Training in progress, step 4500, checkpoint

checkpoint-4500/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b526598c876a1217a0fdbc902768d29e56ab8c30f4d8c0a24f593e6424d35229
+oid sha256:c913361aa159377d3d737d35196b0e48eda3669d2a4d41a478d15e5527ac4afa
 size 1711848436

checkpoint-4500/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0a998e37f1fa5203a5785f84580c856ef10c9a61f839ea21b48dc86e177b280c
+oid sha256:c4dbfb4c3027aaa2ce519641b38d02032b2e32c2525750fa5d68cad5dadce52f
 size 3424043887

checkpoint-4500/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7ef99e08af1d8081528f028944c1e2b864a409536cf4835c14325ef28f3c415f
+oid sha256:510f35f120a516dc8542b9fd7f1dbea2ac61d60efde7590c4fba85afd28ba517
 size 623

checkpoint-4500/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.7490401864051819,
-  "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e4l57-l/checkpoint-4000",
-  "epoch": 0.5065856129685917,
+  "best_metric": 0.7251659035682678,
+  "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e4l57-l/checkpoint-4500",
+  "epoch": 0.47994880546075086,
   "eval_steps": 500,
   "global_step": 4500,
   "is_hyper_param_search": false,
@@ -9,145 +9,145 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.056287290329843524,
-      "grad_norm": 150.1642303466797,
-      "learning_rate": 4.929640887087696e-07,
-      "loss": 0.3486,
+      "epoch": 0.05332764505119454,
+      "grad_norm": 414.2552795410156,
+      "learning_rate": 4.911120591581342e-07,
+      "loss": 0.3809,
       "step": 500
     },
     {
-      "epoch": 0.056287290329843524,
-      "eval_loss": 1.1265727281570435,
-      "eval_runtime": 123.7806,
-      "eval_samples_per_second": 15.948,
-      "eval_steps_per_second": 1.995,
+      "epoch": 0.05332764505119454,
+      "eval_loss": 1.1109352111816406,
+      "eval_runtime": 63.6734,
+      "eval_samples_per_second": 15.501,
+      "eval_steps_per_second": 1.947,
       "step": 500
     },
     {
-      "epoch": 0.11257458065968705,
-      "grad_norm": 366.6913146972656,
-      "learning_rate": 4.859281774175391e-07,
-      "loss": 0.2733,
+      "epoch": 0.10665529010238908,
+      "grad_norm": 45.880027770996094,
+      "learning_rate": 4.822241183162685e-07,
+      "loss": 0.2649,
       "step": 1000
     },
     {
-      "epoch": 0.11257458065968705,
-      "eval_loss": 0.9742079377174377,
-      "eval_runtime": 125.8047,
-      "eval_samples_per_second": 15.691,
-      "eval_steps_per_second": 1.963,
+      "epoch": 0.10665529010238908,
+      "eval_loss": 0.9588962197303772,
+      "eval_runtime": 62.6825,
+      "eval_samples_per_second": 15.746,
+      "eval_steps_per_second": 1.978,
       "step": 1000
     },
     {
-      "epoch": 0.16886187098953057,
-      "grad_norm": 555.5016479492188,
-      "learning_rate": 4.788922661263087e-07,
-      "loss": 0.1851,
+      "epoch": 0.1599829351535836,
+      "grad_norm": 432.9915466308594,
+      "learning_rate": 4.733361774744027e-07,
+      "loss": 0.2104,
       "step": 1500
     },
     {
-      "epoch": 0.16886187098953057,
-      "eval_loss": 0.9162159562110901,
-      "eval_runtime": 126.7606,
-      "eval_samples_per_second": 15.573,
-      "eval_steps_per_second": 1.949,
+      "epoch": 0.1599829351535836,
+      "eval_loss": 0.90110844373703,
+      "eval_runtime": 62.6757,
+      "eval_samples_per_second": 15.748,
+      "eval_steps_per_second": 1.978,
       "step": 1500
     },
     {
-      "epoch": 0.2251491613193741,
-      "grad_norm": 0.423260897397995,
-      "learning_rate": 4.7185635483507824e-07,
-      "loss": 0.1973,
+      "epoch": 0.21331058020477817,
+      "grad_norm": 0.014208819717168808,
+      "learning_rate": 4.6444823663253695e-07,
+      "loss": 0.1849,
       "step": 2000
     },
     {
-      "epoch": 0.2251491613193741,
-      "eval_loss": 0.8716973662376404,
-      "eval_runtime": 127.6955,
-      "eval_samples_per_second": 15.459,
-      "eval_steps_per_second": 1.934,
+      "epoch": 0.21331058020477817,
+      "eval_loss": 0.8570474982261658,
+      "eval_runtime": 63.7296,
+      "eval_samples_per_second": 15.487,
+      "eval_steps_per_second": 1.946,
       "step": 2000
     },
     {
-      "epoch": 0.2814364516492176,
-      "grad_norm": 6.840139389038086,
-      "learning_rate": 4.6482044354384774e-07,
-      "loss": 0.1881,
+      "epoch": 0.2666382252559727,
+      "grad_norm": 174.73324584960938,
+      "learning_rate": 4.5556029579067116e-07,
+      "loss": 0.2056,
       "step": 2500
     },
     {
-      "epoch": 0.2814364516492176,
-      "eval_loss": 0.830047070980072,
-      "eval_runtime": 128.0771,
-      "eval_samples_per_second": 15.413,
-      "eval_steps_per_second": 1.929,
+      "epoch": 0.2666382252559727,
+      "eval_loss": 0.7895939350128174,
+      "eval_runtime": 63.8387,
+      "eval_samples_per_second": 15.461,
+      "eval_steps_per_second": 1.942,
       "step": 2500
     },
     {
-      "epoch": 0.33772374197906113,
-      "grad_norm": 729.8377685546875,
-      "learning_rate": 4.577845322526173e-07,
-      "loss": 0.1695,
+      "epoch": 0.3199658703071672,
+      "grad_norm": 9.852073823424234e-09,
+      "learning_rate": 4.4667235494880547e-07,
+      "loss": 0.159,
       "step": 3000
     },
     {
-      "epoch": 0.33772374197906113,
-      "eval_loss": 0.8265627026557922,
-      "eval_runtime": 128.052,
-      "eval_samples_per_second": 15.416,
-      "eval_steps_per_second": 1.929,
+      "epoch": 0.3199658703071672,
+      "eval_loss": 0.7647623419761658,
+      "eval_runtime": 63.9155,
+      "eval_samples_per_second": 15.442,
+      "eval_steps_per_second": 1.94,
       "step": 3000
     },
     {
-      "epoch": 0.39401103230890466,
-      "grad_norm": 7.96999938756926e-06,
-      "learning_rate": 4.507486209613869e-07,
-      "loss": 0.155,
+      "epoch": 0.37329351535836175,
+      "grad_norm": 0.0025215104687958956,
+      "learning_rate": 4.377844141069397e-07,
+      "loss": 0.1696,
       "step": 3500
     },
     {
-      "epoch": 0.39401103230890466,
-      "eval_loss": 0.7721038460731506,
-      "eval_runtime": 128.1871,
-      "eval_samples_per_second": 15.399,
-      "eval_steps_per_second": 1.927,
+      "epoch": 0.37329351535836175,
+      "eval_loss": 0.7638036012649536,
+      "eval_runtime": 63.8003,
+      "eval_samples_per_second": 15.47,
+      "eval_steps_per_second": 1.944,
       "step": 3500
     },
     {
-      "epoch": 0.4502983226387482,
-      "grad_norm": 0.006555848754942417,
-      "learning_rate": 4.4371270967015645e-07,
-      "loss": 0.1216,
+      "epoch": 0.42662116040955633,
+      "grad_norm": 2.4611830711364746,
+      "learning_rate": 4.2889647326507393e-07,
+      "loss": 0.1125,
       "step": 4000
     },
     {
-      "epoch": 0.4502983226387482,
-      "eval_loss": 0.7490401864051819,
-      "eval_runtime": 127.5516,
-      "eval_samples_per_second": 15.476,
-      "eval_steps_per_second": 1.936,
+      "epoch": 0.42662116040955633,
+      "eval_loss": 0.7486374378204346,
+      "eval_runtime": 63.9847,
+      "eval_samples_per_second": 15.426,
+      "eval_steps_per_second": 1.938,
       "step": 4000
     },
     {
-      "epoch": 0.5065856129685917,
-      "grad_norm": 0.0004515685432124883,
-      "learning_rate": 4.36676798378926e-07,
-      "loss": 0.1531,
+      "epoch": 0.47994880546075086,
+      "grad_norm": 5.230295658111572,
+      "learning_rate": 4.2000853242320814e-07,
+      "loss": 0.1322,
       "step": 4500
     },
     {
-      "epoch": 0.5065856129685917,
-      "eval_loss": 0.7533182501792908,
-      "eval_runtime": 127.4797,
-      "eval_samples_per_second": 15.485,
-      "eval_steps_per_second": 1.938,
+      "epoch": 0.47994880546075086,
+      "eval_loss": 0.7251659035682678,
+      "eval_runtime": 64.4782,
+      "eval_samples_per_second": 15.308,
+      "eval_steps_per_second": 1.923,
       "step": 4500
     }
   ],
   "logging_steps": 500,
-  "max_steps": 35532,
+  "max_steps": 28128,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 4,
+  "num_train_epochs": 3,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {

checkpoint-4500/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:319be2a769ef7117b6826908976f080f9082f159f4f0ad75da1a151ad29d9bff
+oid sha256:057f6bbf1ff3e34abda0863071bbf5fdcc7edb1102f04dba05c7a15087bc3396
 size 4847