masks_loss": 0.0233, + "llm_learning_rate": 1.752320067613028e-08, + "loss": 0.6871, + "output_text_mask_loss": 0.0071, + "predictor_learning_rate": 1.752320067613028e-06, + "step": 9403 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1088, + "image_masks_loss": 0.0282, + "llm_learning_rate": 1.7425283018058236e-08, + "loss": 0.6992, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 1.7425283018058236e-06, + "step": 9404 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1088, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.7327638746643947e-08, + "loss": 0.7169, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.7327638746643948e-06, + "step": 9405 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1087, + "image_masks_loss": 0.0023, + "llm_learning_rate": 1.7230267872639926e-08, + "loss": 0.6933, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.7230267872639926e-06, + "step": 9406 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1087, + "image_masks_loss": 0.0202, + "llm_learning_rate": 1.7133170406768706e-08, + "loss": 0.7403, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.7133170406768705e-06, + "step": 9407 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1087, + "image_masks_loss": 0.0105, + "llm_learning_rate": 1.70363463597234e-08, + "loss": 0.6804, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.7036346359723399e-06, + "step": 9408 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1087, + "image_masks_loss": 0.0186, + "llm_learning_rate": 1.693979574216631e-08, + "loss": 0.6895, + "output_text_mask_loss": 0.0119, + "predictor_learning_rate": 1.6939795742166308e-06, + "step": 9409 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1086, + "image_masks_loss": 0.005, + "llm_learning_rate": 1.684351856472949e-08, + "loss": 0.7342, + "output_text_mask_loss": 0.0149, + "predictor_learning_rate": 1.6843518564729487e-06, + "step": 9410 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1086, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.6747514838015566e-08, + "loss": 0.745, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.6747514838015565e-06, + "step": 9411 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1086, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.665178457259664e-08, + "loss": 0.7395, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.665178457259664e-06, + "step": 9412 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1086, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.655632777901428e-08, + "loss": 0.7718, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 1.655632777901428e-06, + "step": 9413 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1085, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.646114446778091e-08, + "loss": 0.6292, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.646114446778091e-06, + "step": 9414 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1085, + "image_masks_loss": 0.0075, + "llm_learning_rate": 1.6366234649378144e-08, + "loss": 0.7803, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.6366234649378143e-06, + "step": 9415 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1085, + "image_masks_loss": 0.0146, + "llm_learning_rate": 1.6271598334257898e-08, + "loss": 0.692, + "output_text_mask_loss": 0.0029, + "predictor_learning_rate": 1.6271598334257899e-06, + "step": 9416 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1085, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.6177235532841284e-08, + "loss": 0.761, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.6177235532841283e-06, + "step": 9417 + }, + { + "epoch": 0.96, + "gumbel_tau": 0.1084, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.6083146255520266e-08, + "loss": 0.719, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.6083146255520265e-06, + "step": 9418 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1084, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.598933051265572e-08, + "loss": 0.725, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 1.598933051265572e-06, + "step": 9419 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1084, + "image_masks_loss": 0.0011, + "llm_learning_rate": 1.5895788314579107e-08, + "loss": 0.7376, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 1.5895788314579106e-06, + "step": 9420 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1084, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.5802519671591355e-08, + "loss": 0.7207, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.5802519671591353e-06, + "step": 9421 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1083, + "image_masks_loss": 0.0216, + "llm_learning_rate": 1.570952459396341e-08, + "loss": 0.7571, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.570952459396341e-06, + "step": 9422 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1083, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.561680309193625e-08, + "loss": 0.7758, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.561680309193625e-06, + "step": 9423 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1083, + "image_masks_loss": 0.0103, + "llm_learning_rate": 1.5524355175720318e-08, + "loss": 0.752, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.5524355175720318e-06, + "step": 9424 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1082, + "image_masks_loss": 0.0112, + "llm_learning_rate": 1.5432180855496636e-08, + "loss": 0.6154, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 1.5432180855496636e-06, + "step": 9425 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1082, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.5340280141415142e-08, + "loss": 0.7076, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.5340280141415142e-06, + "step": 9426 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1082, + "image_masks_loss": 0.0093, + "llm_learning_rate": 1.524865304359635e-08, + "loss": 0.7854, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 1.524865304359635e-06, + "step": 9427 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1082, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.5157299572130522e-08, + "loss": 0.7593, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 1.5157299572130524e-06, + "step": 9428 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1081, + "image_masks_loss": 0.0107, + "llm_learning_rate": 1.506621973707767e-08, + "loss": 0.6983, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.5066219737077668e-06, + "step": 9429 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1081, + "image_masks_loss": 0.0036, + "llm_learning_rate": 1.4975413548467544e-08, + "loss": 0.701, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 1.4975413548467542e-06, + "step": 9430 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1081, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.4884881016300202e-08, + "loss": 0.6713, + "output_text_mask_loss": 0.021, + "predictor_learning_rate": 1.48848810163002e-06, + "step": 9431 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1081, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.4794622150544891e-08, + "loss": 0.6484, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.479462215054489e-06, + "step": 9432 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.108, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.4704636961141439e-08, + "loss": 0.6806, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.4704636961141438e-06, + "step": 9433 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.108, + "image_masks_loss": 0.0035, + "llm_learning_rate": 1.4614925457999418e-08, + "loss": 0.7017, + "output_text_mask_loss": 0.0073, + "predictor_learning_rate": 1.4614925457999418e-06, + "step": 9434 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.108, + "image_masks_loss": 0.0056, + "llm_learning_rate": 1.4525487650997593e-08, + "loss": 0.7486, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.4525487650997593e-06, + "step": 9435 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.108, + "image_masks_loss": 0.0239, + "llm_learning_rate": 1.443632354998531e-08, + "loss": 0.7689, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.4436323549985308e-06, + "step": 9436 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1079, + "image_masks_loss": 0.0192, + "llm_learning_rate": 1.4347433164781378e-08, + "loss": 0.7009, + "output_text_mask_loss": 0.0303, + "predictor_learning_rate": 1.4347433164781376e-06, + "step": 9437 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1079, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.4258816505174911e-08, + "loss": 0.6675, + "output_text_mask_loss": 0.0326, + "predictor_learning_rate": 1.4258816505174909e-06, + "step": 9438 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1079, + "image_masks_loss": 0.0143, + "llm_learning_rate": 1.4170473580923937e-08, + "loss": 0.6566, + "output_text_mask_loss": 0.0877, + "predictor_learning_rate": 1.4170473580923937e-06, + "step": 9439 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1079, + "image_masks_loss": 0.0052, + "llm_learning_rate": 1.4082404401757621e-08, + "loss": 0.7161, + "output_text_mask_loss": 0.0195, + "predictor_learning_rate": 1.408240440175762e-06, + "step": 9440 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1078, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.3994608977374036e-08, + "loss": 0.774, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 1.3994608977374034e-06, + "step": 9441 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1078, + "image_masks_loss": 0.0223, + "llm_learning_rate": 1.390708731744156e-08, + "loss": 0.6611, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.3907087317441558e-06, + "step": 9442 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1078, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.3819839431598037e-08, + "loss": 0.7271, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.3819839431598035e-06, + "step": 9443 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1078, + "image_masks_loss": 0.01, + "llm_learning_rate": 1.3732865329451617e-08, + "loss": 0.6753, + "output_text_mask_loss": NaN, + "predictor_learning_rate": 1.3732865329451616e-06, + "step": 9444 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1077, + "image_masks_loss": 0.0067, + "llm_learning_rate": 1.3646165020580194e-08, + "loss": 0.7971, + "output_text_mask_loss": 0.0427, + "predictor_learning_rate": 1.3646165020580192e-06, + "step": 9445 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1077, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.3559738514530852e-08, + "loss": 0.7181, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 1.355973851453085e-06, + "step": 9446 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1077, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.3473585820821533e-08, + "loss": 0.7348, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.3473585820821532e-06, + "step": 9447 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1077, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.338770694893965e-08, + "loss": 0.7176, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.338770694893965e-06, + "step": 9448 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1076, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.330210190834208e-08, + "loss": 0.8041, + "output_text_mask_loss": 0.0217, + "predictor_learning_rate": 1.330210190834208e-06, + "step": 9449 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1076, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.3216770708456006e-08, + "loss": 0.7007, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 1.3216770708456004e-06, + "step": 9450 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1076, + "image_masks_loss": 0.0141, + "llm_learning_rate": 1.3131713358678356e-08, + "loss": 0.7401, + "output_text_mask_loss": 0.0187, + "predictor_learning_rate": 1.3131713358678355e-06, + "step": 9451 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1076, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.3046929868375524e-08, + "loss": 0.7128, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 1.3046929868375523e-06, + "step": 9452 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1075, + "image_masks_loss": 0.0038, + "llm_learning_rate": 1.296242024688421e-08, + "loss": 0.7038, + "output_text_mask_loss": 0.0202, + "predictor_learning_rate": 1.2962420246884211e-06, + "step": 9453 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1075, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.287818450351086e-08, + "loss": 0.744, + "output_text_mask_loss": 0.0141, + "predictor_learning_rate": 1.2878184503510858e-06, + "step": 9454 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1075, + "image_masks_loss": 0.02, + "llm_learning_rate": 1.2794222647531663e-08, + "loss": 0.6751, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 1.279422264753166e-06, + "step": 9455 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1075, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.2710534688192833e-08, + "loss": 0.8432, + "output_text_mask_loss": 0.0778, + "predictor_learning_rate": 1.2710534688192833e-06, + "step": 9456 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1074, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.2627120634710055e-08, + "loss": 0.6252, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 1.2627120634710054e-06, + "step": 9457 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1074, + "image_masks_loss": 0.012, + "llm_learning_rate": 1.2543980496269037e-08, + "loss": 0.6782, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 1.2543980496269037e-06, + "step": 9458 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1074, + "image_masks_loss": 0.0087, + "llm_learning_rate": 1.2461114282025789e-08, + "loss": 0.6872, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.2461114282025787e-06, + "step": 9459 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1074, + "image_masks_loss": 0.0153, + "llm_learning_rate": 1.237852200110523e-08, + "loss": 0.711, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.237852200110523e-06, + "step": 9460 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1073, + "image_masks_loss": 0.0211, + "llm_learning_rate": 1.2296203662602869e-08, + "loss": 0.6727, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.2296203662602867e-06, + "step": 9461 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1073, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.2214159275583392e-08, + "loss": 0.6871, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.2214159275583393e-06, + "step": 9462 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1073, + "image_masks_loss": 0.0082, + "llm_learning_rate": 1.2132388849082355e-08, + "loss": 0.7519, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 1.2132388849082354e-06, + "step": 9463 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1073, + "image_masks_loss": 0.0078, + "llm_learning_rate": 1.2050892392104219e-08, + "loss": 0.7578, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 1.2050892392104218e-06, + "step": 9464 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1072, + "image_masks_loss": 0.0134, + "llm_learning_rate": 1.1969669913623194e-08, + "loss": 0.8035, + "output_text_mask_loss": 0.0132, + "predictor_learning_rate": 1.1969669913623193e-06, + "step": 9465 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1072, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.1888721422584071e-08, + "loss": 0.7567, + "output_text_mask_loss": 0.0496, + "predictor_learning_rate": 1.188872142258407e-06, + "step": 9466 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1072, + "image_masks_loss": 0.0044, + "llm_learning_rate": 1.1808046927901107e-08, + "loss": 0.7605, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 1.1808046927901107e-06, + "step": 9467 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1072, + "image_masks_loss": 0.0042, + "llm_learning_rate": 1.1727646438458307e-08, + "loss": 0.6406, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 1.1727646438458306e-06, + "step": 9468 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1071, + "image_masks_loss": 0.0206, + "llm_learning_rate": 1.1647519963109421e-08, + "loss": 0.7933, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 1.1647519963109422e-06, + "step": 9469 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1071, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.1567667510678226e-08, + "loss": 0.8251, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.1567667510678226e-06, + "step": 9470 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1071, + "image_masks_loss": 0.0117, + "llm_learning_rate": 1.1488089089958243e-08, + "loss": 0.6264, + "output_text_mask_loss": 0.0054, + "predictor_learning_rate": 1.1488089089958243e-06, + "step": 9471 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1071, + "image_masks_loss": 0.0041, + "llm_learning_rate": 1.1408784709712739e-08, + "loss": 0.6907, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1408784709712738e-06, + "step": 9472 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.107, + "image_masks_loss": 0.0163, + "llm_learning_rate": 1.1329754378675007e-08, + "loss": 0.6789, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.1329754378675006e-06, + "step": 9473 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.107, + "image_masks_loss": 0.0016, + "llm_learning_rate": 1.1250998105548083e-08, + "loss": 0.7523, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 1.1250998105548083e-06, + "step": 9474 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.107, + "image_masks_loss": 0.0063, + "llm_learning_rate": 1.1172515899004754e-08, + "loss": 0.743, + "output_text_mask_loss": 0.0115, + "predictor_learning_rate": 1.1172515899004753e-06, + "step": 9475 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.107, + "image_masks_loss": 0.0038, + "llm_learning_rate": 1.1094307767687551e-08, + "loss": 0.717, + "output_text_mask_loss": 0.037, + "predictor_learning_rate": 1.109430776768755e-06, + "step": 9476 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1069, + "image_masks_loss": 0.0091, + "llm_learning_rate": 1.1016373720209027e-08, + "loss": 0.7372, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 1.1016373720209027e-06, + "step": 9477 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1069, + "image_masks_loss": 0.0164, + "llm_learning_rate": 1.093871376515121e-08, + "loss": 0.7109, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 1.093871376515121e-06, + "step": 9478 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1069, + "image_masks_loss": 0.0092, + "llm_learning_rate": 1.08613279110667e-08, + "loss": 0.7232, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 1.08613279110667e-06, + "step": 9479 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1069, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.0784216166477012e-08, + "loss": 0.7273, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.0784216166477013e-06, + "step": 9480 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1068, + "image_masks_loss": 0.0029, + "llm_learning_rate": 1.070737853987397e-08, + "loss": 0.8027, + "output_text_mask_loss": 0.0139, + "predictor_learning_rate": 1.0707378539873969e-06, + "step": 9481 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1068, + "image_masks_loss": 0.0108, + "llm_learning_rate": 1.0630815039719133e-08, + "loss": 0.7165, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 1.0630815039719133e-06, + "step": 9482 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1068, + "image_masks_loss": 0.0043, + "llm_learning_rate": 1.0554525674443539e-08, + "loss": 0.7227, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.0554525674443538e-06, + "step": 9483 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1068, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.0478510452448798e-08, + "loss": 0.7112, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.0478510452448797e-06, + "step": 9484 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1067, + "image_masks_loss": 0.0334, + "llm_learning_rate": 1.0402769382105716e-08, + "loss": 0.7993, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.0402769382105715e-06, + "step": 9485 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1067, + "image_masks_loss": 0.014, + "llm_learning_rate": 1.032730247175484e-08, + "loss": 0.7241, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 1.032730247175484e-06, + "step": 9486 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1067, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.0252109729707305e-08, + "loss": 0.7724, + "output_text_mask_loss": 0.0158, + "predictor_learning_rate": 1.0252109729707304e-06, + "step": 9487 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1067, + "image_masks_loss": 0.0271, + "llm_learning_rate": 1.0177191164242872e-08, + "loss": 0.7076, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 1.0177191164242872e-06, + "step": 9488 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1066, + "image_masks_loss": 0.0099, + "llm_learning_rate": 1.0102546783611888e-08, + "loss": 0.7093, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.0102546783611888e-06, + "step": 9489 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1066, + "image_masks_loss": 0.0054, + "llm_learning_rate": 1.0028176596034722e-08, + "loss": 0.8161, + "output_text_mask_loss": 0.02, + "predictor_learning_rate": 1.0028176596034722e-06, + "step": 9490 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1066, + "image_masks_loss": 0.0112, + "llm_learning_rate": 9.954080609700933e-09, + "loss": 0.7995, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 9.954080609700933e-07, + "step": 9491 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1066, + "image_masks_loss": 0.0082, + "llm_learning_rate": 9.880258832770106e-09, + "loss": 0.7981, + "output_text_mask_loss": 0.0565, + "predictor_learning_rate": 9.880258832770106e-07, + "step": 9492 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1065, + "image_masks_loss": 0.0215, + "llm_learning_rate": 9.806711273371572e-09, + "loss": 0.7432, + "output_text_mask_loss": 0.004, + "predictor_learning_rate": 9.806711273371572e-07, + "step": 9493 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1065, + "image_masks_loss": 0.0052, + "llm_learning_rate": 9.733437939604962e-09, + "loss": 0.5389, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 9.733437939604962e-07, + "step": 9494 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1065, + "image_masks_loss": 0.0036, + "llm_learning_rate": 9.66043883953882e-09, + "loss": 0.7599, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 9.660438839538821e-07, + "step": 9495 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1065, + "image_masks_loss": 0.0093, + "llm_learning_rate": 9.587713981212276e-09, + "loss": 0.6785, + "output_text_mask_loss": 0.0181, + "predictor_learning_rate": 9.587713981212276e-07, + "step": 9496 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1064, + "image_masks_loss": 0.0101, + "llm_learning_rate": 9.515263372633643e-09, + "loss": 0.6633, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 9.515263372633642e-07, + "step": 9497 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1064, + "image_masks_loss": 0.0152, + "llm_learning_rate": 9.443087021781538e-09, + "loss": 0.6733, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 9.443087021781538e-07, + "step": 9498 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1064, + "image_masks_loss": 0.0075, + "llm_learning_rate": 9.371184936604328e-09, + "loss": 0.6806, + "output_text_mask_loss": 0.0389, + "predictor_learning_rate": 9.371184936604327e-07, + "step": 9499 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1064, + "image_masks_loss": 0.0121, + "llm_learning_rate": 9.299557125019843e-09, + "loss": 0.7538, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 9.299557125019842e-07, + "step": 9500 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1063, + "image_masks_loss": 0.0115, + "llm_learning_rate": 9.228203594915942e-09, + "loss": 0.7859, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 9.22820359491594e-07, + "step": 9501 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1063, + "image_masks_loss": 0.0082, + "llm_learning_rate": 9.157124354150504e-09, + "loss": 0.7657, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 9.157124354150504e-07, + "step": 9502 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1063, + "image_masks_loss": 0.0031, + "llm_learning_rate": 9.086319410550327e-09, + "loss": 0.7258, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 9.086319410550326e-07, + "step": 9503 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1063, + "image_masks_loss": 0.0074, + "llm_learning_rate": 9.015788771913059e-09, + "loss": 0.6558, + "output_text_mask_loss": 0.0084, + "predictor_learning_rate": 9.015788771913058e-07, + "step": 9504 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1062, + "image_masks_loss": 0.0051, + "llm_learning_rate": 8.945532446005545e-09, + "loss": 0.709, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 8.945532446005544e-07, + "step": 9505 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1062, + "image_masks_loss": 0.0041, + "llm_learning_rate": 8.875550440564651e-09, + "loss": 0.7156, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 8.875550440564651e-07, + "step": 9506 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1062, + "image_masks_loss": 0.0019, + "llm_learning_rate": 8.80584276329699e-09, + "loss": 0.7065, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 8.805842763296989e-07, + "step": 9507 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1062, + "image_masks_loss": 0.0083, + "llm_learning_rate": 8.736409421878644e-09, + "loss": 0.6996, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.736409421878643e-07, + "step": 9508 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1061, + "image_masks_loss": 0.0147, + "llm_learning_rate": 8.667250423955997e-09, + "loss": 0.6954, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 8.667250423955996e-07, + "step": 9509 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1061, + "image_masks_loss": 0.0128, + "llm_learning_rate": 8.59836577714518e-09, + "loss": 0.7199, + "output_text_mask_loss": 0.0156, + "predictor_learning_rate": 8.598365777145178e-07, + "step": 9510 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1061, + "image_masks_loss": 0.0074, + "llm_learning_rate": 8.52975548903151e-09, + "loss": 0.6872, + "output_text_mask_loss": 0.0097, + "predictor_learning_rate": 8.529755489031509e-07, + "step": 9511 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.1061, + "image_masks_loss": 0.0068, + "llm_learning_rate": 8.461419567170615e-09, + "loss": 0.6899, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 8.461419567170614e-07, + "step": 9512 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.106, + "image_masks_loss": 0.0095, + "llm_learning_rate": 8.393358019088138e-09, + "loss": 0.7349, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.393358019088137e-07, + "step": 9513 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.106, + "image_masks_loss": 0.0025, + "llm_learning_rate": 8.32557085227892e-09, + "loss": 0.7422, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 8.325570852278918e-07, + "step": 9514 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.106, + "image_masks_loss": 0.0113, + "llm_learning_rate": 8.258058074207543e-09, + "loss": 0.5734, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 8.258058074207542e-07, + "step": 9515 + }, + { + "epoch": 0.97, + "gumbel_tau": 0.106, + "image_masks_loss": 0.007, + "llm_learning_rate": 8.190819692309172e-09, + "loss": 0.7306, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 8.19081969230917e-07, + "step": 9516 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1059, + "image_masks_loss": 0.0028, + "llm_learning_rate": 8.123855713987883e-09, + "loss": 0.7439, + "output_text_mask_loss": 0.0081, + "predictor_learning_rate": 8.123855713987882e-07, + "step": 9517 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1059, + "image_masks_loss": 0.0087, + "llm_learning_rate": 8.057166146618057e-09, + "loss": 0.7185, + "output_text_mask_loss": 0.0641, + "predictor_learning_rate": 8.057166146618056e-07, + "step": 9518 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1059, + "image_masks_loss": 0.015, + "llm_learning_rate": 7.990750997543816e-09, + "loss": 0.7379, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 7.990750997543816e-07, + "step": 9519 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1059, + "image_masks_loss": 0.013, + "llm_learning_rate": 7.924610274078758e-09, + "loss": 0.7712, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 7.924610274078758e-07, + "step": 9520 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1058, + "image_masks_loss": 0.016, + "llm_learning_rate": 7.858743983506223e-09, + "loss": 0.6799, + "output_text_mask_loss": 0.0021, + "predictor_learning_rate": 7.858743983506222e-07, + "step": 9521 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1058, + "image_masks_loss": 0.0031, + "llm_learning_rate": 7.793152133080128e-09, + "loss": 0.6732, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 7.793152133080129e-07, + "step": 9522 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1058, + "image_masks_loss": 0.009, + "llm_learning_rate": 7.72783473002331e-09, + "loss": 0.7643, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 7.727834730023308e-07, + "step": 9523 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1058, + "image_masks_loss": 0.0123, + "llm_learning_rate": 7.662791781528344e-09, + "loss": 0.6868, + "output_text_mask_loss": 0.0223, + "predictor_learning_rate": 7.662791781528344e-07, + "step": 9524 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1057, + "image_masks_loss": 0.0093, + "llm_learning_rate": 7.598023294758671e-09, + "loss": 0.7788, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 7.598023294758671e-07, + "step": 9525 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1057, + "image_masks_loss": 0.0136, + "llm_learning_rate": 7.533529276846085e-09, + "loss": 0.8171, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 7.533529276846085e-07, + "step": 9526 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1057, + "image_masks_loss": 0.0078, + "llm_learning_rate": 7.46930973489296e-09, + "loss": 0.706, + "output_text_mask_loss": 0.0259, + "predictor_learning_rate": 7.469309734892959e-07, + "step": 9527 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1057, + "image_masks_loss": 0.0046, + "llm_learning_rate": 7.405364675971138e-09, + "loss": 0.7613, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 7.405364675971138e-07, + "step": 9528 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1056, + "image_masks_loss": 0.012, + "llm_learning_rate": 7.341694107123043e-09, + "loss": 0.6777, + "output_text_mask_loss": 0.0179, + "predictor_learning_rate": 7.341694107123043e-07, + "step": 9529 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1056, + "image_masks_loss": 0.016, + "llm_learning_rate": 7.278298035359455e-09, + "loss": 0.7314, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 7.278298035359454e-07, + "step": 9530 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1056, + "image_masks_loss": 0.0217, + "llm_learning_rate": 7.2151764676620105e-09, + "loss": 0.7885, + "output_text_mask_loss": 0.0271, + "predictor_learning_rate": 7.215176467662011e-07, + "step": 9531 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1056, + "image_masks_loss": 0.0284, + "llm_learning_rate": 7.152329410981818e-09, + "loss": 0.6933, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 7.152329410981817e-07, + "step": 9532 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1055, + "image_masks_loss": 0.0163, + "llm_learning_rate": 7.089756872240006e-09, + "loss": 0.722, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 7.089756872240005e-07, + "step": 9533 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1055, + "image_masks_loss": 0.0085, + "llm_learning_rate": 7.027458858326619e-09, + "loss": 0.6964, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.027458858326618e-07, + "step": 9534 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1055, + "image_masks_loss": 0.0237, + "llm_learning_rate": 6.965435376102281e-09, + "loss": 0.7707, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 6.96543537610228e-07, + "step": 9535 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1055, + "image_masks_loss": 0.0061, + "llm_learning_rate": 6.903686432397361e-09, + "loss": 0.752, + "output_text_mask_loss": 0.0416, + "predictor_learning_rate": 6.90368643239736e-07, + "step": 9536 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1054, + "image_masks_loss": 0.0065, + "llm_learning_rate": 6.842212034011419e-09, + "loss": 0.7303, + "output_text_mask_loss": 0.0185, + "predictor_learning_rate": 6.842212034011419e-07, + "step": 9537 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1054, + "image_masks_loss": 0.002, + "llm_learning_rate": 6.781012187714597e-09, + "loss": 0.7437, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.781012187714597e-07, + "step": 9538 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1054, + "image_masks_loss": 0.0068, + "llm_learning_rate": 6.7200869002462256e-09, + "loss": 0.723, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.720086900246225e-07, + "step": 9539 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1054, + "image_masks_loss": 0.0054, + "llm_learning_rate": 6.659436178315381e-09, + "loss": 0.7806, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 6.659436178315382e-07, + "step": 9540 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1053, + "image_masks_loss": 0.0086, + "llm_learning_rate": 6.59906002860089e-09, + "loss": 0.7147, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 6.59906002860089e-07, + "step": 9541 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1053, + "image_masks_loss": 0.0093, + "llm_learning_rate": 6.538958457752154e-09, + "loss": 0.6926, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 6.538958457752153e-07, + "step": 9542 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1053, + "image_masks_loss": 0.0098, + "llm_learning_rate": 6.4791314723869346e-09, + "loss": 0.691, + "output_text_mask_loss": 0.0387, + "predictor_learning_rate": 6.479131472386935e-07, + "step": 9543 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1053, + "image_masks_loss": 0.006, + "llm_learning_rate": 6.419579079093852e-09, + "loss": 0.7193, + "output_text_mask_loss": 0.0084, + "predictor_learning_rate": 6.419579079093851e-07, + "step": 9544 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1052, + "image_masks_loss": 0.0078, + "llm_learning_rate": 6.360301284431269e-09, + "loss": 0.7542, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 6.360301284431269e-07, + "step": 9545 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1052, + "image_masks_loss": 0.0095, + "llm_learning_rate": 6.3012980949264666e-09, + "loss": 0.7429, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 6.301298094926467e-07, + "step": 9546 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1052, + "image_masks_loss": 0.0116, + "llm_learning_rate": 6.242569517077302e-09, + "loss": 0.7427, + "output_text_mask_loss": 0.0277, + "predictor_learning_rate": 6.242569517077301e-07, + "step": 9547 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1052, + "image_masks_loss": 0.0036, + "llm_learning_rate": 6.184115557350823e-09, + "loss": 0.7102, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 6.184115557350823e-07, + "step": 9548 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1051, + "image_masks_loss": 0.0068, + "llm_learning_rate": 6.125936222184381e-09, + "loss": 0.7207, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 6.125936222184381e-07, + "step": 9549 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1051, + "image_masks_loss": 0.0173, + "llm_learning_rate": 6.068031517984796e-09, + "loss": 0.7004, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 6.068031517984796e-07, + "step": 9550 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1051, + "image_masks_loss": 0.007, + "llm_learning_rate": 6.010401451128634e-09, + "loss": 0.7839, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 6.010401451128633e-07, + "step": 9551 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1051, + "image_masks_loss": 0.0111, + "llm_learning_rate": 5.9530460279619284e-09, + "loss": 0.7132, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 5.953046027961928e-07, + "step": 9552 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.105, + "image_masks_loss": 0.0103, + "llm_learning_rate": 5.895965254801295e-09, + "loss": 0.6932, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.895965254801294e-07, + "step": 9553 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.105, + "image_masks_loss": 0.0089, + "llm_learning_rate": 5.839159137932537e-09, + "loss": 0.7177, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 5.839159137932537e-07, + "step": 9554 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.105, + "image_masks_loss": 0.0136, + "llm_learning_rate": 5.782627683610931e-09, + "loss": 0.6676, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 5.782627683610931e-07, + "step": 9555 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.105, + "image_masks_loss": 0.0233, + "llm_learning_rate": 5.726370898061773e-09, + "loss": 0.746, + "output_text_mask_loss": 0.004, + "predictor_learning_rate": 5.726370898061773e-07, + "step": 9556 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1049, + "image_masks_loss": 0.0147, + "llm_learning_rate": 5.670388787480663e-09, + "loss": 0.7514, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 5.670388787480663e-07, + "step": 9557 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1049, + "image_masks_loss": 0.0225, + "llm_learning_rate": 5.614681358032114e-09, + "loss": 0.7489, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 5.614681358032115e-07, + "step": 9558 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1049, + "image_masks_loss": 0.0051, + "llm_learning_rate": 5.55924861585122e-09, + "loss": 0.7195, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 5.55924861585122e-07, + "step": 9559 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1049, + "image_masks_loss": 0.0057, + "llm_learning_rate": 5.504090567041709e-09, + "loss": 0.6691, + "output_text_mask_loss": 0.0172, + "predictor_learning_rate": 5.504090567041709e-07, + "step": 9560 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1048, + "image_masks_loss": 0.0059, + "llm_learning_rate": 5.449207217678165e-09, + "loss": 0.8099, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.449207217678165e-07, + "step": 9561 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1048, + "image_masks_loss": 0.0088, + "llm_learning_rate": 5.394598573804088e-09, + "loss": 0.6878, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 5.394598573804088e-07, + "step": 9562 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1048, + "image_masks_loss": 0.0134, + "llm_learning_rate": 5.3402646414335564e-09, + "loss": 0.7082, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 5.340264641433556e-07, + "step": 9563 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1048, + "image_masks_loss": 0.011, + "llm_learning_rate": 5.286205426549562e-09, + "loss": 0.7942, + "output_text_mask_loss": 0.0063, + "predictor_learning_rate": 5.286205426549561e-07, + "step": 9564 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1047, + "image_masks_loss": 0.0105, + "llm_learning_rate": 5.232420935105675e-09, + "loss": 0.671, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.232420935105675e-07, + "step": 9565 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1047, + "image_masks_loss": 0.0066, + "llm_learning_rate": 5.1789111730241036e-09, + "loss": 0.7919, + "output_text_mask_loss": 0.0046, + "predictor_learning_rate": 5.178911173024104e-07, + "step": 9566 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1047, + "image_masks_loss": 0.0085, + "llm_learning_rate": 5.125676146198189e-09, + "loss": 0.7283, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 5.125676146198188e-07, + "step": 9567 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1047, + "image_masks_loss": 0.0056, + "llm_learning_rate": 5.07271586048963e-09, + "loss": 0.6855, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 5.07271586048963e-07, + "step": 9568 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1046, + "image_masks_loss": 0.0151, + "llm_learning_rate": 5.0200303217309845e-09, + "loss": 0.8597, + "output_text_mask_loss": 0.0153, + "predictor_learning_rate": 5.020030321730984e-07, + "step": 9569 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1046, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.967619535723723e-09, + "loss": 0.6853, + "output_text_mask_loss": 0.0336, + "predictor_learning_rate": 4.967619535723722e-07, + "step": 9570 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1046, + "image_masks_loss": 0.0078, + "llm_learning_rate": 4.9154835082398934e-09, + "loss": 0.7219, + "output_text_mask_loss": 0.0055, + "predictor_learning_rate": 4.915483508239893e-07, + "step": 9571 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1046, + "image_masks_loss": 0.0129, + "llm_learning_rate": 4.8636222450207384e-09, + "loss": 0.7875, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.863622245020738e-07, + "step": 9572 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1045, + "image_masks_loss": 0.0133, + "llm_learning_rate": 4.812035751776967e-09, + "loss": 0.6822, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 4.812035751776966e-07, + "step": 9573 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1045, + "image_masks_loss": 0.0099, + "llm_learning_rate": 4.76072403418959e-09, + "loss": 0.6069, + "output_text_mask_loss": 0.0027, + "predictor_learning_rate": 4.7607240341895894e-07, + "step": 9574 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1045, + "image_masks_loss": 0.0049, + "llm_learning_rate": 4.7096870979093655e-09, + "loss": 0.6701, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.709687097909365e-07, + "step": 9575 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1045, + "image_masks_loss": 0.015, + "llm_learning_rate": 4.658924948556243e-09, + "loss": 0.6731, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.6589249485562424e-07, + "step": 9576 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1044, + "image_masks_loss": 0.0071, + "llm_learning_rate": 4.608437591720749e-09, + "loss": 0.7673, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 4.6084375917207487e-07, + "step": 9577 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1044, + "image_masks_loss": 0.0153, + "llm_learning_rate": 4.558225032962049e-09, + "loss": 0.7163, + "output_text_mask_loss": 0.017, + "predictor_learning_rate": 4.5582250329620487e-07, + "step": 9578 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1044, + "image_masks_loss": 0.0087, + "llm_learning_rate": 4.508287277810441e-09, + "loss": 0.7608, + "output_text_mask_loss": 0.0437, + "predictor_learning_rate": 4.508287277810441e-07, + "step": 9579 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1044, + "image_masks_loss": 0.0045, + "llm_learning_rate": 4.458624331764305e-09, + "loss": 0.7649, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 4.458624331764305e-07, + "step": 9580 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1043, + "image_masks_loss": 0.0026, + "llm_learning_rate": 4.4092362002931545e-09, + "loss": 0.7371, + "output_text_mask_loss": 0.0044, + "predictor_learning_rate": 4.409236200293154e-07, + "step": 9581 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1043, + "image_masks_loss": 0.0071, + "llm_learning_rate": 4.360122888835694e-09, + "loss": 0.7227, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 4.360122888835694e-07, + "step": 9582 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1043, + "image_masks_loss": 0.0123, + "llm_learning_rate": 4.3112844028003755e-09, + "loss": 0.6831, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 4.311284402800375e-07, + "step": 9583 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1043, + "image_masks_loss": 0.0048, + "llm_learning_rate": 4.262720747565119e-09, + "loss": 0.7082, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.262720747565119e-07, + "step": 9584 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1042, + "image_masks_loss": 0.0163, + "llm_learning_rate": 4.214431928478424e-09, + "loss": 0.7911, + "output_text_mask_loss": 0.0039, + "predictor_learning_rate": 4.2144319284784237e-07, + "step": 9585 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1042, + "image_masks_loss": 0.0052, + "llm_learning_rate": 4.166417950857426e-09, + "loss": 0.6688, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.166417950857426e-07, + "step": 9586 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1042, + "image_masks_loss": 0.009, + "llm_learning_rate": 4.118678819989841e-09, + "loss": 0.7317, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.118678819989841e-07, + "step": 9587 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1042, + "image_masks_loss": 0.0148, + "llm_learning_rate": 4.071214541132573e-09, + "loss": 0.6789, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 4.071214541132573e-07, + "step": 9588 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1041, + "image_masks_loss": 0.0134, + "llm_learning_rate": 4.024025119512554e-09, + "loss": 0.7773, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 4.0240251195125533e-07, + "step": 9589 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1041, + "image_masks_loss": 0.0112, + "llm_learning_rate": 3.9771105603264586e-09, + "loss": 0.7382, + "output_text_mask_loss": 0.0261, + "predictor_learning_rate": 3.977110560326458e-07, + "step": 9590 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1041, + "image_masks_loss": 0.0162, + "llm_learning_rate": 3.930470868740988e-09, + "loss": 0.6728, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.9304708687409873e-07, + "step": 9591 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1041, + "image_masks_loss": 0.0122, + "llm_learning_rate": 3.884106049891478e-09, + "loss": 0.7442, + "output_text_mask_loss": 0.0056, + "predictor_learning_rate": 3.884106049891478e-07, + "step": 9592 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.104, + "image_masks_loss": 0.016, + "llm_learning_rate": 3.838016108884124e-09, + "loss": 0.6716, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.8380161088841236e-07, + "step": 9593 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.104, + "image_masks_loss": 0.0033, + "llm_learning_rate": 3.792201050794586e-09, + "loss": 0.7743, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 3.792201050794586e-07, + "step": 9594 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.104, + "image_masks_loss": 0.0039, + "llm_learning_rate": 3.7466608806677186e-09, + "loss": 0.7623, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 3.746660880667718e-07, + "step": 9595 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.104, + "image_masks_loss": 0.0025, + "llm_learning_rate": 3.7013956035189536e-09, + "loss": 0.6973, + "output_text_mask_loss": 0.0368, + "predictor_learning_rate": 3.7013956035189533e-07, + "step": 9596 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1039, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.6564052243326376e-09, + "loss": 0.7614, + "output_text_mask_loss": 0.0111, + "predictor_learning_rate": 3.6564052243326373e-07, + "step": 9597 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1039, + "image_masks_loss": 0.014, + "llm_learning_rate": 3.611689748063696e-09, + "loss": 0.7671, + "output_text_mask_loss": 0.0292, + "predictor_learning_rate": 3.611689748063696e-07, + "step": 9598 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1039, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.56724917963569e-09, + "loss": 0.7531, + "output_text_mask_loss": 0.0534, + "predictor_learning_rate": 3.56724917963569e-07, + "step": 9599 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1039, + "image_masks_loss": 0.0141, + "llm_learning_rate": 3.5230835239427607e-09, + "loss": 0.7188, + "output_text_mask_loss": 0.0041, + "predictor_learning_rate": 3.5230835239427606e-07, + "step": 9600 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1038, + "image_masks_loss": 0.0065, + "llm_learning_rate": 3.4791927858485175e-09, + "loss": 0.7208, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.479192785848517e-07, + "step": 9601 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1038, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.435576970186594e-09, + "loss": 0.6954, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.4355769701865934e-07, + "step": 9602 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1038, + "image_masks_loss": 0.0053, + "llm_learning_rate": 3.3922360817595366e-09, + "loss": 0.8373, + "output_text_mask_loss": 0.0198, + "predictor_learning_rate": 3.392236081759537e-07, + "step": 9603 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1038, + "image_masks_loss": 0.0077, + "llm_learning_rate": 3.3491701253407504e-09, + "loss": 0.833, + "output_text_mask_loss": 0.0101, + "predictor_learning_rate": 3.34917012534075e-07, + "step": 9604 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1037, + "image_masks_loss": 0.0131, + "llm_learning_rate": 3.3063791056719973e-09, + "loss": 0.7593, + "output_text_mask_loss": 0.0173, + "predictor_learning_rate": 3.306379105671997e-07, + "step": 9605 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1037, + "image_masks_loss": 0.0099, + "llm_learning_rate": 3.263863027466174e-09, + "loss": 0.6951, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.263863027466174e-07, + "step": 9606 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1037, + "image_masks_loss": 0.0109, + "llm_learning_rate": 3.2216218954050916e-09, + "loss": 0.7534, + "output_text_mask_loss": 0.0248, + "predictor_learning_rate": 3.221621895405091e-07, + "step": 9607 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1037, + "image_masks_loss": 0.0064, + "llm_learning_rate": 3.179655714140306e-09, + "loss": 0.6679, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.179655714140306e-07, + "step": 9608 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1037, + "image_masks_loss": 0.0113, + "llm_learning_rate": 3.1379644882931216e-09, + "loss": 0.7262, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.137964488293121e-07, + "step": 9609 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1036, + "image_masks_loss": 0.0036, + "llm_learning_rate": 3.0965482224551423e-09, + "loss": 0.6993, + "output_text_mask_loss": 0.071, + "predictor_learning_rate": 3.0965482224551424e-07, + "step": 9610 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1036, + "image_masks_loss": 0.0026, + "llm_learning_rate": 3.05540692118661e-09, + "loss": 0.6734, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 3.05540692118661e-07, + "step": 9611 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1036, + "image_masks_loss": 0.0104, + "llm_learning_rate": 3.014540589018622e-09, + "loss": 0.7218, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 3.014540589018622e-07, + "step": 9612 + }, + { + "epoch": 0.98, + "gumbel_tau": 0.1036, + "image_masks_loss": 0.0208, + "llm_learning_rate": 2.9739492304511895e-09, + "loss": 0.7609, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 2.9739492304511897e-07, + "step": 9613 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1035, + "image_masks_loss": 0.0074, + "llm_learning_rate": 2.9336328499543487e-09, + "loss": 0.7746, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.9336328499543484e-07, + "step": 9614 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1035, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.8935914519676035e-09, + "loss": 0.726, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 2.893591451967603e-07, + "step": 9615 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1035, + "image_masks_loss": 0.0067, + "llm_learning_rate": 2.853825040901037e-09, + "loss": 0.7299, + "output_text_mask_loss": 0.0263, + "predictor_learning_rate": 2.8538250409010367e-07, + "step": 9616 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1035, + "image_masks_loss": 0.0128, + "llm_learning_rate": 2.814333621133092e-09, + "loss": 0.7588, + "output_text_mask_loss": 0.0088, + "predictor_learning_rate": 2.8143336211330915e-07, + "step": 9617 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1034, + "image_masks_loss": 0.0181, + "llm_learning_rate": 2.775117197013344e-09, + "loss": 0.6073, + "output_text_mask_loss": 0.0235, + "predictor_learning_rate": 2.775117197013344e-07, + "step": 9618 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1034, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.7361757728597283e-09, + "loss": 0.6724, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 2.736175772859728e-07, + "step": 9619 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1034, + "image_masks_loss": 0.0189, + "llm_learning_rate": 2.6975093529610364e-09, + "loss": 0.7688, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 2.697509352961036e-07, + "step": 9620 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1034, + "image_masks_loss": 0.0069, + "llm_learning_rate": 2.659117941575251e-09, + "loss": 0.7298, + "output_text_mask_loss": 0.0553, + "predictor_learning_rate": 2.6591179415752507e-07, + "step": 9621 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1033, + "image_masks_loss": 0.0046, + "llm_learning_rate": 2.6210015429298243e-09, + "loss": 0.7126, + "output_text_mask_loss": 0.0443, + "predictor_learning_rate": 2.6210015429298237e-07, + "step": 9622 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1033, + "image_masks_loss": 0.0027, + "llm_learning_rate": 2.5831601612225087e-09, + "loss": 0.6828, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.583160161222509e-07, + "step": 9623 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1033, + "image_masks_loss": 0.0151, + "llm_learning_rate": 2.5455938006202495e-09, + "loss": 0.7071, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 2.5455938006202494e-07, + "step": 9624 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1033, + "image_masks_loss": 0.0036, + "llm_learning_rate": 2.508302465260293e-09, + "loss": 0.7213, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.5083024652602924e-07, + "step": 9625 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1032, + "image_masks_loss": 0.006, + "llm_learning_rate": 2.471286159248798e-09, + "loss": 0.7468, + "output_text_mask_loss": 0.0147, + "predictor_learning_rate": 2.471286159248798e-07, + "step": 9626 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1032, + "image_masks_loss": 0.008, + "llm_learning_rate": 2.4345448866622268e-09, + "loss": 0.6657, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.4345448866622265e-07, + "step": 9627 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1032, + "image_masks_loss": 0.0038, + "llm_learning_rate": 2.398078651546787e-09, + "loss": 0.7376, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.3980786515467866e-07, + "step": 9628 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1032, + "image_masks_loss": 0.0048, + "llm_learning_rate": 2.361887457918155e-09, + "loss": 0.7201, + "output_text_mask_loss": 0.0256, + "predictor_learning_rate": 2.3618874579181548e-07, + "step": 9629 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1031, + "image_masks_loss": 0.011, + "llm_learning_rate": 2.325971309761754e-09, + "loss": 0.7561, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.325971309761754e-07, + "step": 9630 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1031, + "image_masks_loss": 0.0078, + "llm_learning_rate": 2.2903302110327542e-09, + "loss": 0.698, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 2.290330211032754e-07, + "step": 9631 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1031, + "image_masks_loss": 0.0051, + "llm_learning_rate": 2.2549641656560705e-09, + "loss": 0.6771, + "output_text_mask_loss": 0.0265, + "predictor_learning_rate": 2.2549641656560704e-07, + "step": 9632 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1031, + "image_masks_loss": 0.0111, + "llm_learning_rate": 2.2198731775260883e-09, + "loss": 0.7454, + "output_text_mask_loss": 0.0237, + "predictor_learning_rate": 2.2198731775260884e-07, + "step": 9633 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.103, + "image_masks_loss": 0.0194, + "llm_learning_rate": 2.1850572505072167e-09, + "loss": 0.7646, + "output_text_mask_loss": 0.0275, + "predictor_learning_rate": 2.1850572505072165e-07, + "step": 9634 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.103, + "image_masks_loss": 0.0018, + "llm_learning_rate": 2.15051638843361e-09, + "loss": 0.7056, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 2.1505163884336098e-07, + "step": 9635 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.103, + "image_masks_loss": 0.0077, + "llm_learning_rate": 2.116250595108893e-09, + "loss": 0.6895, + "output_text_mask_loss": 0.0107, + "predictor_learning_rate": 2.116250595108893e-07, + "step": 9636 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.103, + "image_masks_loss": 0.0098, + "llm_learning_rate": 2.082259874306436e-09, + "loss": 0.7141, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 2.0822598743064357e-07, + "step": 9637 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1029, + "image_masks_loss": 0.0108, + "llm_learning_rate": 2.0485442297696336e-09, + "loss": 0.7013, + "output_text_mask_loss": 0.0668, + "predictor_learning_rate": 2.0485442297696334e-07, + "step": 9638 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1029, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.0151036652107936e-09, + "loss": 0.6746, + "output_text_mask_loss": 0.0515, + "predictor_learning_rate": 2.0151036652107936e-07, + "step": 9639 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1029, + "image_masks_loss": 0.0113, + "llm_learning_rate": 1.9819381843130815e-09, + "loss": 0.7021, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 1.9819381843130813e-07, + "step": 9640 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1029, + "image_masks_loss": 0.0076, + "llm_learning_rate": 1.9490477907282977e-09, + "loss": 0.6997, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 1.9490477907282978e-07, + "step": 9641 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1028, + "image_masks_loss": 0.0046, + "llm_learning_rate": 1.9164324880785458e-09, + "loss": 0.686, + "output_text_mask_loss": 0.0164, + "predictor_learning_rate": 1.9164324880785454e-07, + "step": 9642 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1028, + "image_masks_loss": 0.002, + "llm_learning_rate": 1.8840922799553964e-09, + "loss": 0.6648, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.8840922799553961e-07, + "step": 9643 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1028, + "image_masks_loss": 0.0086, + "llm_learning_rate": 1.8520271699204452e-09, + "loss": 0.7362, + "output_text_mask_loss": 0.0238, + "predictor_learning_rate": 1.852027169920445e-07, + "step": 9644 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1028, + "image_masks_loss": 0.0145, + "llm_learning_rate": 1.8202371615047566e-09, + "loss": 0.7153, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.8202371615047563e-07, + "step": 9645 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1028, + "image_masks_loss": 0.0138, + "llm_learning_rate": 1.7887222582088638e-09, + "loss": 0.6743, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.7887222582088635e-07, + "step": 9646 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1027, + "image_masks_loss": 0.0191, + "llm_learning_rate": 1.7574824635036015e-09, + "loss": 0.7306, + "output_text_mask_loss": 0.0839, + "predictor_learning_rate": 1.7574824635036013e-07, + "step": 9647 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1027, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.7265177808287181e-09, + "loss": 0.7034, + "output_text_mask_loss": 0.0023, + "predictor_learning_rate": 1.726517780828718e-07, + "step": 9648 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1027, + "image_masks_loss": 0.0071, + "llm_learning_rate": 1.695828213594819e-09, + "loss": 0.7793, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.695828213594819e-07, + "step": 9649 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1027, + "image_masks_loss": 0.0132, + "llm_learning_rate": 1.6654137651805901e-09, + "loss": 0.7122, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 1.66541376518059e-07, + "step": 9650 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1026, + "image_masks_loss": 0.0073, + "llm_learning_rate": 1.6352744389361297e-09, + "loss": 0.6965, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.6352744389361295e-07, + "step": 9651 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1026, + "image_masks_loss": 0.0074, + "llm_learning_rate": 1.6054102381801718e-09, + "loss": 0.6937, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 1.6054102381801716e-07, + "step": 9652 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1026, + "image_masks_loss": 0.0282, + "llm_learning_rate": 1.575821166201197e-09, + "loss": 0.7278, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.575821166201197e-07, + "step": 9653 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1026, + "image_masks_loss": 0.0096, + "llm_learning_rate": 1.5465072262579872e-09, + "loss": 0.6164, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.546507226257987e-07, + "step": 9654 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1025, + "image_masks_loss": 0.006, + "llm_learning_rate": 1.5174684215782386e-09, + "loss": 0.7107, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.5174684215782386e-07, + "step": 9655 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1025, + "image_masks_loss": 0.0084, + "llm_learning_rate": 1.4887047553602262e-09, + "loss": 0.7082, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.4887047553602263e-07, + "step": 9656 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1025, + "image_masks_loss": 0.0121, + "llm_learning_rate": 1.4602162307711388e-09, + "loss": 0.7083, + "output_text_mask_loss": 0.0193, + "predictor_learning_rate": 1.4602162307711387e-07, + "step": 9657 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1025, + "image_masks_loss": 0.0055, + "llm_learning_rate": 1.4320028509484663e-09, + "loss": 0.7718, + "output_text_mask_loss": 0.0269, + "predictor_learning_rate": 1.432002850948466e-07, + "step": 9658 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1024, + "image_masks_loss": 0.0072, + "llm_learning_rate": 1.4040646189988905e-09, + "loss": 0.7497, + "output_text_mask_loss": 0.028, + "predictor_learning_rate": 1.4040646189988905e-07, + "step": 9659 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1024, + "image_masks_loss": 0.0031, + "llm_learning_rate": 1.3764015379991168e-09, + "loss": 0.7468, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.376401537999117e-07, + "step": 9660 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1024, + "image_masks_loss": 0.0048, + "llm_learning_rate": 1.3490136109955975e-09, + "loss": 0.7537, + "output_text_mask_loss": 0.0034, + "predictor_learning_rate": 1.3490136109955976e-07, + "step": 9661 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1024, + "image_masks_loss": 0.0172, + "llm_learning_rate": 1.3219008410042533e-09, + "loss": 0.6985, + "output_text_mask_loss": 0.0542, + "predictor_learning_rate": 1.3219008410042532e-07, + "step": 9662 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1023, + "image_masks_loss": 0.0088, + "llm_learning_rate": 1.295063231010474e-09, + "loss": 0.7202, + "output_text_mask_loss": 0.0124, + "predictor_learning_rate": 1.2950632310104738e-07, + "step": 9663 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1023, + "image_masks_loss": 0.0106, + "llm_learning_rate": 1.268500783970228e-09, + "loss": 0.7463, + "output_text_mask_loss": 0.0599, + "predictor_learning_rate": 1.2685007839702278e-07, + "step": 9664 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1023, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.2422135028083982e-09, + "loss": 0.7448, + "output_text_mask_loss": 0.0093, + "predictor_learning_rate": 1.2422135028083982e-07, + "step": 9665 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1023, + "image_masks_loss": 0.0039, + "llm_learning_rate": 1.2162013904196134e-09, + "loss": 0.7411, + "output_text_mask_loss": 0.0162, + "predictor_learning_rate": 1.2162013904196135e-07, + "step": 9666 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1022, + "image_masks_loss": 0.0142, + "llm_learning_rate": 1.1904644496685268e-09, + "loss": 0.7211, + "output_text_mask_loss": 0.0397, + "predictor_learning_rate": 1.1904644496685268e-07, + "step": 9667 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1022, + "image_masks_loss": 0.0135, + "llm_learning_rate": 1.1650026833895377e-09, + "loss": 0.727, + "output_text_mask_loss": 0.0307, + "predictor_learning_rate": 1.1650026833895377e-07, + "step": 9668 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1022, + "image_masks_loss": 0.0111, + "llm_learning_rate": 1.139816094386237e-09, + "loss": 0.7152, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 1.1398160943862368e-07, + "step": 9669 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1022, + "image_masks_loss": 0.0049, + "llm_learning_rate": 1.1149046854325162e-09, + "loss": 0.7573, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 1.1149046854325163e-07, + "step": 9670 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1021, + "image_masks_loss": 0.0081, + "llm_learning_rate": 1.0902684592714596e-09, + "loss": 0.7173, + "output_text_mask_loss": 0.0542, + "predictor_learning_rate": 1.0902684592714595e-07, + "step": 9671 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1021, + "image_masks_loss": 0.0085, + "llm_learning_rate": 1.0659074186164519e-09, + "loss": 0.7513, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 1.065907418616452e-07, + "step": 9672 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1021, + "image_masks_loss": 0.0058, + "llm_learning_rate": 1.0418215661495146e-09, + "loss": 0.6956, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.0418215661495145e-07, + "step": 9673 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1021, + "image_masks_loss": 0.0153, + "llm_learning_rate": 1.0180109045235254e-09, + "loss": 0.7323, + "output_text_mask_loss": 0.016, + "predictor_learning_rate": 1.0180109045235253e-07, + "step": 9674 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.102, + "image_masks_loss": 0.0117, + "llm_learning_rate": 9.94475436360276e-10, + "loss": 0.707, + "output_text_mask_loss": 0.0572, + "predictor_learning_rate": 9.944754363602759e-08, + "step": 9675 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.102, + "image_masks_loss": 0.0107, + "llm_learning_rate": 9.712151642518598e-10, + "loss": 0.678, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 9.712151642518597e-08, + "step": 9676 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.102, + "image_masks_loss": 0.0052, + "llm_learning_rate": 9.48230090759561e-10, + "loss": 0.7983, + "output_text_mask_loss": 0.0095, + "predictor_learning_rate": 9.482300907595609e-08, + "step": 9677 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.102, + "image_masks_loss": 0.0116, + "llm_learning_rate": 9.255202184146883e-10, + "loss": 0.6854, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 9.255202184146883e-08, + "step": 9678 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.102, + "image_masks_loss": 0.0092, + "llm_learning_rate": 9.03085549718019e-10, + "loss": 0.7729, + "output_text_mask_loss": 0.1205, + "predictor_learning_rate": 9.030855497180191e-08, + "step": 9679 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1019, + "image_masks_loss": 0.0097, + "llm_learning_rate": 8.809260871397995e-10, + "loss": 0.7736, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 8.809260871397995e-08, + "step": 9680 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1019, + "image_masks_loss": 0.0067, + "llm_learning_rate": 8.59041833120855e-10, + "loss": 0.6878, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 8.590418331208549e-08, + "step": 9681 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1019, + "image_masks_loss": 0.0023, + "llm_learning_rate": 8.374327900706469e-10, + "loss": 0.7229, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 8.374327900706468e-08, + "step": 9682 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1019, + "image_masks_loss": 0.0062, + "llm_learning_rate": 8.16098960369216e-10, + "loss": 0.8922, + "output_text_mask_loss": 0.0467, + "predictor_learning_rate": 8.160989603692159e-08, + "step": 9683 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1018, + "image_masks_loss": 0.0048, + "llm_learning_rate": 7.950403463655165e-10, + "loss": 0.7287, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 7.950403463655165e-08, + "step": 9684 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1018, + "image_masks_loss": 0.0068, + "llm_learning_rate": 7.742569503788044e-10, + "loss": 0.7709, + "output_text_mask_loss": 0.0006, + "predictor_learning_rate": 7.742569503788044e-08, + "step": 9685 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1018, + "image_masks_loss": 0.0049, + "llm_learning_rate": 7.537487746978045e-10, + "loss": 0.6817, + "output_text_mask_loss": 0.0231, + "predictor_learning_rate": 7.537487746978045e-08, + "step": 9686 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1018, + "image_masks_loss": 0.006, + "llm_learning_rate": 7.335158215807104e-10, + "loss": 0.7865, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 7.335158215807103e-08, + "step": 9687 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1017, + "image_masks_loss": 0.0156, + "llm_learning_rate": 7.135580932557395e-10, + "loss": 0.7434, + "output_text_mask_loss": 0.034, + "predictor_learning_rate": 7.135580932557395e-08, + "step": 9688 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1017, + "image_masks_loss": 0.0075, + "llm_learning_rate": 6.938755919208562e-10, + "loss": 0.698, + "output_text_mask_loss": 0.0025, + "predictor_learning_rate": 6.938755919208562e-08, + "step": 9689 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1017, + "image_masks_loss": 0.0058, + "llm_learning_rate": 6.744683197432155e-10, + "loss": 0.6906, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.744683197432156e-08, + "step": 9690 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1017, + "image_masks_loss": 0.0028, + "llm_learning_rate": 6.553362788602746e-10, + "loss": 0.6923, + "output_text_mask_loss": 0.0099, + "predictor_learning_rate": 6.553362788602745e-08, + "step": 9691 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1016, + "image_masks_loss": 0.0105, + "llm_learning_rate": 6.364794713786814e-10, + "loss": 0.7966, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 6.364794713786815e-08, + "step": 9692 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1016, + "image_masks_loss": 0.0259, + "llm_learning_rate": 6.178978993751083e-10, + "loss": 0.6711, + "output_text_mask_loss": 0.0097, + "predictor_learning_rate": 6.178978993751084e-08, + "step": 9693 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1016, + "image_masks_loss": 0.0188, + "llm_learning_rate": 5.995915648956963e-10, + "loss": 0.7857, + "output_text_mask_loss": 0.0121, + "predictor_learning_rate": 5.995915648956962e-08, + "step": 9694 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1016, + "image_masks_loss": 0.0141, + "llm_learning_rate": 5.815604699566102e-10, + "loss": 0.7162, + "output_text_mask_loss": 0.0221, + "predictor_learning_rate": 5.8156046995661014e-08, + "step": 9695 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1015, + "image_masks_loss": 0.0024, + "llm_learning_rate": 5.638046165432065e-10, + "loss": 0.8127, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 5.638046165432065e-08, + "step": 9696 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1015, + "image_masks_loss": 0.0162, + "llm_learning_rate": 5.463240066108655e-10, + "loss": 0.6937, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 5.463240066108655e-08, + "step": 9697 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1015, + "image_masks_loss": 0.0189, + "llm_learning_rate": 5.291186420847139e-10, + "loss": 0.7804, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 5.2911864208471386e-08, + "step": 9698 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1015, + "image_masks_loss": 0.0032, + "llm_learning_rate": 5.121885248593473e-10, + "loss": 0.7266, + "output_text_mask_loss": 0.0236, + "predictor_learning_rate": 5.121885248593472e-08, + "step": 9699 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1014, + "image_masks_loss": 0.0069, + "llm_learning_rate": 4.955336567991076e-10, + "loss": 0.679, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.9553365679910754e-08, + "step": 9700 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1014, + "image_masks_loss": 0.0046, + "llm_learning_rate": 4.79154039738361e-10, + "loss": 0.7152, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.791540397383609e-08, + "step": 9701 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1014, + "image_masks_loss": 0.0213, + "llm_learning_rate": 4.63049675480387e-10, + "loss": 0.7413, + "output_text_mask_loss": 0.0069, + "predictor_learning_rate": 4.6304967548038704e-08, + "step": 9702 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1014, + "image_masks_loss": 0.0057, + "llm_learning_rate": 4.472205657990447e-10, + "loss": 0.7238, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 4.472205657990447e-08, + "step": 9703 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1014, + "image_masks_loss": 0.0035, + "llm_learning_rate": 4.3166671243738414e-10, + "loss": 0.8686, + "output_text_mask_loss": 0.0322, + "predictor_learning_rate": 4.316667124373841e-08, + "step": 9704 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1013, + "image_masks_loss": 0.0206, + "llm_learning_rate": 4.163881171079243e-10, + "loss": 0.6276, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.1638811710792425e-08, + "step": 9705 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1013, + "image_masks_loss": 0.0121, + "llm_learning_rate": 4.013847814934857e-10, + "loss": 0.7388, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 4.013847814934857e-08, + "step": 9706 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1013, + "image_masks_loss": 0.0106, + "llm_learning_rate": 3.8665670724635784e-10, + "loss": 0.7962, + "output_text_mask_loss": 0.0103, + "predictor_learning_rate": 3.8665670724635784e-08, + "step": 9707 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1013, + "image_masks_loss": 0.0121, + "llm_learning_rate": 3.7220389598802143e-10, + "loss": 0.7799, + "output_text_mask_loss": 0.0154, + "predictor_learning_rate": 3.722038959880214e-08, + "step": 9708 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1012, + "image_masks_loss": 0.0146, + "llm_learning_rate": 3.5802634931053625e-10, + "loss": 0.8393, + "output_text_mask_loss": 0.0353, + "predictor_learning_rate": 3.580263493105362e-08, + "step": 9709 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1012, + "image_masks_loss": 0.0127, + "llm_learning_rate": 3.441240687748759e-10, + "loss": 0.8106, + "output_text_mask_loss": 0.0343, + "predictor_learning_rate": 3.441240687748759e-08, + "step": 9710 + }, + { + "epoch": 0.99, + "gumbel_tau": 0.1012, + "image_masks_loss": 0.0061, + "llm_learning_rate": 3.304970559117604e-10, + "loss": 0.8019, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 3.304970559117604e-08, + "step": 9711 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1012, + "image_masks_loss": 0.0088, + "llm_learning_rate": 3.1714531222221123e-10, + "loss": 0.7062, + "output_text_mask_loss": 0.0078, + "predictor_learning_rate": 3.171453122222112e-08, + "step": 9712 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1011, + "image_masks_loss": 0.014, + "llm_learning_rate": 3.040688391767188e-10, + "loss": 0.7514, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 3.0406883917671877e-08, + "step": 9713 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1011, + "image_masks_loss": 0.0164, + "llm_learning_rate": 2.912676382146873e-10, + "loss": 0.6928, + "output_text_mask_loss": 0.001, + "predictor_learning_rate": 2.9126763821468726e-08, + "step": 9714 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1011, + "image_masks_loss": 0.0122, + "llm_learning_rate": 2.7874171074637745e-10, + "loss": 0.744, + "output_text_mask_loss": 0.0137, + "predictor_learning_rate": 2.7874171074637744e-08, + "step": 9715 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1011, + "image_masks_loss": 0.0023, + "llm_learning_rate": 2.664910581509639e-10, + "loss": 0.7469, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 2.6649105815096387e-08, + "step": 9716 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.101, + "image_masks_loss": 0.0037, + "llm_learning_rate": 2.5451568177736754e-10, + "loss": 0.7414, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 2.545156817773675e-08, + "step": 9717 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.101, + "image_masks_loss": 0.0047, + "llm_learning_rate": 2.4281558294453334e-10, + "loss": 0.788, + "output_text_mask_loss": 0.046, + "predictor_learning_rate": 2.4281558294453332e-08, + "step": 9718 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.101, + "image_masks_loss": 0.0068, + "llm_learning_rate": 2.313907629405976e-10, + "loss": 0.7961, + "output_text_mask_loss": 0.0058, + "predictor_learning_rate": 2.3139076294059757e-08, + "step": 9719 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.101, + "image_masks_loss": 0.01, + "llm_learning_rate": 2.2024122302427565e-10, + "loss": 0.6596, + "output_text_mask_loss": 0.0557, + "predictor_learning_rate": 2.2024122302427563e-08, + "step": 9720 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1009, + "image_masks_loss": 0.004, + "llm_learning_rate": 2.0936696442291904e-10, + "loss": 0.6441, + "output_text_mask_loss": 0.0233, + "predictor_learning_rate": 2.0936696442291902e-08, + "step": 9721 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1009, + "image_masks_loss": 0.0129, + "llm_learning_rate": 1.9876798833390332e-10, + "loss": 0.79, + "output_text_mask_loss": 0.0332, + "predictor_learning_rate": 1.9876798833390332e-08, + "step": 9722 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1009, + "image_masks_loss": 0.0176, + "llm_learning_rate": 1.8844429592490554e-10, + "loss": 0.8387, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 1.8844429592490554e-08, + "step": 9723 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1009, + "image_masks_loss": 0.0064, + "llm_learning_rate": 1.7839588833251653e-10, + "loss": 0.6821, + "output_text_mask_loss": 0.0048, + "predictor_learning_rate": 1.7839588833251652e-08, + "step": 9724 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1009, + "image_masks_loss": 0.0162, + "llm_learning_rate": 1.686227666633511e-10, + "loss": 0.708, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.686227666633511e-08, + "step": 9725 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1008, + "image_masks_loss": 0.0176, + "llm_learning_rate": 1.5912493199349287e-10, + "loss": 0.7068, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 1.5912493199349286e-08, + "step": 9726 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1008, + "image_masks_loss": 0.0149, + "llm_learning_rate": 1.4990238536904955e-10, + "loss": 0.7205, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.4990238536904953e-08, + "step": 9727 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1008, + "image_masks_loss": 0.0062, + "llm_learning_rate": 1.4095512780532005e-10, + "loss": 0.809, + "output_text_mask_loss": 0.0208, + "predictor_learning_rate": 1.4095512780532004e-08, + "step": 9728 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1008, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.3228316028818243e-10, + "loss": 0.7183, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 1.3228316028818244e-08, + "step": 9729 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1007, + "image_masks_loss": 0.0116, + "llm_learning_rate": 1.238864837718734e-10, + "loss": 0.7368, + "output_text_mask_loss": 0.0031, + "predictor_learning_rate": 1.238864837718734e-08, + "step": 9730 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1007, + "image_masks_loss": 0.0135, + "llm_learning_rate": 1.1576509918176381e-10, + "loss": 0.6587, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.1576509918176382e-08, + "step": 9731 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1007, + "image_masks_loss": 0.0165, + "llm_learning_rate": 1.0791900741186079e-10, + "loss": 0.7329, + "output_text_mask_loss": 0.0191, + "predictor_learning_rate": 1.0791900741186078e-08, + "step": 9732 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1007, + "image_masks_loss": 0.0057, + "llm_learning_rate": 1.0034820932591782e-10, + "loss": 0.6922, + "output_text_mask_loss": 0.0019, + "predictor_learning_rate": 1.0034820932591782e-08, + "step": 9733 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1006, + "image_masks_loss": 0.0103, + "llm_learning_rate": 9.305270575826752e-11, + "loss": 0.752, + "output_text_mask_loss": 0.0122, + "predictor_learning_rate": 9.305270575826752e-09, + "step": 9734 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1006, + "image_masks_loss": 0.0099, + "llm_learning_rate": 8.60324975118787e-11, + "loss": 0.6419, + "output_text_mask_loss": 0.0298, + "predictor_learning_rate": 8.60324975118787e-09, + "step": 9735 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1006, + "image_masks_loss": 0.0141, + "llm_learning_rate": 7.928758535974412e-11, + "loss": 0.7124, + "output_text_mask_loss": 0.0004, + "predictor_learning_rate": 7.928758535974413e-09, + "step": 9736 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1006, + "image_masks_loss": 0.0079, + "llm_learning_rate": 7.281797004515812e-11, + "loss": 0.759, + "output_text_mask_loss": 0.0376, + "predictor_learning_rate": 7.281797004515811e-09, + "step": 9737 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1005, + "image_masks_loss": 0.0123, + "llm_learning_rate": 6.662365228005119e-11, + "loss": 0.6841, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 6.662365228005118e-09, + "step": 9738 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1005, + "image_masks_loss": 0.0047, + "llm_learning_rate": 6.070463274665539e-11, + "loss": 0.7152, + "output_text_mask_loss": 0.004, + "predictor_learning_rate": 6.0704632746655386e-09, + "step": 9739 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1005, + "image_masks_loss": 0.0044, + "llm_learning_rate": 5.506091209694919e-11, + "loss": 0.7196, + "output_text_mask_loss": 0.005, + "predictor_learning_rate": 5.506091209694919e-09, + "step": 9740 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1005, + "image_masks_loss": 0.0104, + "llm_learning_rate": 4.969249095210238e-11, + "loss": 0.6895, + "output_text_mask_loss": 0.0175, + "predictor_learning_rate": 4.969249095210238e-09, + "step": 9741 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1004, + "image_masks_loss": 0.0062, + "llm_learning_rate": 4.459936990386382e-11, + "loss": 0.6866, + "output_text_mask_loss": 0.0008, + "predictor_learning_rate": 4.459936990386382e-09, + "step": 9742 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1004, + "image_masks_loss": 0.0153, + "llm_learning_rate": 3.978154951261859e-11, + "loss": 0.7409, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.978154951261859e-09, + "step": 9743 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1004, + "image_masks_loss": 0.0057, + "llm_learning_rate": 3.523903030933085e-11, + "loss": 0.6704, + "output_text_mask_loss": 0.0576, + "predictor_learning_rate": 3.5239030309330844e-09, + "step": 9744 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1004, + "image_masks_loss": 0.0028, + "llm_learning_rate": 3.0971812793600954e-11, + "loss": 0.7797, + "output_text_mask_loss": 0.0429, + "predictor_learning_rate": 3.097181279360095e-09, + "step": 9745 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1004, + "image_masks_loss": 0.004, + "llm_learning_rate": 2.6979897435885915e-11, + "loss": 0.7027, + "output_text_mask_loss": 0.0065, + "predictor_learning_rate": 2.6979897435885913e-09, + "step": 9746 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1003, + "image_masks_loss": 0.0096, + "llm_learning_rate": 2.3263284675556498e-11, + "loss": 0.7351, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 2.3263284675556496e-09, + "step": 9747 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1003, + "image_masks_loss": 0.0101, + "llm_learning_rate": 1.9821974922007436e-11, + "loss": 0.8101, + "output_text_mask_loss": 0.0017, + "predictor_learning_rate": 1.9821974922007436e-09, + "step": 9748 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1003, + "image_masks_loss": 0.011, + "llm_learning_rate": 1.6655968554379897e-11, + "loss": 0.7385, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 1.6655968554379897e-09, + "step": 9749 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1003, + "image_masks_loss": 0.008, + "llm_learning_rate": 1.37652659207288e-11, + "loss": 0.7242, + "output_text_mask_loss": 0.0341, + "predictor_learning_rate": 1.3765265920728798e-09, + "step": 9750 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1002, + "image_masks_loss": 0.0065, + "llm_learning_rate": 1.1149867339965704e-11, + "loss": 0.6849, + "output_text_mask_loss": 0.0381, + "predictor_learning_rate": 1.1149867339965703e-09, + "step": 9751 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1002, + "image_masks_loss": 0.0055, + "llm_learning_rate": 8.809773099638375e-12, + "loss": 0.7118, + "output_text_mask_loss": 0.05, + "predictor_learning_rate": 8.809773099638374e-10, + "step": 9752 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1002, + "image_masks_loss": 0.0288, + "llm_learning_rate": 6.744983457873666e-12, + "loss": 0.7047, + "output_text_mask_loss": 0.013, + "predictor_learning_rate": 6.744983457873666e-10, + "step": 9753 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1002, + "image_masks_loss": 0.0064, + "llm_learning_rate": 4.955498641989742e-12, + "loss": 0.6901, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 4.955498641989741e-10, + "step": 9754 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1001, + "image_masks_loss": 0.0204, + "llm_learning_rate": 3.441318848773634e-12, + "loss": 0.6843, + "output_text_mask_loss": 0.0, + "predictor_learning_rate": 3.4413188487736334e-10, + "step": 9755 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1001, + "image_masks_loss": 0.0117, + "llm_learning_rate": 2.2024442450363505e-12, + "loss": 0.7701, + "output_text_mask_loss": 0.0118, + "predictor_learning_rate": 2.2024442450363502e-10, + "step": 9756 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1001, + "image_masks_loss": 0.0126, + "llm_learning_rate": 1.238874967335324e-12, + "loss": 0.6589, + "output_text_mask_loss": 0.0002, + "predictor_learning_rate": 1.238874967335324e-10, + "step": 9757 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1001, + "image_masks_loss": 0.0082, + "llm_learning_rate": 5.506111219744093e-13, + "loss": 0.723, + "output_text_mask_loss": 0.0116, + "predictor_learning_rate": 5.506111219744092e-11, + "step": 9758 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1, + "image_masks_loss": 0.0109, + "llm_learning_rate": 1.376527841712161e-13, + "loss": 0.7294, + "output_text_mask_loss": 0.0399, + "predictor_learning_rate": 1.3765278417121608e-11, + "step": 9759 + }, + { + "epoch": 1.0, + "gumbel_tau": 0.1, + "image_masks_loss": 0.0053, + "llm_learning_rate": 0.0, + "loss": 0.6993, + "output_text_mask_loss": 0.0038, + "predictor_learning_rate": 0.0, + "step": 9760 + }, + { + "epoch": 1.0, + "step": 9760, + "total_flos": 1446408346820608.0, + "train_loss": 0.7873950216980254, + "train_runtime": 63786.8455, + "train_samples_per_second": 9.792, + "train_steps_per_second": 0.153 + } + ], + "logging_steps": 1.0, + "max_steps": 9760, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 40000, + "total_flos": 1446408346820608.0, + "train_batch_size": 8, + "trial_name": null, + "trial_params": null +}