dummy-checkpoints / trainer_state.json
sabaa96's picture
Add files using upload-large-folder tool
2e0d91a verified
Raw
History Blame Contribute Delete
79.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.99709911361805,
"eval_steps": 100,
"global_step": 12400,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.032232070910556,
"grad_norm": 8.376817069736529,
"learning_rate": 2.222222222222222e-06,
"loss": 3.8567,
"step": 100
},
{
"epoch": 0.032232070910556,
"eval_loss": 3.3241677284240723,
"eval_runtime": 43.0421,
"eval_samples_per_second": 23.233,
"eval_steps_per_second": 5.808,
"eval_text_contr_loss": 0.40461004638671877,
"eval_total_custom_loss": 6.231916015625,
"eval_total_image_loss": 7.38401416015625,
"eval_total_text_loss": 1.925572509765625,
"eval_vision_contr_loss": 5.827306640625,
"step": 100
},
{
"epoch": 0.064464141821112,
"grad_norm": 5.36489946474968,
"learning_rate": 4.444444444444444e-06,
"loss": 2.8427,
"step": 200
},
{
"epoch": 0.064464141821112,
"eval_loss": 2.458766222000122,
"eval_runtime": 43.0713,
"eval_samples_per_second": 23.217,
"eval_steps_per_second": 5.804,
"eval_text_contr_loss": 0.17913763427734375,
"eval_total_custom_loss": 4.625837890625,
"eval_total_image_loss": 5.636779296875,
"eval_total_text_loss": 0.8462789306640625,
"eval_vision_contr_loss": 4.4467001953125,
"step": 200
},
{
"epoch": 0.09669621273166801,
"grad_norm": 6.147009169947269,
"learning_rate": 6.666666666666667e-06,
"loss": 2.4532,
"step": 300
},
{
"epoch": 0.09669621273166801,
"eval_loss": 2.3321802616119385,
"eval_runtime": 43.3413,
"eval_samples_per_second": 23.073,
"eval_steps_per_second": 5.768,
"eval_text_contr_loss": 0.15445872497558594,
"eval_total_custom_loss": 4.38212109375,
"eval_total_image_loss": 5.35918359375,
"eval_total_text_loss": 0.7293816528320313,
"eval_vision_contr_loss": 4.227662109375,
"step": 300
},
{
"epoch": 0.128928283642224,
"grad_norm": 3.5097872406813813,
"learning_rate": 8.888888888888888e-06,
"loss": 2.3181,
"step": 400
},
{
"epoch": 0.128928283642224,
"eval_loss": 2.226626396179199,
"eval_runtime": 43.1188,
"eval_samples_per_second": 23.192,
"eval_steps_per_second": 5.798,
"eval_text_contr_loss": 0.14448023986816405,
"eval_total_custom_loss": 4.18251220703125,
"eval_total_image_loss": 5.11893896484375,
"eval_total_text_loss": 0.68219287109375,
"eval_vision_contr_loss": 4.03803125,
"step": 400
},
{
"epoch": 0.16116035455278002,
"grad_norm": 4.265746378158,
"learning_rate": 9.999755224249292e-06,
"loss": 2.2306,
"step": 500
},
{
"epoch": 0.16116035455278002,
"eval_loss": 2.111872911453247,
"eval_runtime": 43.1542,
"eval_samples_per_second": 23.173,
"eval_steps_per_second": 5.793,
"eval_text_contr_loss": 0.13938406372070314,
"eval_total_custom_loss": 3.968442626953125,
"eval_total_image_loss": 4.85413525390625,
"eval_total_text_loss": 0.658082275390625,
"eval_vision_contr_loss": 3.829057861328125,
"step": 500
},
{
"epoch": 0.19339242546333602,
"grad_norm": 4.544534340107222,
"learning_rate": 9.997797178014507e-06,
"loss": 2.1246,
"step": 600
},
{
"epoch": 0.19339242546333602,
"eval_loss": 2.0428073406219482,
"eval_runtime": 43.371,
"eval_samples_per_second": 23.057,
"eval_steps_per_second": 5.764,
"eval_text_contr_loss": 0.13477769470214843,
"eval_total_custom_loss": 3.8388486328125,
"eval_total_image_loss": 4.69573974609375,
"eval_total_text_loss": 0.6362749633789062,
"eval_vision_contr_loss": 3.704071044921875,
"step": 600
},
{
"epoch": 0.22562449637389204,
"grad_norm": 3.5412226982166963,
"learning_rate": 9.993881937579076e-06,
"loss": 2.0743,
"step": 700
},
{
"epoch": 0.22562449637389204,
"eval_loss": 1.9881960153579712,
"eval_runtime": 43.7215,
"eval_samples_per_second": 22.872,
"eval_steps_per_second": 5.718,
"eval_text_contr_loss": 0.13119364929199218,
"eval_total_custom_loss": 3.736088623046875,
"eval_total_image_loss": 4.57005029296875,
"eval_total_text_loss": 0.6193148193359375,
"eval_vision_contr_loss": 3.60489501953125,
"step": 700
},
{
"epoch": 0.257856567284448,
"grad_norm": 4.897001360182838,
"learning_rate": 9.98801120664051e-06,
"loss": 2.0124,
"step": 800
},
{
"epoch": 0.257856567284448,
"eval_loss": 1.964741587638855,
"eval_runtime": 43.0246,
"eval_samples_per_second": 23.242,
"eval_steps_per_second": 5.811,
"eval_text_contr_loss": 0.12790821838378907,
"eval_total_custom_loss": 3.69105322265625,
"eval_total_image_loss": 4.5171328125,
"eval_total_text_loss": 0.6036654663085937,
"eval_vision_contr_loss": 3.563144775390625,
"step": 800
},
{
"epoch": 0.290088638195004,
"grad_norm": 4.196177043276632,
"learning_rate": 9.980187539818349e-06,
"loss": 1.979,
"step": 900
},
{
"epoch": 0.290088638195004,
"eval_loss": 1.9333972930908203,
"eval_runtime": 43.6492,
"eval_samples_per_second": 22.91,
"eval_steps_per_second": 5.727,
"eval_text_contr_loss": 0.12648372650146483,
"eval_total_custom_loss": 3.63245361328125,
"eval_total_image_loss": 4.444640625,
"eval_total_text_loss": 0.5969776000976562,
"eval_vision_contr_loss": 3.505969482421875,
"step": 900
},
{
"epoch": 0.32232070910556004,
"grad_norm": 2.517458079297082,
"learning_rate": 9.970414341542523e-06,
"loss": 1.9419,
"step": 1000
},
{
"epoch": 0.32232070910556004,
"eval_loss": 1.9163663387298584,
"eval_runtime": 42.8571,
"eval_samples_per_second": 23.333,
"eval_steps_per_second": 5.833,
"eval_text_contr_loss": 0.12434061431884766,
"eval_total_custom_loss": 3.599889404296875,
"eval_total_image_loss": 4.4060947265625,
"eval_total_text_loss": 0.5867772827148438,
"eval_vision_contr_loss": 3.47554931640625,
"step": 1000
},
{
"epoch": 0.354552780016116,
"grad_norm": 2.239292209724041,
"learning_rate": 9.958695864571933e-06,
"loss": 1.9184,
"step": 1100
},
{
"epoch": 0.354552780016116,
"eval_loss": 1.8987705707550049,
"eval_runtime": 43.2237,
"eval_samples_per_second": 23.135,
"eval_steps_per_second": 5.784,
"eval_text_contr_loss": 0.12296145629882813,
"eval_total_custom_loss": 3.5664951171875,
"eval_total_image_loss": 4.3655224609375,
"eval_total_text_loss": 0.5803037109375,
"eval_vision_contr_loss": 3.44353466796875,
"step": 1100
},
{
"epoch": 0.38678485092667203,
"grad_norm": 2.6680379186401098,
"learning_rate": 9.945037208143885e-06,
"loss": 1.9488,
"step": 1200
},
{
"epoch": 0.38678485092667203,
"eval_loss": 1.8872787952423096,
"eval_runtime": 43.681,
"eval_samples_per_second": 22.893,
"eval_steps_per_second": 5.723,
"eval_text_contr_loss": 0.12142788696289063,
"eval_total_custom_loss": 3.544423828125,
"eval_total_image_loss": 4.3394892578125,
"eval_total_text_loss": 0.57302685546875,
"eval_vision_contr_loss": 3.42299609375,
"step": 1200
},
{
"epoch": 0.41901692183722805,
"grad_norm": 2.8407509665859063,
"learning_rate": 9.929444315755184e-06,
"loss": 1.8984,
"step": 1300
},
{
"epoch": 0.41901692183722805,
"eval_loss": 1.8737540245056152,
"eval_runtime": 42.7457,
"eval_samples_per_second": 23.394,
"eval_steps_per_second": 5.849,
"eval_text_contr_loss": 0.12054175567626953,
"eval_total_custom_loss": 3.5189599609375,
"eval_total_image_loss": 4.3083427734375,
"eval_total_text_loss": 0.5688414916992187,
"eval_vision_contr_loss": 3.39841845703125,
"step": 1300
},
{
"epoch": 0.4512489927477841,
"grad_norm": 2.7099504564117214,
"learning_rate": 9.911923972575844e-06,
"loss": 1.9214,
"step": 1400
},
{
"epoch": 0.4512489927477841,
"eval_loss": 1.8677526712417603,
"eval_runtime": 42.6477,
"eval_samples_per_second": 23.448,
"eval_steps_per_second": 5.862,
"eval_text_contr_loss": 0.11904737091064453,
"eval_total_custom_loss": 3.506833251953125,
"eval_total_image_loss": 4.29482470703125,
"eval_total_text_loss": 0.5617998046875,
"eval_vision_contr_loss": 3.3877861328125,
"step": 1400
},
{
"epoch": 0.48348106365834004,
"grad_norm": 2.1559236619034126,
"learning_rate": 9.892483802496567e-06,
"loss": 1.8909,
"step": 1500
},
{
"epoch": 0.48348106365834004,
"eval_loss": 1.8571125268936157,
"eval_runtime": 42.6701,
"eval_samples_per_second": 23.436,
"eval_steps_per_second": 5.859,
"eval_text_contr_loss": 0.11790892791748046,
"eval_total_custom_loss": 3.486457275390625,
"eval_total_image_loss": 4.270455078125,
"eval_total_text_loss": 0.556384521484375,
"eval_vision_contr_loss": 3.36854736328125,
"step": 1500
},
{
"epoch": 0.515713134568896,
"grad_norm": 4.470842059114152,
"learning_rate": 9.87113226481123e-06,
"loss": 1.8716,
"step": 1600
},
{
"epoch": 0.515713134568896,
"eval_loss": 1.8490651845932007,
"eval_runtime": 42.803,
"eval_samples_per_second": 23.363,
"eval_steps_per_second": 5.841,
"eval_text_contr_loss": 0.11713390350341797,
"eval_total_custom_loss": 3.471124755859375,
"eval_total_image_loss": 4.25204296875,
"eval_total_text_loss": 0.5526325073242188,
"eval_vision_contr_loss": 3.35399072265625,
"step": 1600
},
{
"epoch": 0.547945205479452,
"grad_norm": 2.577700492879793,
"learning_rate": 9.847878650535872e-06,
"loss": 1.8718,
"step": 1700
},
{
"epoch": 0.547945205479452,
"eval_loss": 1.8381274938583374,
"eval_runtime": 43.4452,
"eval_samples_per_second": 23.017,
"eval_steps_per_second": 5.754,
"eval_text_contr_loss": 0.11611717987060546,
"eval_total_custom_loss": 3.450317626953125,
"eval_total_image_loss": 4.2269052734375,
"eval_total_text_loss": 0.5478128662109375,
"eval_vision_contr_loss": 3.334201171875,
"step": 1700
},
{
"epoch": 0.580177276390008,
"grad_norm": 2.2420338794696875,
"learning_rate": 9.822733078365759e-06,
"loss": 1.8512,
"step": 1800
},
{
"epoch": 0.580177276390008,
"eval_loss": 1.8326914310455322,
"eval_runtime": 43.0625,
"eval_samples_per_second": 23.222,
"eval_steps_per_second": 5.806,
"eval_text_contr_loss": 0.11557137298583985,
"eval_total_custom_loss": 3.439968017578125,
"eval_total_image_loss": 4.21451025390625,
"eval_total_text_loss": 0.5452426147460937,
"eval_vision_contr_loss": 3.324396240234375,
"step": 1800
},
{
"epoch": 0.6124093473005641,
"grad_norm": 3.2369310814743195,
"learning_rate": 9.79570649027228e-06,
"loss": 1.858,
"step": 1900
},
{
"epoch": 0.6124093473005641,
"eval_loss": 1.8280653953552246,
"eval_runtime": 43.1494,
"eval_samples_per_second": 23.175,
"eval_steps_per_second": 5.794,
"eval_text_contr_loss": 0.11462459564208985,
"eval_total_custom_loss": 3.43068505859375,
"eval_total_image_loss": 4.20393994140625,
"eval_total_text_loss": 0.5407430419921875,
"eval_vision_contr_loss": 3.316060302734375,
"step": 1900
},
{
"epoch": 0.6446414182111201,
"grad_norm": 2.1087018207267123,
"learning_rate": 9.766810646741595e-06,
"loss": 1.8538,
"step": 2000
},
{
"epoch": 0.6446414182111201,
"eval_loss": 1.8192895650863647,
"eval_runtime": 43.4977,
"eval_samples_per_second": 22.99,
"eval_steps_per_second": 5.747,
"eval_text_contr_loss": 0.11377765655517579,
"eval_total_custom_loss": 3.414066650390625,
"eval_total_image_loss": 4.1839619140625,
"eval_total_text_loss": 0.5367439575195313,
"eval_vision_contr_loss": 3.3002890625,
"step": 2000
},
{
"epoch": 0.6768734891216761,
"grad_norm": 1.7994274749762051,
"learning_rate": 9.736058121657131e-06,
"loss": 1.8419,
"step": 2100
},
{
"epoch": 0.6768734891216761,
"eval_loss": 1.8163025379180908,
"eval_runtime": 43.2677,
"eval_samples_per_second": 23.112,
"eval_steps_per_second": 5.778,
"eval_text_contr_loss": 0.11306369018554688,
"eval_total_custom_loss": 3.408015625,
"eval_total_image_loss": 4.1771630859375,
"eval_total_text_loss": 0.533274169921875,
"eval_vision_contr_loss": 3.294951904296875,
"step": 2100
},
{
"epoch": 0.709105560032232,
"grad_norm": 1.5532153517574525,
"learning_rate": 9.703462296828106e-06,
"loss": 1.8411,
"step": 2200
},
{
"epoch": 0.709105560032232,
"eval_loss": 1.8098113536834717,
"eval_runtime": 43.1134,
"eval_samples_per_second": 23.195,
"eval_steps_per_second": 5.799,
"eval_text_contr_loss": 0.11252593994140625,
"eval_total_custom_loss": 3.39573583984375,
"eval_total_image_loss": 4.16230810546875,
"eval_total_text_loss": 0.5307877197265625,
"eval_vision_contr_loss": 3.28320947265625,
"step": 2200
},
{
"epoch": 0.741337630942788,
"grad_norm": 2.6894191960932177,
"learning_rate": 9.669037356166514e-06,
"loss": 1.8233,
"step": 2300
},
{
"epoch": 0.741337630942788,
"eval_loss": 1.8054653406143188,
"eval_runtime": 43.2697,
"eval_samples_per_second": 23.111,
"eval_steps_per_second": 5.778,
"eval_text_contr_loss": 0.1121211166381836,
"eval_total_custom_loss": 3.3874794921875,
"eval_total_image_loss": 4.1523466796875,
"eval_total_text_loss": 0.5288152465820313,
"eval_vision_contr_loss": 3.2753583984375,
"step": 2300
},
{
"epoch": 0.7735697018533441,
"grad_norm": 1.7558102211650422,
"learning_rate": 9.632798279515048e-06,
"loss": 1.8282,
"step": 2400
},
{
"epoch": 0.7735697018533441,
"eval_loss": 1.7984919548034668,
"eval_runtime": 43.0473,
"eval_samples_per_second": 23.23,
"eval_steps_per_second": 5.808,
"eval_text_contr_loss": 0.11167151641845703,
"eval_total_custom_loss": 3.3743779296875,
"eval_total_image_loss": 4.1363369140625,
"eval_total_text_loss": 0.5266166381835937,
"eval_vision_contr_loss": 3.2627060546875,
"step": 2400
},
{
"epoch": 0.8058017727639001,
"grad_norm": 2.189718486325206,
"learning_rate": 9.594760836128718e-06,
"loss": 1.8272,
"step": 2500
},
{
"epoch": 0.8058017727639001,
"eval_loss": 1.796999454498291,
"eval_runtime": 43.2243,
"eval_samples_per_second": 23.135,
"eval_steps_per_second": 5.784,
"eval_text_contr_loss": 0.11117787170410157,
"eval_total_custom_loss": 3.3712236328125,
"eval_total_image_loss": 4.132939453125,
"eval_total_text_loss": 0.5243654174804687,
"eval_vision_contr_loss": 3.260046142578125,
"step": 2500
},
{
"epoch": 0.8380338436744561,
"grad_norm": 2.07083629988929,
"learning_rate": 9.55494157781292e-06,
"loss": 1.828,
"step": 2600
},
{
"epoch": 0.8380338436744561,
"eval_loss": 1.7920169830322266,
"eval_runtime": 43.0701,
"eval_samples_per_second": 23.218,
"eval_steps_per_second": 5.804,
"eval_text_contr_loss": 0.11067650604248047,
"eval_total_custom_loss": 3.3617353515625,
"eval_total_image_loss": 4.1215615234375,
"eval_total_text_loss": 0.5220291748046875,
"eval_vision_contr_loss": 3.25105859375,
"step": 2600
},
{
"epoch": 0.8702659145850121,
"grad_norm": 2.437655358902012,
"learning_rate": 9.513357831721002e-06,
"loss": 1.816,
"step": 2700
},
{
"epoch": 0.8702659145850121,
"eval_loss": 1.7867939472198486,
"eval_runtime": 43.0641,
"eval_samples_per_second": 23.221,
"eval_steps_per_second": 5.805,
"eval_text_contr_loss": 0.11017485046386719,
"eval_total_custom_loss": 3.35179833984375,
"eval_total_image_loss": 4.1096103515625,
"eval_total_text_loss": 0.519625732421875,
"eval_vision_contr_loss": 3.241623291015625,
"step": 2700
},
{
"epoch": 0.9024979854955681,
"grad_norm": 1.9142874919425596,
"learning_rate": 9.470027692814453e-06,
"loss": 1.7628,
"step": 2800
},
{
"epoch": 0.9024979854955681,
"eval_loss": 1.7855441570281982,
"eval_runtime": 43.2125,
"eval_samples_per_second": 23.141,
"eval_steps_per_second": 5.785,
"eval_text_contr_loss": 0.10965695190429688,
"eval_total_custom_loss": 3.349080078125,
"eval_total_image_loss": 4.10679736328125,
"eval_total_text_loss": 0.5172005615234375,
"eval_vision_contr_loss": 3.2394228515625,
"step": 2800
},
{
"epoch": 0.934730056406124,
"grad_norm": 1.8433712863938758,
"learning_rate": 9.424970015988945e-06,
"loss": 1.8155,
"step": 2900
},
{
"epoch": 0.934730056406124,
"eval_loss": 1.7826323509216309,
"eval_runtime": 43.4344,
"eval_samples_per_second": 23.023,
"eval_steps_per_second": 5.756,
"eval_text_contr_loss": 0.10918792724609375,
"eval_total_custom_loss": 3.34338037109375,
"eval_total_image_loss": 4.100208984375,
"eval_total_text_loss": 0.5150006103515625,
"eval_vision_contr_loss": 3.23419287109375,
"step": 2900
},
{
"epoch": 0.9669621273166801,
"grad_norm": 1.7431036135091218,
"learning_rate": 9.378204407869748e-06,
"loss": 1.7956,
"step": 3000
},
{
"epoch": 0.9669621273166801,
"eval_loss": 1.7770920991897583,
"eval_runtime": 43.1363,
"eval_samples_per_second": 23.182,
"eval_steps_per_second": 5.796,
"eval_text_contr_loss": 0.108552001953125,
"eval_total_custom_loss": 3.3327802734375,
"eval_total_image_loss": 4.08755419921875,
"eval_total_text_loss": 0.5119426574707031,
"eval_vision_contr_loss": 3.224228759765625,
"step": 3000
},
{
"epoch": 0.9991941982272361,
"grad_norm": 2.1897189823570646,
"learning_rate": 9.329751218279976e-06,
"loss": 1.8034,
"step": 3100
},
{
"epoch": 0.9991941982272361,
"eval_loss": 1.7757591009140015,
"eval_runtime": 43.4575,
"eval_samples_per_second": 23.011,
"eval_steps_per_second": 5.753,
"eval_text_contr_loss": 0.10834796142578125,
"eval_total_custom_loss": 3.33019140625,
"eval_total_image_loss": 4.084512451171875,
"eval_total_text_loss": 0.5109953002929688,
"eval_vision_contr_loss": 3.22184375,
"step": 3100
},
{
"epoch": 1.031426269137792,
"grad_norm": 1.6481602504338035,
"learning_rate": 9.279631531385484e-06,
"loss": 1.7484,
"step": 3200
},
{
"epoch": 1.031426269137792,
"eval_loss": 1.7711955308914185,
"eval_runtime": 43.1879,
"eval_samples_per_second": 23.155,
"eval_steps_per_second": 5.789,
"eval_text_contr_loss": 0.10868955993652343,
"eval_total_custom_loss": 3.3221103515625,
"eval_total_image_loss": 4.0738515625,
"eval_total_text_loss": 0.5125972900390625,
"eval_vision_contr_loss": 3.213421142578125,
"step": 3200
},
{
"epoch": 1.0636583400483481,
"grad_norm": 1.7900383558924777,
"learning_rate": 9.227867156520188e-06,
"loss": 1.7429,
"step": 3300
},
{
"epoch": 1.0636583400483481,
"eval_loss": 1.7699233293533325,
"eval_runtime": 43.0499,
"eval_samples_per_second": 23.229,
"eval_steps_per_second": 5.807,
"eval_text_contr_loss": 0.10875718688964844,
"eval_total_custom_loss": 3.31983349609375,
"eval_total_image_loss": 4.0709267578125,
"eval_total_text_loss": 0.51294189453125,
"eval_vision_contr_loss": 3.21107666015625,
"step": 3300
},
{
"epoch": 1.095890410958904,
"grad_norm": 1.4167928322005499,
"learning_rate": 9.17448061869583e-06,
"loss": 1.7704,
"step": 3400
},
{
"epoch": 1.095890410958904,
"eval_loss": 1.769239068031311,
"eval_runtime": 43.3181,
"eval_samples_per_second": 23.085,
"eval_steps_per_second": 5.771,
"eval_text_contr_loss": 0.10824945068359375,
"eval_total_custom_loss": 3.31817138671875,
"eval_total_image_loss": 4.069458251953125,
"eval_total_text_loss": 0.51047802734375,
"eval_vision_contr_loss": 3.20992236328125,
"step": 3400
},
{
"epoch": 1.1281224818694602,
"grad_norm": 1.859931203369117,
"learning_rate": 9.119495148800357e-06,
"loss": 1.7449,
"step": 3500
},
{
"epoch": 1.1281224818694602,
"eval_loss": 1.763574481010437,
"eval_runtime": 43.105,
"eval_samples_per_second": 23.199,
"eval_steps_per_second": 5.8,
"eval_text_contr_loss": 0.10772032928466797,
"eval_total_custom_loss": 3.30740576171875,
"eval_total_image_loss": 4.0564755859375,
"eval_total_text_loss": 0.50800439453125,
"eval_vision_contr_loss": 3.199684814453125,
"step": 3500
},
{
"epoch": 1.160354552780016,
"grad_norm": 2.068989185774414,
"learning_rate": 9.062934673489092e-06,
"loss": 1.7687,
"step": 3600
},
{
"epoch": 1.160354552780016,
"eval_loss": 1.7613393068313599,
"eval_runtime": 42.9491,
"eval_samples_per_second": 23.283,
"eval_steps_per_second": 5.821,
"eval_text_contr_loss": 0.10758199310302734,
"eval_total_custom_loss": 3.303072265625,
"eval_total_image_loss": 4.05116064453125,
"eval_total_text_loss": 0.5073269958496094,
"eval_vision_contr_loss": 3.195490234375,
"step": 3600
},
{
"epoch": 1.192586623690572,
"grad_norm": 2.141082949004269,
"learning_rate": 9.004823804773181e-06,
"loss": 1.7297,
"step": 3700
},
{
"epoch": 1.192586623690572,
"eval_loss": 1.7592214345932007,
"eval_runtime": 43.069,
"eval_samples_per_second": 23.219,
"eval_steps_per_second": 5.805,
"eval_text_contr_loss": 0.10733989715576171,
"eval_total_custom_loss": 3.2991220703125,
"eval_total_image_loss": 4.046507080078125,
"eval_total_text_loss": 0.5061781921386719,
"eval_vision_contr_loss": 3.1917822265625,
"step": 3700
},
{
"epoch": 1.2248186946011281,
"grad_norm": 1.8450064802531847,
"learning_rate": 8.945187829309785e-06,
"loss": 1.7589,
"step": 3800
},
{
"epoch": 1.2248186946011281,
"eval_loss": 1.7595419883728027,
"eval_runtime": 43.6782,
"eval_samples_per_second": 22.895,
"eval_steps_per_second": 5.724,
"eval_text_contr_loss": 0.10679227447509766,
"eval_total_custom_loss": 3.29919677734375,
"eval_total_image_loss": 4.04727490234375,
"eval_total_text_loss": 0.5035033569335937,
"eval_vision_contr_loss": 3.192405029296875,
"step": 3800
},
{
"epoch": 1.257050765511684,
"grad_norm": 1.5670039690273154,
"learning_rate": 8.884052697398737e-06,
"loss": 1.7627,
"step": 3900
},
{
"epoch": 1.257050765511684,
"eval_loss": 1.7528612613677979,
"eval_runtime": 43.1783,
"eval_samples_per_second": 23.16,
"eval_steps_per_second": 5.79,
"eval_text_contr_loss": 0.10662190246582032,
"eval_total_custom_loss": 3.286930908203125,
"eval_total_image_loss": 4.031930908203125,
"eval_total_text_loss": 0.502760498046875,
"eval_vision_contr_loss": 3.180308349609375,
"step": 3900
},
{
"epoch": 1.2892828364222402,
"grad_norm": 1.5003342439172258,
"learning_rate": 8.82144501169037e-06,
"loss": 1.7692,
"step": 4000
},
{
"epoch": 1.2892828364222402,
"eval_loss": 1.7518749237060547,
"eval_runtime": 43.142,
"eval_samples_per_second": 23.179,
"eval_steps_per_second": 5.795,
"eval_text_contr_loss": 0.10632988739013671,
"eval_total_custom_loss": 3.2848466796875,
"eval_total_image_loss": 4.02964599609375,
"eval_total_text_loss": 0.501397216796875,
"eval_vision_contr_loss": 3.178516357421875,
"step": 4000
},
{
"epoch": 1.321514907332796,
"grad_norm": 1.588181678069688,
"learning_rate": 8.757392015609536e-06,
"loss": 1.7336,
"step": 4100
},
{
"epoch": 1.321514907332796,
"eval_loss": 1.7515907287597656,
"eval_runtime": 43.2497,
"eval_samples_per_second": 23.122,
"eval_steps_per_second": 5.78,
"eval_text_contr_loss": 0.10626918029785157,
"eval_total_custom_loss": 3.284319091796875,
"eval_total_image_loss": 4.029078125,
"eval_total_text_loss": 0.5010674743652344,
"eval_vision_contr_loss": 3.17805078125,
"step": 4100
},
{
"epoch": 1.3537469782433522,
"grad_norm": 1.7924143802949446,
"learning_rate": 8.691921581500737e-06,
"loss": 1.7931,
"step": 4200
},
{
"epoch": 1.3537469782433522,
"eval_loss": 1.746930718421936,
"eval_runtime": 42.8299,
"eval_samples_per_second": 23.348,
"eval_steps_per_second": 5.837,
"eval_text_contr_loss": 0.10574241638183594,
"eval_total_custom_loss": 3.27539111328125,
"eval_total_image_loss": 4.018408203125,
"eval_total_text_loss": 0.4985403747558594,
"eval_vision_contr_loss": 3.16964892578125,
"step": 4200
},
{
"epoch": 1.3859790491539081,
"grad_norm": 1.2696679831034872,
"learning_rate": 8.62506219849963e-06,
"loss": 1.747,
"step": 4300
},
{
"epoch": 1.3859790491539081,
"eval_loss": 1.7456879615783691,
"eval_runtime": 42.9935,
"eval_samples_per_second": 23.259,
"eval_steps_per_second": 5.815,
"eval_text_contr_loss": 0.10543916320800781,
"eval_total_custom_loss": 3.272862548828125,
"eval_total_image_loss": 4.0155888671875,
"eval_total_text_loss": 0.49712225341796873,
"eval_vision_contr_loss": 3.167423095703125,
"step": 4300
},
{
"epoch": 1.418211120064464,
"grad_norm": 1.4038670024142585,
"learning_rate": 8.556842960136109e-06,
"loss": 1.7542,
"step": 4400
},
{
"epoch": 1.418211120064464,
"eval_loss": 1.7419202327728271,
"eval_runtime": 43.1291,
"eval_samples_per_second": 23.186,
"eval_steps_per_second": 5.797,
"eval_text_contr_loss": 0.10515576171875,
"eval_total_custom_loss": 3.2657568359375,
"eval_total_image_loss": 4.006947265625,
"eval_total_text_loss": 0.49575506591796875,
"eval_vision_contr_loss": 3.160601318359375,
"step": 4400
},
{
"epoch": 1.4504431909750202,
"grad_norm": 1.553147953245719,
"learning_rate": 8.487293551674407e-06,
"loss": 1.7504,
"step": 4500
},
{
"epoch": 1.4504431909750202,
"eval_loss": 1.740807056427002,
"eval_runtime": 42.8972,
"eval_samples_per_second": 23.312,
"eval_steps_per_second": 5.828,
"eval_text_contr_loss": 0.10501337432861328,
"eval_total_custom_loss": 3.263591796875,
"eval_total_image_loss": 4.004369140625,
"eval_total_text_loss": 0.495142578125,
"eval_vision_contr_loss": 3.15857861328125,
"step": 4500
},
{
"epoch": 1.482675261885576,
"grad_norm": 1.5490326450615954,
"learning_rate": 8.416444237195702e-06,
"loss": 1.754,
"step": 4600
},
{
"epoch": 1.482675261885576,
"eval_loss": 1.7383344173431396,
"eval_runtime": 43.0597,
"eval_samples_per_second": 23.224,
"eval_steps_per_second": 5.806,
"eval_text_contr_loss": 0.10456053924560547,
"eval_total_custom_loss": 3.25873291015625,
"eval_total_image_loss": 3.99877001953125,
"eval_total_text_loss": 0.4929710693359375,
"eval_vision_contr_loss": 3.154172119140625,
"step": 4600
},
{
"epoch": 1.514907332796132,
"grad_norm": 1.306313286524459,
"learning_rate": 8.34432584642884e-06,
"loss": 1.7429,
"step": 4700
},
{
"epoch": 1.514907332796132,
"eval_loss": 1.7368396520614624,
"eval_runtime": 43.2244,
"eval_samples_per_second": 23.135,
"eval_steps_per_second": 5.784,
"eval_text_contr_loss": 0.10414806365966797,
"eval_total_custom_loss": 3.2556689453125,
"eval_total_image_loss": 3.99540576171875,
"eval_total_text_loss": 0.49097271728515623,
"eval_vision_contr_loss": 3.151520263671875,
"step": 4700
},
{
"epoch": 1.5471394037066881,
"grad_norm": 1.2887661649855016,
"learning_rate": 8.270969761334946e-06,
"loss": 1.7619,
"step": 4800
},
{
"epoch": 1.5471394037066881,
"eval_loss": 1.7330701351165771,
"eval_runtime": 43.0538,
"eval_samples_per_second": 23.227,
"eval_steps_per_second": 5.807,
"eval_text_contr_loss": 0.10417289733886718,
"eval_total_custom_loss": 3.24881494140625,
"eval_total_image_loss": 3.98670556640625,
"eval_total_text_loss": 0.4910880126953125,
"eval_vision_contr_loss": 3.144642822265625,
"step": 4800
},
{
"epoch": 1.5793714746172443,
"grad_norm": 1.3604255924953426,
"learning_rate": 8.1964079024517e-06,
"loss": 1.7679,
"step": 4900
},
{
"epoch": 1.5793714746172443,
"eval_loss": 1.729392170906067,
"eval_runtime": 43.2858,
"eval_samples_per_second": 23.102,
"eval_steps_per_second": 5.776,
"eval_text_contr_loss": 0.10385750579833984,
"eval_total_custom_loss": 3.241834716796875,
"eval_total_image_loss": 3.97827783203125,
"eval_total_text_loss": 0.4896236572265625,
"eval_vision_contr_loss": 3.137977294921875,
"step": 4900
},
{
"epoch": 1.6116035455278002,
"grad_norm": 1.1049881715481917,
"learning_rate": 8.120672715003294e-06,
"loss": 1.7428,
"step": 5000
},
{
"epoch": 1.6116035455278002,
"eval_loss": 1.7302324771881104,
"eval_runtime": 43.5243,
"eval_samples_per_second": 22.976,
"eval_steps_per_second": 5.744,
"eval_text_contr_loss": 0.10343788909912109,
"eval_total_custom_loss": 3.243067138671875,
"eval_total_image_loss": 3.9803583984375,
"eval_total_text_loss": 0.4876519775390625,
"eval_vision_contr_loss": 3.13962939453125,
"step": 5000
},
{
"epoch": 1.643835616438356,
"grad_norm": 1.4871373362175029,
"learning_rate": 8.043797154782041e-06,
"loss": 1.7438,
"step": 5100
},
{
"epoch": 1.643835616438356,
"eval_loss": 1.7294504642486572,
"eval_runtime": 43.2467,
"eval_samples_per_second": 23.123,
"eval_steps_per_second": 5.781,
"eval_text_contr_loss": 0.10337482452392578,
"eval_total_custom_loss": 3.24157080078125,
"eval_total_image_loss": 3.978525634765625,
"eval_total_text_loss": 0.4873544921875,
"eval_vision_contr_loss": 3.138195556640625,
"step": 5100
},
{
"epoch": 1.6760676873489122,
"grad_norm": 1.4862996756725702,
"learning_rate": 7.965814673807827e-06,
"loss": 1.7258,
"step": 5200
},
{
"epoch": 1.6760676873489122,
"eval_loss": 1.7246971130371094,
"eval_runtime": 43.1195,
"eval_samples_per_second": 23.191,
"eval_steps_per_second": 5.798,
"eval_text_contr_loss": 0.10332577514648437,
"eval_total_custom_loss": 3.232852294921875,
"eval_total_image_loss": 3.96752294921875,
"eval_total_text_loss": 0.4871480712890625,
"eval_vision_contr_loss": 3.12952587890625,
"step": 5200
},
{
"epoch": 1.7082997582594683,
"grad_norm": 1.412355388159593,
"learning_rate": 7.886759205771625e-06,
"loss": 1.7256,
"step": 5300
},
{
"epoch": 1.7082997582594683,
"eval_loss": 1.7225898504257202,
"eval_runtime": 42.8073,
"eval_samples_per_second": 23.361,
"eval_steps_per_second": 5.84,
"eval_text_contr_loss": 0.10268486022949219,
"eval_total_custom_loss": 3.228478759765625,
"eval_total_image_loss": 3.96278955078125,
"eval_total_text_loss": 0.48410064697265626,
"eval_vision_contr_loss": 3.12579345703125,
"step": 5300
},
{
"epoch": 1.740531829170024,
"grad_norm": 1.7627888517566388,
"learning_rate": 7.806665151269424e-06,
"loss": 1.7288,
"step": 5400
},
{
"epoch": 1.740531829170024,
"eval_loss": 1.722500205039978,
"eval_runtime": 42.9419,
"eval_samples_per_second": 23.287,
"eval_steps_per_second": 5.822,
"eval_text_contr_loss": 0.10271249389648437,
"eval_total_custom_loss": 3.228342041015625,
"eval_total_image_loss": 3.96258447265625,
"eval_total_text_loss": 0.4842587890625,
"eval_vision_contr_loss": 3.1256298828125,
"step": 5400
},
{
"epoch": 1.7727639000805802,
"grad_norm": 1.1625897768555602,
"learning_rate": 7.725567362832988e-06,
"loss": 1.7119,
"step": 5500
},
{
"epoch": 1.7727639000805802,
"eval_loss": 1.7217448949813843,
"eval_runtime": 43.5739,
"eval_samples_per_second": 22.95,
"eval_steps_per_second": 5.737,
"eval_text_contr_loss": 0.102185302734375,
"eval_total_custom_loss": 3.22654345703125,
"eval_total_image_loss": 3.960973876953125,
"eval_total_text_loss": 0.48181881713867186,
"eval_vision_contr_loss": 3.124358154296875,
"step": 5500
},
{
"epoch": 1.8049959709911363,
"grad_norm": 1.2532474256069803,
"learning_rate": 7.643501129763925e-06,
"loss": 1.7249,
"step": 5600
},
{
"epoch": 1.8049959709911363,
"eval_loss": 1.719870686531067,
"eval_runtime": 43.0446,
"eval_samples_per_second": 23.232,
"eval_steps_per_second": 5.808,
"eval_text_contr_loss": 0.1021416015625,
"eval_total_custom_loss": 3.2230703125,
"eval_total_image_loss": 3.95665185546875,
"eval_total_text_loss": 0.48166421508789065,
"eval_vision_contr_loss": 3.120928955078125,
"step": 5600
},
{
"epoch": 1.8372280419016922,
"grad_norm": 1.186401836865202,
"learning_rate": 7.56050216277774e-06,
"loss": 1.7236,
"step": 5700
},
{
"epoch": 1.8372280419016922,
"eval_loss": 1.7167900800704956,
"eval_runtime": 43.0532,
"eval_samples_per_second": 23.227,
"eval_steps_per_second": 5.807,
"eval_text_contr_loss": 0.10181983947753906,
"eval_total_custom_loss": 3.217169677734375,
"eval_total_image_loss": 3.94959619140625,
"eval_total_text_loss": 0.4800431213378906,
"eval_vision_contr_loss": 3.11535009765625,
"step": 5700
},
{
"epoch": 1.869460112812248,
"grad_norm": 1.8611578087256904,
"learning_rate": 7.476606578464497e-06,
"loss": 1.7196,
"step": 5800
},
{
"epoch": 1.869460112812248,
"eval_loss": 1.7161837816238403,
"eval_runtime": 43.0611,
"eval_samples_per_second": 23.223,
"eval_steps_per_second": 5.806,
"eval_text_contr_loss": 0.10173680114746093,
"eval_total_custom_loss": 3.216015380859375,
"eval_total_image_loss": 3.94821630859375,
"eval_total_text_loss": 0.4796446533203125,
"eval_vision_contr_loss": 3.1142783203125,
"step": 5800
},
{
"epoch": 1.9016921837228042,
"grad_norm": 1.1970358284464848,
"learning_rate": 7.391850883572849e-06,
"loss": 1.7154,
"step": 5900
},
{
"epoch": 1.9016921837228042,
"eval_loss": 1.7155755758285522,
"eval_runtime": 43.668,
"eval_samples_per_second": 22.9,
"eval_steps_per_second": 5.725,
"eval_text_contr_loss": 0.101461669921875,
"eval_total_custom_loss": 3.21466650390625,
"eval_total_image_loss": 3.946880859375,
"eval_total_text_loss": 0.4783258056640625,
"eval_vision_contr_loss": 3.113205322265625,
"step": 5900
},
{
"epoch": 1.9339242546333602,
"grad_norm": 1.2468389300551599,
"learning_rate": 7.306271959124313e-06,
"loss": 1.7497,
"step": 6000
},
{
"epoch": 1.9339242546333602,
"eval_loss": 1.711749792098999,
"eval_runtime": 43.0167,
"eval_samples_per_second": 23.247,
"eval_steps_per_second": 5.812,
"eval_text_contr_loss": 0.10137190246582031,
"eval_total_custom_loss": 3.207614990234375,
"eval_total_image_loss": 3.93804150390625,
"eval_total_text_loss": 0.4779098205566406,
"eval_vision_contr_loss": 3.10624267578125,
"step": 6000
},
{
"epoch": 1.966156325543916,
"grad_norm": 1.3879414435610702,
"learning_rate": 7.219907044364683e-06,
"loss": 1.7212,
"step": 6100
},
{
"epoch": 1.966156325543916,
"eval_loss": 1.7115145921707153,
"eval_runtime": 43.2586,
"eval_samples_per_second": 23.117,
"eval_steps_per_second": 5.779,
"eval_text_contr_loss": 0.10102347564697266,
"eval_total_custom_loss": 3.206907958984375,
"eval_total_image_loss": 3.937591796875,
"eval_total_text_loss": 0.476210205078125,
"eval_vision_contr_loss": 3.105884765625,
"step": 6100
},
{
"epoch": 1.9983883964544722,
"grad_norm": 1.2689982880943955,
"learning_rate": 7.132793720559548e-06,
"loss": 1.7293,
"step": 6200
},
{
"epoch": 1.9983883964544722,
"eval_loss": 1.710171103477478,
"eval_runtime": 43.0121,
"eval_samples_per_second": 23.249,
"eval_steps_per_second": 5.812,
"eval_text_contr_loss": 0.100876708984375,
"eval_total_custom_loss": 3.204313232421875,
"eval_total_image_loss": 3.93449755859375,
"eval_total_text_loss": 0.47554855346679686,
"eval_vision_contr_loss": 3.103436767578125,
"step": 6200
},
{
"epoch": 2.0306204673650283,
"grad_norm": 1.3950821678120036,
"learning_rate": 7.044969894640985e-06,
"loss": 1.6634,
"step": 6300
},
{
"epoch": 2.0306204673650283,
"eval_loss": 1.7124313116073608,
"eval_runtime": 43.1043,
"eval_samples_per_second": 23.2,
"eval_steps_per_second": 5.8,
"eval_text_contr_loss": 0.10253661346435547,
"eval_total_custom_loss": 3.209766357421875,
"eval_total_image_loss": 3.93932373046875,
"eval_total_text_loss": 0.4833622741699219,
"eval_vision_contr_loss": 3.10723046875,
"step": 6300
},
{
"epoch": 2.062852538275584,
"grad_norm": 1.6330256517316408,
"learning_rate": 6.9564737827125624e-06,
"loss": 1.654,
"step": 6400
},
{
"epoch": 2.062852538275584,
"eval_loss": 1.7117656469345093,
"eval_runtime": 43.2236,
"eval_samples_per_second": 23.136,
"eval_steps_per_second": 5.784,
"eval_text_contr_loss": 0.10246735382080079,
"eval_total_custom_loss": 3.208496337890625,
"eval_total_image_loss": 3.93780859375,
"eval_total_text_loss": 0.48304425048828126,
"eval_vision_contr_loss": 3.106028564453125,
"step": 6400
},
{
"epoch": 2.09508460918614,
"grad_norm": 1.0726622018398422,
"learning_rate": 6.867343893419779e-06,
"loss": 1.6844,
"step": 6500
},
{
"epoch": 2.09508460918614,
"eval_loss": 1.710357427597046,
"eval_runtime": 43.1642,
"eval_samples_per_second": 23.167,
"eval_steps_per_second": 5.792,
"eval_text_contr_loss": 0.1026227798461914,
"eval_total_custom_loss": 3.206042236328125,
"eval_total_image_loss": 3.93451171875,
"eval_total_text_loss": 0.483800048828125,
"eval_vision_contr_loss": 3.10341943359375,
"step": 6500
},
{
"epoch": 2.1273166800966963,
"grad_norm": 1.1677656833173125,
"learning_rate": 6.777619011193214e-06,
"loss": 1.6812,
"step": 6600
},
{
"epoch": 2.1273166800966963,
"eval_loss": 1.708493709564209,
"eval_runtime": 43.1562,
"eval_samples_per_second": 23.172,
"eval_steps_per_second": 5.793,
"eval_text_contr_loss": 0.10243350219726563,
"eval_total_custom_loss": 3.20248828125,
"eval_total_image_loss": 3.93024072265625,
"eval_total_text_loss": 0.482855224609375,
"eval_vision_contr_loss": 3.1000546875,
"step": 6600
},
{
"epoch": 2.1595487510072524,
"grad_norm": 1.6646093520541214,
"learning_rate": 6.6873381793716865e-06,
"loss": 1.6777,
"step": 6700
},
{
"epoch": 2.1595487510072524,
"eval_loss": 1.7090308666229248,
"eval_runtime": 42.7292,
"eval_samples_per_second": 23.403,
"eval_steps_per_second": 5.851,
"eval_text_contr_loss": 0.10214550018310548,
"eval_total_custom_loss": 3.2032421875,
"eval_total_image_loss": 3.931541748046875,
"eval_total_text_loss": 0.4815105590820313,
"eval_vision_contr_loss": 3.10109619140625,
"step": 6700
},
{
"epoch": 2.191780821917808,
"grad_norm": 1.118793791053252,
"learning_rate": 6.596540683212729e-06,
"loss": 1.662,
"step": 6800
},
{
"epoch": 2.191780821917808,
"eval_loss": 1.707641363143921,
"eval_runtime": 43.0524,
"eval_samples_per_second": 23.228,
"eval_steps_per_second": 5.807,
"eval_text_contr_loss": 0.10229474639892579,
"eval_total_custom_loss": 3.200827392578125,
"eval_total_image_loss": 3.9283203125,
"eval_total_text_loss": 0.48220172119140625,
"eval_vision_contr_loss": 3.09853271484375,
"step": 6800
},
{
"epoch": 2.2240128928283642,
"grad_norm": 1.5524306551850715,
"learning_rate": 6.505266032797806e-06,
"loss": 1.6846,
"step": 6900
},
{
"epoch": 2.2240128928283642,
"eval_loss": 1.7059853076934814,
"eval_runtime": 43.1533,
"eval_samples_per_second": 23.173,
"eval_steps_per_second": 5.793,
"eval_text_contr_loss": 0.10189845275878906,
"eval_total_custom_loss": 3.1974970703125,
"eval_total_image_loss": 3.924572265625,
"eval_total_text_loss": 0.4803218994140625,
"eval_vision_contr_loss": 3.09559814453125,
"step": 6900
},
{
"epoch": 2.2562449637389204,
"grad_norm": 1.0266753499509045,
"learning_rate": 6.413553945839697e-06,
"loss": 1.6714,
"step": 7000
},
{
"epoch": 2.2562449637389204,
"eval_loss": 1.705426812171936,
"eval_runtime": 43.1904,
"eval_samples_per_second": 23.153,
"eval_steps_per_second": 5.788,
"eval_text_contr_loss": 0.10180580139160156,
"eval_total_custom_loss": 3.196394287109375,
"eval_total_image_loss": 3.923303955078125,
"eval_total_text_loss": 0.47996240234375,
"eval_vision_contr_loss": 3.09458837890625,
"step": 7000
},
{
"epoch": 2.288477034649476,
"grad_norm": 1.2088814629290832,
"learning_rate": 6.321444330399532e-06,
"loss": 1.66,
"step": 7100
},
{
"epoch": 2.288477034649476,
"eval_loss": 1.7046881914138794,
"eval_runtime": 43.2519,
"eval_samples_per_second": 23.12,
"eval_steps_per_second": 5.78,
"eval_text_contr_loss": 0.10193362426757813,
"eval_total_custom_loss": 3.1951435546875,
"eval_total_image_loss": 3.92156982421875,
"eval_total_text_loss": 0.4805397338867187,
"eval_vision_contr_loss": 3.0932099609375,
"step": 7100
},
{
"epoch": 2.320709105560032,
"grad_norm": 1.3694715406974758,
"learning_rate": 6.228977267520991e-06,
"loss": 1.665,
"step": 7200
},
{
"epoch": 2.320709105560032,
"eval_loss": 1.7047983407974243,
"eval_runtime": 42.9671,
"eval_samples_per_second": 23.274,
"eval_steps_per_second": 5.818,
"eval_text_contr_loss": 0.1015860595703125,
"eval_total_custom_loss": 3.1950625,
"eval_total_image_loss": 3.92192333984375,
"eval_total_text_loss": 0.47889395141601565,
"eval_vision_contr_loss": 3.093476806640625,
"step": 7200
},
{
"epoch": 2.3529411764705883,
"grad_norm": 1.1989074798894601,
"learning_rate": 6.136192993789241e-06,
"loss": 1.6754,
"step": 7300
},
{
"epoch": 2.3529411764705883,
"eval_loss": 1.7038034200668335,
"eval_runtime": 43.194,
"eval_samples_per_second": 23.151,
"eval_steps_per_second": 5.788,
"eval_text_contr_loss": 0.1014683837890625,
"eval_total_custom_loss": 3.193150390625,
"eval_total_image_loss": 3.9196484375,
"eval_total_text_loss": 0.4783086547851563,
"eval_vision_contr_loss": 3.09168212890625,
"step": 7300
},
{
"epoch": 2.385173247381144,
"grad_norm": 1.1841765547179885,
"learning_rate": 6.043131883822186e-06,
"loss": 1.6716,
"step": 7400
},
{
"epoch": 2.385173247381144,
"eval_loss": 1.7007592916488647,
"eval_runtime": 43.0911,
"eval_samples_per_second": 23.207,
"eval_steps_per_second": 5.802,
"eval_text_contr_loss": 0.10157785034179688,
"eval_total_custom_loss": 3.1876787109375,
"eval_total_image_loss": 3.912532958984375,
"eval_total_text_loss": 0.4788715515136719,
"eval_vision_contr_loss": 3.0861005859375,
"step": 7400
},
{
"epoch": 2.4174053182917,
"grad_norm": 1.09623045255461,
"learning_rate": 5.949834432701642e-06,
"loss": 1.6904,
"step": 7500
},
{
"epoch": 2.4174053182917,
"eval_loss": 1.7012134790420532,
"eval_runtime": 43.2095,
"eval_samples_per_second": 23.143,
"eval_steps_per_second": 5.786,
"eval_text_contr_loss": 0.10139631652832032,
"eval_total_custom_loss": 3.188372314453125,
"eval_total_image_loss": 3.91363525390625,
"eval_total_text_loss": 0.47799374389648436,
"eval_vision_contr_loss": 3.086975830078125,
"step": 7500
},
{
"epoch": 2.4496373892022563,
"grad_norm": 1.2798135292777189,
"learning_rate": 5.856341238352114e-06,
"loss": 1.6686,
"step": 7600
},
{
"epoch": 2.4496373892022563,
"eval_loss": 1.699338674545288,
"eval_runtime": 43.1463,
"eval_samples_per_second": 23.177,
"eval_steps_per_second": 5.794,
"eval_text_contr_loss": 0.10125347900390624,
"eval_total_custom_loss": 3.18483154296875,
"eval_total_image_loss": 3.909332275390625,
"eval_total_text_loss": 0.47738201904296873,
"eval_vision_contr_loss": 3.083578125,
"step": 7600
},
{
"epoch": 2.4821917808219176,
"grad_norm": 1.2387160690879455,
"learning_rate": 5.762692983874807e-06,
"loss": 1.6875,
"step": 7700
},
{
"epoch": 2.4821917808219176,
"eval_loss": 1.699604868888855,
"eval_runtime": 43.2954,
"eval_samples_per_second": 23.097,
"eval_steps_per_second": 5.774,
"eval_text_contr_loss": 0.10113139343261719,
"eval_total_custom_loss": 3.18522021484375,
"eval_total_image_loss": 3.90997802734375,
"eval_total_text_loss": 0.47675457763671875,
"eval_vision_contr_loss": 3.084088134765625,
"step": 7700
},
{
"epoch": 2.5144238517324737,
"grad_norm": 1.2888845613016715,
"learning_rate": 5.668930419844569e-06,
"loss": 1.6694,
"step": 7800
},
{
"epoch": 2.5144238517324737,
"eval_loss": 1.6977527141571045,
"eval_runtime": 42.8068,
"eval_samples_per_second": 23.361,
"eval_steps_per_second": 5.84,
"eval_text_contr_loss": 0.10074325561523438,
"eval_total_custom_loss": 3.1815302734375,
"eval_total_image_loss": 3.9057802734375,
"eval_total_text_loss": 0.47487213134765627,
"eval_vision_contr_loss": 3.080787353515625,
"step": 7800
},
{
"epoch": 2.54665592264303,
"grad_norm": 1.0961945452459818,
"learning_rate": 5.575094346577487e-06,
"loss": 1.6843,
"step": 7900
},
{
"epoch": 2.54665592264303,
"eval_loss": 1.6972546577453613,
"eval_runtime": 43.1625,
"eval_samples_per_second": 23.168,
"eval_steps_per_second": 5.792,
"eval_text_contr_loss": 0.10105512237548828,
"eval_total_custom_loss": 3.180876708984375,
"eval_total_image_loss": 3.904545166015625,
"eval_total_text_loss": 0.47634335327148436,
"eval_vision_contr_loss": 3.07982177734375,
"step": 7900
},
{
"epoch": 2.578887993553586,
"grad_norm": 1.0730302928206912,
"learning_rate": 5.481225596376832e-06,
"loss": 1.6531,
"step": 8000
},
{
"epoch": 2.578887993553586,
"eval_loss": 1.6951245069503784,
"eval_runtime": 42.7536,
"eval_samples_per_second": 23.39,
"eval_steps_per_second": 5.847,
"eval_text_contr_loss": 0.10080006408691407,
"eval_total_custom_loss": 3.1767880859375,
"eval_total_image_loss": 3.899679443359375,
"eval_total_text_loss": 0.4752117919921875,
"eval_vision_contr_loss": 3.075987548828125,
"step": 8000
},
{
"epoch": 2.6111200644641417,
"grad_norm": 1.2012985411125285,
"learning_rate": 5.387365015765072e-06,
"loss": 1.6645,
"step": 8100
},
{
"epoch": 2.6111200644641417,
"eval_loss": 1.6954604387283325,
"eval_runtime": 42.8389,
"eval_samples_per_second": 23.343,
"eval_steps_per_second": 5.836,
"eval_text_contr_loss": 0.10047319793701172,
"eval_total_custom_loss": 3.17713623046875,
"eval_total_image_loss": 3.90052880859375,
"eval_total_text_loss": 0.47367898559570315,
"eval_vision_contr_loss": 3.0766630859375,
"step": 8100
},
{
"epoch": 2.643352135374698,
"grad_norm": 1.0726584383882933,
"learning_rate": 5.2935534477097296e-06,
"loss": 1.6677,
"step": 8200
},
{
"epoch": 2.643352135374698,
"eval_loss": 1.6943846940994263,
"eval_runtime": 42.5793,
"eval_samples_per_second": 23.486,
"eval_steps_per_second": 5.871,
"eval_text_contr_loss": 0.10054948425292969,
"eval_total_custom_loss": 3.175241943359375,
"eval_total_image_loss": 3.898025146484375,
"eval_total_text_loss": 0.4740495300292969,
"eval_vision_contr_loss": 3.07469189453125,
"step": 8200
},
{
"epoch": 2.675584206285254,
"grad_norm": 1.0454265169766854,
"learning_rate": 5.199831713850748e-06,
"loss": 1.6843,
"step": 8300
},
{
"epoch": 2.675584206285254,
"eval_loss": 1.693207025527954,
"eval_runtime": 42.8178,
"eval_samples_per_second": 23.355,
"eval_steps_per_second": 5.839,
"eval_text_contr_loss": 0.10044010162353516,
"eval_total_custom_loss": 3.17299267578125,
"eval_total_image_loss": 3.895315673828125,
"eval_total_text_loss": 0.4735074768066406,
"eval_vision_contr_loss": 3.072552490234375,
"step": 8300
},
{
"epoch": 2.7078162771958096,
"grad_norm": 1.0376317976485014,
"learning_rate": 5.106240596737168e-06,
"loss": 1.6772,
"step": 8400
},
{
"epoch": 2.7078162771958096,
"eval_loss": 1.6923362016677856,
"eval_runtime": 42.9651,
"eval_samples_per_second": 23.275,
"eval_steps_per_second": 5.819,
"eval_text_contr_loss": 0.1002273178100586,
"eval_total_custom_loss": 3.17123046875,
"eval_total_image_loss": 3.893354736328125,
"eval_total_text_loss": 0.4724638671875,
"eval_vision_contr_loss": 3.07100341796875,
"step": 8400
},
{
"epoch": 2.7400483481063658,
"grad_norm": 1.1505481111073788,
"learning_rate": 5.012820822080799e-06,
"loss": 1.6677,
"step": 8500
},
{
"epoch": 2.7400483481063658,
"eval_loss": 1.6918821334838867,
"eval_runtime": 43.5347,
"eval_samples_per_second": 22.97,
"eval_steps_per_second": 5.743,
"eval_text_contr_loss": 0.10038319396972656,
"eval_total_custom_loss": 3.170523193359375,
"eval_total_image_loss": 3.892265625,
"eval_total_text_loss": 0.473227294921875,
"eval_vision_contr_loss": 3.070140380859375,
"step": 8500
},
{
"epoch": 2.772280419016922,
"grad_norm": 1.2546342286772034,
"learning_rate": 4.919613041034628e-06,
"loss": 1.6525,
"step": 8600
},
{
"epoch": 2.772280419016922,
"eval_loss": 1.6920381784439087,
"eval_runtime": 43.2028,
"eval_samples_per_second": 23.147,
"eval_steps_per_second": 5.787,
"eval_text_contr_loss": 0.10001284027099609,
"eval_total_custom_loss": 3.170507568359375,
"eval_total_image_loss": 3.892719482421875,
"eval_total_text_loss": 0.47142926025390625,
"eval_vision_contr_loss": 3.07049462890625,
"step": 8600
},
{
"epoch": 2.804512489927478,
"grad_norm": 1.0449956888200327,
"learning_rate": 4.8266578125036675e-06,
"loss": 1.655,
"step": 8700
},
{
"epoch": 2.804512489927478,
"eval_loss": 1.6917293071746826,
"eval_runtime": 42.7617,
"eval_samples_per_second": 23.385,
"eval_steps_per_second": 5.846,
"eval_text_contr_loss": 0.10005457305908202,
"eval_total_custom_loss": 3.16998046875,
"eval_total_image_loss": 3.89201171875,
"eval_total_text_loss": 0.47171200561523435,
"eval_vision_contr_loss": 3.06992578125,
"step": 8700
},
{
"epoch": 2.8367445608380337,
"grad_norm": 1.1304255420893947,
"learning_rate": 4.733995585495955e-06,
"loss": 1.6619,
"step": 8800
},
{
"epoch": 2.8367445608380337,
"eval_loss": 1.6887074708938599,
"eval_runtime": 42.8364,
"eval_samples_per_second": 23.345,
"eval_steps_per_second": 5.836,
"eval_text_contr_loss": 0.09995537567138672,
"eval_total_custom_loss": 3.1643916015625,
"eval_total_image_loss": 3.8850234375,
"eval_total_text_loss": 0.471197998046875,
"eval_vision_contr_loss": 3.06443603515625,
"step": 8800
},
{
"epoch": 2.86897663174859,
"grad_norm": 1.1635596710167195,
"learning_rate": 4.6416666815213646e-06,
"loss": 1.6539,
"step": 8900
},
{
"epoch": 2.86897663174859,
"eval_loss": 1.6882895231246948,
"eval_runtime": 42.7082,
"eval_samples_per_second": 23.415,
"eval_steps_per_second": 5.854,
"eval_text_contr_loss": 0.099897216796875,
"eval_total_custom_loss": 3.16357763671875,
"eval_total_image_loss": 3.88407275390625,
"eval_total_text_loss": 0.47090353393554685,
"eval_vision_contr_loss": 3.0636806640625,
"step": 8900
},
{
"epoch": 2.901208702659146,
"grad_norm": 1.4475543234008788,
"learning_rate": 4.549711277045895e-06,
"loss": 1.6513,
"step": 9000
},
{
"epoch": 2.901208702659146,
"eval_loss": 1.6892811059951782,
"eval_runtime": 42.7435,
"eval_samples_per_second": 23.395,
"eval_steps_per_second": 5.849,
"eval_text_contr_loss": 0.09978788757324218,
"eval_total_custom_loss": 3.16530078125,
"eval_total_image_loss": 3.88640869140625,
"eval_total_text_loss": 0.4704385681152344,
"eval_vision_contr_loss": 3.06551318359375,
"step": 9000
},
{
"epoch": 2.9334407735697017,
"grad_norm": 1.082082592198345,
"learning_rate": 4.458169386009095e-06,
"loss": 1.6623,
"step": 9100
},
{
"epoch": 2.9334407735697017,
"eval_loss": 1.6865910291671753,
"eval_runtime": 42.7896,
"eval_samples_per_second": 23.37,
"eval_steps_per_second": 5.843,
"eval_text_contr_loss": 0.0996867904663086,
"eval_total_custom_loss": 3.16030908203125,
"eval_total_image_loss": 3.880197021484375,
"eval_total_text_loss": 0.46994622802734376,
"eval_vision_contr_loss": 3.06062255859375,
"step": 9100
},
{
"epoch": 2.965672844480258,
"grad_norm": 1.252913071666507,
"learning_rate": 4.367080842412176e-06,
"loss": 1.667,
"step": 9200
},
{
"epoch": 2.965672844480258,
"eval_loss": 1.685819149017334,
"eval_runtime": 42.9823,
"eval_samples_per_second": 23.265,
"eval_steps_per_second": 5.816,
"eval_text_contr_loss": 0.09963002014160156,
"eval_total_custom_loss": 3.1588515625,
"eval_total_image_loss": 3.878440185546875,
"eval_total_text_loss": 0.46969296264648436,
"eval_vision_contr_loss": 3.059221435546875,
"step": 9200
},
{
"epoch": 2.997904915390814,
"grad_norm": 1.2123646326849216,
"learning_rate": 4.2764852829844575e-06,
"loss": 1.6493,
"step": 9300
},
{
"epoch": 2.997904915390814,
"eval_loss": 1.684321403503418,
"eval_runtime": 43.2065,
"eval_samples_per_second": 23.145,
"eval_steps_per_second": 5.786,
"eval_text_contr_loss": 0.09949755859375,
"eval_total_custom_loss": 3.156007568359375,
"eval_total_image_loss": 3.875003662109375,
"eval_total_text_loss": 0.469060302734375,
"eval_vision_contr_loss": 3.056509033203125,
"step": 9300
},
{
"epoch": 3.03013698630137,
"grad_norm": 1.2765862298497435,
"learning_rate": 4.186422129935634e-06,
"loss": 1.6081,
"step": 9400
},
{
"epoch": 3.03013698630137,
"eval_loss": 1.6938881874084473,
"eval_runtime": 43.1138,
"eval_samples_per_second": 23.194,
"eval_steps_per_second": 5.799,
"eval_text_contr_loss": 0.10148472595214844,
"eval_total_custom_loss": 3.175074462890625,
"eval_total_image_loss": 3.896648681640625,
"eval_total_text_loss": 0.4784152221679687,
"eval_vision_contr_loss": 3.07359033203125,
"step": 9400
},
{
"epoch": 3.0623690572119258,
"grad_norm": 1.2098762882131135,
"learning_rate": 4.096930573801397e-06,
"loss": 1.634,
"step": 9500
},
{
"epoch": 3.0623690572119258,
"eval_loss": 1.6933133602142334,
"eval_runtime": 42.9251,
"eval_samples_per_second": 23.296,
"eval_steps_per_second": 5.824,
"eval_text_contr_loss": 0.101427734375,
"eval_total_custom_loss": 3.1739755859375,
"eval_total_image_loss": 3.89532421875,
"eval_total_text_loss": 0.4781582641601563,
"eval_vision_contr_loss": 3.07254736328125,
"step": 9500
},
{
"epoch": 3.094601128122482,
"grad_norm": 1.3163746415227902,
"learning_rate": 4.008049556389867e-06,
"loss": 1.5951,
"step": 9600
},
{
"epoch": 3.094601128122482,
"eval_loss": 1.6907294988632202,
"eval_runtime": 42.9029,
"eval_samples_per_second": 23.308,
"eval_steps_per_second": 5.827,
"eval_text_contr_loss": 0.10167989349365235,
"eval_total_custom_loss": 3.16946728515625,
"eval_total_image_loss": 3.88928515625,
"eval_total_text_loss": 0.47932699584960936,
"eval_vision_contr_loss": 3.067786376953125,
"step": 9600
},
{
"epoch": 3.126833199033038,
"grad_norm": 1.2525529317293345,
"learning_rate": 3.919817753836259e-06,
"loss": 1.6167,
"step": 9700
},
{
"epoch": 3.126833199033038,
"eval_loss": 1.692821741104126,
"eval_runtime": 43.003,
"eval_samples_per_second": 23.254,
"eval_steps_per_second": 5.814,
"eval_text_contr_loss": 0.10175180816650391,
"eval_total_custom_loss": 3.173349365234375,
"eval_total_image_loss": 3.894118896484375,
"eval_total_text_loss": 0.4796273193359375,
"eval_vision_contr_loss": 3.0715966796875,
"step": 9700
},
{
"epoch": 3.1590652699435937,
"grad_norm": 1.126369009113395,
"learning_rate": 3.832273559773153e-06,
"loss": 1.6043,
"step": 9800
},
{
"epoch": 3.1590652699435937,
"eval_loss": 1.6909953355789185,
"eval_runtime": 42.8655,
"eval_samples_per_second": 23.329,
"eval_steps_per_second": 5.832,
"eval_text_contr_loss": 0.10167758178710938,
"eval_total_custom_loss": 3.169955078125,
"eval_total_image_loss": 3.889908203125,
"eval_total_text_loss": 0.4793105163574219,
"eval_vision_contr_loss": 3.06827734375,
"step": 9800
},
{
"epoch": 3.19129734085415,
"grad_norm": 1.123771397464815,
"learning_rate": 3.7454550686236945e-06,
"loss": 1.6073,
"step": 9900
},
{
"epoch": 3.19129734085415,
"eval_loss": 1.6896721124649048,
"eval_runtime": 42.8425,
"eval_samples_per_second": 23.341,
"eval_steps_per_second": 5.835,
"eval_text_contr_loss": 0.1015400161743164,
"eval_total_custom_loss": 3.16741552734375,
"eval_total_image_loss": 3.88685693359375,
"eval_total_text_loss": 0.47866622924804686,
"eval_vision_contr_loss": 3.065875732421875,
"step": 9900
},
{
"epoch": 3.223529411764706,
"grad_norm": 1.0983682310218152,
"learning_rate": 3.6594000590249946e-06,
"loss": 1.62,
"step": 10000
},
{
"epoch": 3.223529411764706,
"eval_loss": 1.6913846731185913,
"eval_runtime": 42.7475,
"eval_samples_per_second": 23.393,
"eval_steps_per_second": 5.848,
"eval_text_contr_loss": 0.10158514404296876,
"eval_total_custom_loss": 3.17058740234375,
"eval_total_image_loss": 3.890841552734375,
"eval_total_text_loss": 0.4788506469726562,
"eval_vision_contr_loss": 3.06900244140625,
"step": 10000
},
{
"epoch": 3.255761482675262,
"grad_norm": 1.1391042191009226,
"learning_rate": 3.5741459773889415e-06,
"loss": 1.6032,
"step": 10100
},
{
"epoch": 3.255761482675262,
"eval_loss": 1.6893099546432495,
"eval_runtime": 42.7757,
"eval_samples_per_second": 23.378,
"eval_steps_per_second": 5.844,
"eval_text_contr_loss": 0.10135261535644531,
"eval_total_custom_loss": 3.1666064453125,
"eval_total_image_loss": 3.886076171875,
"eval_total_text_loss": 0.47780508422851564,
"eval_vision_contr_loss": 3.06525390625,
"step": 10100
},
{
"epoch": 3.287993553585818,
"grad_norm": 1.2951870930310643,
"learning_rate": 3.4897299216075838e-06,
"loss": 1.6156,
"step": 10200
},
{
"epoch": 3.287993553585818,
"eval_loss": 1.6886229515075684,
"eval_runtime": 42.9874,
"eval_samples_per_second": 23.263,
"eval_steps_per_second": 5.816,
"eval_text_contr_loss": 0.10141517639160157,
"eval_total_custom_loss": 3.16539208984375,
"eval_total_image_loss": 3.884467041015625,
"eval_total_text_loss": 0.4781048889160156,
"eval_vision_contr_loss": 3.063976806640625,
"step": 10200
},
{
"epoch": 3.320225624496374,
"grad_norm": 1.1576760253285912,
"learning_rate": 3.40618862491016e-06,
"loss": 1.607,
"step": 10300
},
{
"epoch": 3.320225624496374,
"eval_loss": 1.6859066486358643,
"eval_runtime": 42.9508,
"eval_samples_per_second": 23.282,
"eval_steps_per_second": 5.821,
"eval_text_contr_loss": 0.10142912292480469,
"eval_total_custom_loss": 3.160448486328125,
"eval_total_image_loss": 3.878161376953125,
"eval_total_text_loss": 0.47815496826171877,
"eval_vision_contr_loss": 3.05901953125,
"step": 10300
},
{
"epoch": 3.35245769540693,
"grad_norm": 1.1233070454350762,
"learning_rate": 3.3235584398788225e-06,
"loss": 1.6302,
"step": 10400
},
{
"epoch": 3.35245769540693,
"eval_loss": 1.686716079711914,
"eval_runtime": 42.7155,
"eval_samples_per_second": 23.411,
"eval_steps_per_second": 5.853,
"eval_text_contr_loss": 0.10136524200439453,
"eval_total_custom_loss": 3.161879638671875,
"eval_total_image_loss": 3.880064208984375,
"eval_total_text_loss": 0.4778822631835937,
"eval_vision_contr_loss": 3.0605146484375,
"step": 10400
},
{
"epoch": 3.3846897663174857,
"grad_norm": 1.1327092213015983,
"learning_rate": 3.241875322629986e-06,
"loss": 1.6036,
"step": 10500
},
{
"epoch": 3.3846897663174857,
"eval_loss": 1.6858470439910889,
"eval_runtime": 42.7744,
"eval_samples_per_second": 23.378,
"eval_steps_per_second": 5.845,
"eval_text_contr_loss": 0.10138286590576172,
"eval_total_custom_loss": 3.160292236328125,
"eval_total_image_loss": 3.8780537109375,
"eval_total_text_loss": 0.47793585205078126,
"eval_vision_contr_loss": 3.0589091796875,
"step": 10500
},
{
"epoch": 3.416921837228042,
"grad_norm": 1.1339295712914674,
"learning_rate": 3.1611748171682016e-06,
"loss": 1.6453,
"step": 10600
},
{
"epoch": 3.416921837228042,
"eval_loss": 1.687591314315796,
"eval_runtime": 42.8807,
"eval_samples_per_second": 23.321,
"eval_steps_per_second": 5.83,
"eval_text_contr_loss": 0.10116822052001953,
"eval_total_custom_loss": 3.16331298828125,
"eval_total_image_loss": 3.882153076171875,
"eval_total_text_loss": 0.4768859558105469,
"eval_vision_contr_loss": 3.06214453125,
"step": 10600
},
{
"epoch": 3.449153908138598,
"grad_norm": 1.0997773503486938,
"learning_rate": 3.081492039919361e-06,
"loss": 1.6054,
"step": 10700
},
{
"epoch": 3.449153908138598,
"eval_loss": 1.6868486404418945,
"eval_runtime": 42.8674,
"eval_samples_per_second": 23.328,
"eval_steps_per_second": 5.832,
"eval_text_contr_loss": 0.10126123046875,
"eval_total_custom_loss": 3.162032470703125,
"eval_total_image_loss": 3.880401123046875,
"eval_total_text_loss": 0.47728997802734374,
"eval_vision_contr_loss": 3.060771484375,
"step": 10700
},
{
"epoch": 3.481385979049154,
"grad_norm": 1.219354719842817,
"learning_rate": 3.0028616644499574e-06,
"loss": 1.5969,
"step": 10800
},
{
"epoch": 3.481385979049154,
"eval_loss": 1.6862396001815796,
"eval_runtime": 42.9267,
"eval_samples_per_second": 23.296,
"eval_steps_per_second": 5.824,
"eval_text_contr_loss": 0.10140933227539063,
"eval_total_custom_loss": 3.16104345703125,
"eval_total_image_loss": 3.878958984375,
"eval_total_text_loss": 0.4780615234375,
"eval_vision_contr_loss": 3.0596337890625,
"step": 10800
},
{
"epoch": 3.51361804995971,
"grad_norm": 1.175772050996846,
"learning_rate": 2.9253179063790525e-06,
"loss": 1.5918,
"step": 10900
},
{
"epoch": 3.51361804995971,
"eval_loss": 1.6859683990478516,
"eval_runtime": 42.9594,
"eval_samples_per_second": 23.278,
"eval_steps_per_second": 5.819,
"eval_text_contr_loss": 0.1010609359741211,
"eval_total_custom_loss": 3.1602666015625,
"eval_total_image_loss": 3.8784208984375,
"eval_total_text_loss": 0.47643841552734373,
"eval_vision_contr_loss": 3.059205810546875,
"step": 10900
},
{
"epoch": 3.545850120870266,
"grad_norm": 1.1237469272528742,
"learning_rate": 2.8488945084895257e-06,
"loss": 1.6129,
"step": 11000
},
{
"epoch": 3.545850120870266,
"eval_loss": 1.6847009658813477,
"eval_runtime": 42.8786,
"eval_samples_per_second": 23.322,
"eval_steps_per_second": 5.83,
"eval_text_contr_loss": 0.10107876586914062,
"eval_total_custom_loss": 3.15795556640625,
"eval_total_image_loss": 3.87548388671875,
"eval_total_text_loss": 0.4764222412109375,
"eval_vision_contr_loss": 3.056876708984375,
"step": 11000
},
{
"epoch": 3.5780821917808217,
"grad_norm": 1.0867638145927627,
"learning_rate": 2.7736247260450542e-06,
"loss": 1.6222,
"step": 11100
},
{
"epoch": 3.5780821917808217,
"eval_loss": 1.6834853887557983,
"eval_runtime": 42.9899,
"eval_samples_per_second": 23.261,
"eval_steps_per_second": 5.815,
"eval_text_contr_loss": 0.10083826446533203,
"eval_total_custom_loss": 3.155547607421875,
"eval_total_image_loss": 3.87272802734375,
"eval_total_text_loss": 0.47536029052734374,
"eval_vision_contr_loss": 3.054708984375,
"step": 11100
},
{
"epoch": 3.610314262691378,
"grad_norm": 1.2223953493974884,
"learning_rate": 2.699541312319264e-06,
"loss": 1.6045,
"step": 11200
},
{
"epoch": 3.610314262691378,
"eval_loss": 1.684789776802063,
"eval_runtime": 42.8174,
"eval_samples_per_second": 23.355,
"eval_steps_per_second": 5.839,
"eval_text_contr_loss": 0.10071804809570313,
"eval_total_custom_loss": 3.1578349609375,
"eval_total_image_loss": 3.875773681640625,
"eval_total_text_loss": 0.4747677001953125,
"eval_vision_contr_loss": 3.0571171875,
"step": 11200
},
{
"epoch": 3.642546333601934,
"grad_norm": 1.076543018855401,
"learning_rate": 2.6266765043433017e-06,
"loss": 1.6068,
"step": 11300
},
{
"epoch": 3.642546333601934,
"eval_loss": 1.6840680837631226,
"eval_runtime": 42.9346,
"eval_samples_per_second": 23.291,
"eval_steps_per_second": 5.823,
"eval_text_contr_loss": 0.10091355895996093,
"eval_total_custom_loss": 3.15666748046875,
"eval_total_image_loss": 3.87404931640625,
"eval_total_text_loss": 0.4756776428222656,
"eval_vision_contr_loss": 3.05575390625,
"step": 11300
},
{
"epoch": 3.67477840451249,
"grad_norm": 1.201376688056067,
"learning_rate": 2.555062008878044e-06,
"loss": 1.6157,
"step": 11400
},
{
"epoch": 3.67477840451249,
"eval_loss": 1.6833868026733398,
"eval_runtime": 42.9832,
"eval_samples_per_second": 23.265,
"eval_steps_per_second": 5.816,
"eval_text_contr_loss": 0.100856201171875,
"eval_total_custom_loss": 3.15538037109375,
"eval_total_image_loss": 3.872487548828125,
"eval_total_text_loss": 0.47540740966796874,
"eval_vision_contr_loss": 3.05452392578125,
"step": 11400
},
{
"epoch": 3.707010475423046,
"grad_norm": 1.032794436813145,
"learning_rate": 2.484728988617063e-06,
"loss": 1.6082,
"step": 11500
},
{
"epoch": 3.707010475423046,
"eval_loss": 1.6835486888885498,
"eval_runtime": 43.5001,
"eval_samples_per_second": 22.988,
"eval_steps_per_second": 5.747,
"eval_text_contr_loss": 0.10072601318359375,
"eval_total_custom_loss": 3.15557373046875,
"eval_total_image_loss": 3.87289892578125,
"eval_total_text_loss": 0.4747868347167969,
"eval_vision_contr_loss": 3.054848388671875,
"step": 11500
},
{
"epoch": 3.739242546333602,
"grad_norm": 1.003411270750372,
"learning_rate": 2.415708048626333e-06,
"loss": 1.5981,
"step": 11600
},
{
"epoch": 3.739242546333602,
"eval_loss": 1.6829206943511963,
"eval_runtime": 42.9967,
"eval_samples_per_second": 23.258,
"eval_steps_per_second": 5.814,
"eval_text_contr_loss": 0.10052241516113282,
"eval_total_custom_loss": 3.154263916015625,
"eval_total_image_loss": 3.8714951171875,
"eval_total_text_loss": 0.4738740234375,
"eval_vision_contr_loss": 3.0537412109375,
"step": 11600
},
{
"epoch": 3.771474617244158,
"grad_norm": 1.063479476495336,
"learning_rate": 2.348029223026585e-06,
"loss": 1.597,
"step": 11700
},
{
"epoch": 3.771474617244158,
"eval_loss": 1.68172025680542,
"eval_runtime": 42.9293,
"eval_samples_per_second": 23.294,
"eval_steps_per_second": 5.824,
"eval_text_contr_loss": 0.10068083190917969,
"eval_total_custom_loss": 3.152196533203125,
"eval_total_image_loss": 3.86867529296875,
"eval_total_text_loss": 0.47459600830078125,
"eval_vision_contr_loss": 3.0515166015625,
"step": 11700
},
{
"epoch": 3.8037066881547137,
"grad_norm": 1.0838405339905979,
"learning_rate": 2.2817219619241105e-06,
"loss": 1.6236,
"step": 11800
},
{
"epoch": 3.8037066881547137,
"eval_loss": 1.6811496019363403,
"eval_runtime": 42.8946,
"eval_samples_per_second": 23.313,
"eval_steps_per_second": 5.828,
"eval_text_contr_loss": 0.10067573547363282,
"eval_total_custom_loss": 3.151148193359375,
"eval_total_image_loss": 3.867359375,
"eval_total_text_loss": 0.4745587158203125,
"eval_vision_contr_loss": 3.05047216796875,
"step": 11800
},
{
"epoch": 3.83593875906527,
"grad_norm": 1.079820928087788,
"learning_rate": 2.216815118595693e-06,
"loss": 1.6317,
"step": 11900
},
{
"epoch": 3.83593875906527,
"eval_loss": 1.6812784671783447,
"eval_runtime": 42.7509,
"eval_samples_per_second": 23.391,
"eval_steps_per_second": 5.848,
"eval_text_contr_loss": 0.10055113220214844,
"eval_total_custom_loss": 3.15128173828125,
"eval_total_image_loss": 3.86768896484375,
"eval_total_text_loss": 0.47393701171875,
"eval_vision_contr_loss": 3.050729736328125,
"step": 11900
},
{
"epoch": 3.868170829975826,
"grad_norm": 1.1619861502916402,
"learning_rate": 2.1533369369332456e-06,
"loss": 1.6183,
"step": 12000
},
{
"epoch": 3.868170829975826,
"eval_loss": 1.6810537576675415,
"eval_runtime": 42.7941,
"eval_samples_per_second": 23.368,
"eval_steps_per_second": 5.842,
"eval_text_contr_loss": 0.1004962387084961,
"eval_total_custom_loss": 3.150823974609375,
"eval_total_image_loss": 3.86718115234375,
"eval_total_text_loss": 0.4737078857421875,
"eval_vision_contr_loss": 3.050328125,
"step": 12000
},
{
"epoch": 3.900402900886382,
"grad_norm": 1.1726466990038384,
"learning_rate": 2.0913150391536236e-06,
"loss": 1.6114,
"step": 12100
},
{
"epoch": 3.900402900886382,
"eval_loss": 1.6809941530227661,
"eval_runtime": 43.1003,
"eval_samples_per_second": 23.202,
"eval_steps_per_second": 5.8,
"eval_text_contr_loss": 0.10044279479980468,
"eval_total_custom_loss": 3.150681640625,
"eval_total_image_loss": 3.86706689453125,
"eval_total_text_loss": 0.47346914672851564,
"eval_vision_contr_loss": 3.05023876953125,
"step": 12100
},
{
"epoch": 3.932634971796938,
"grad_norm": 1.1410317698581134,
"learning_rate": 2.0307764137789505e-06,
"loss": 1.6167,
"step": 12200
},
{
"epoch": 3.932634971796938,
"eval_loss": 1.6800341606140137,
"eval_runtime": 42.8924,
"eval_samples_per_second": 23.314,
"eval_steps_per_second": 5.829,
"eval_text_contr_loss": 0.10050942993164062,
"eval_total_custom_loss": 3.14897265625,
"eval_total_image_loss": 3.864827880859375,
"eval_total_text_loss": 0.47377548217773435,
"eval_vision_contr_loss": 3.048463134765625,
"step": 12200
},
{
"epoch": 3.964867042707494,
"grad_norm": 1.074995807709666,
"learning_rate": 1.9717474038926967e-06,
"loss": 1.6029,
"step": 12300
},
{
"epoch": 3.964867042707494,
"eval_loss": 1.678675889968872,
"eval_runtime": 42.9912,
"eval_samples_per_second": 23.261,
"eval_steps_per_second": 5.815,
"eval_text_contr_loss": 0.10038053894042968,
"eval_total_custom_loss": 3.14638232421875,
"eval_total_image_loss": 3.86171728515625,
"eval_total_text_loss": 0.4731833190917969,
"eval_vision_contr_loss": 3.046001953125,
"step": 12300
},
{
"epoch": 3.99709911361805,
"grad_norm": 1.1178357415374787,
"learning_rate": 1.91425369567662e-06,
"loss": 1.6086,
"step": 12400
},
{
"epoch": 3.99709911361805,
"eval_loss": 1.6787165403366089,
"eval_runtime": 42.7428,
"eval_samples_per_second": 23.396,
"eval_steps_per_second": 5.849,
"eval_text_contr_loss": 0.10042031860351562,
"eval_total_custom_loss": 3.1464853515625,
"eval_total_image_loss": 3.86180322265625,
"eval_total_text_loss": 0.4733684997558594,
"eval_vision_contr_loss": 3.046065185546875,
"step": 12400
}
],
"logging_steps": 100,
"max_steps": 15510,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 400,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 1479806652776448.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}