Training in progress, step 20000, checkpoint

checkpoint-20000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e731e08c4dc4ab73e89ee2177c434fb2b86ae14b6b45dae53fa7947b17be176f
+oid sha256:1b72c847d0955ffc9252a92bb77e7ae80d47871db21cf42d9f7b343299e59756
 size 51007160

checkpoint-20000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f400078591fa119447e1b10dc317036bb066abf3322022f8eac4498a312ea6a6
+oid sha256:d3e08f95cf1004f3ec7d33263c3c78cd3466d732a1aebd38f6450514c50badb5
 size 102078202

checkpoint-20000/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:62e5cee71b08a26812bb2ab1f1794e22f219eb2ff0a5855fd0f09463a54e566f
+oid sha256:28573ea73e5c3832edcbaf40a12817c319e861c6e88257535733329dae7efe4c
 size 14244

checkpoint-20000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:77e587047de45252de55a7eeb235aeebe6c43e35262d15dab334a70b654cd50f
+oid sha256:abe74028345b3917fc341e1c16987110af2c830c3a44fe9b5625ccbe85d05cee
 size 1000

checkpoint-20000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 3.970871686935425,
-  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/fr_clm/childes_42/checkpoint-20000",
-  "epoch": 31.446540880503143,
+  "best_metric": 4.692162990570068,
+  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/de_clm/childes_42/checkpoint-20000",
+  "epoch": 20.964360587002098,
   "eval_steps": 2000,
   "global_step": 20000,
   "is_hyper_param_search": false,
@@ -9,125 +9,125 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 3.1446540880503147,
-      "eval_loss": 6.666613578796387,
-      "eval_runtime": 0.6599,
-      "eval_samples_per_second": 1380.413,
-      "eval_steps_per_second": 86.37,
+      "epoch": 2.0964360587002098,
+      "eval_loss": 7.0908098220825195,
+      "eval_runtime": 0.9834,
+      "eval_samples_per_second": 1398.23,
+      "eval_steps_per_second": 87.453,
       "step": 2000
     },
     {
-      "epoch": 6.289308176100629,
-      "grad_norm": 1.431768536567688,
+      "epoch": 4.1928721174004195,
+      "grad_norm": 1.4699604511260986,
       "learning_rate": 1e-05,
-      "loss": 6.5315,
+      "loss": 6.9765,
       "step": 4000
     },
     {
-      "epoch": 6.289308176100629,
-      "eval_loss": 5.101934432983398,
-      "eval_runtime": 0.6406,
-      "eval_samples_per_second": 1422.145,
-      "eval_steps_per_second": 88.982,
+      "epoch": 4.1928721174004195,
+      "eval_loss": 5.874638557434082,
+      "eval_runtime": 0.973,
+      "eval_samples_per_second": 1413.214,
+      "eval_steps_per_second": 88.39,
       "step": 4000
     },
     {
-      "epoch": 9.433962264150944,
-      "eval_loss": 4.677049160003662,
-      "eval_runtime": 0.6439,
-      "eval_samples_per_second": 1414.724,
-      "eval_steps_per_second": 88.517,
+      "epoch": 6.289308176100629,
+      "eval_loss": 5.544075965881348,
+      "eval_runtime": 0.9898,
+      "eval_samples_per_second": 1389.228,
+      "eval_steps_per_second": 86.89,
       "step": 6000
     },
     {
-      "epoch": 12.578616352201259,
-      "grad_norm": 2.6387760639190674,
-      "learning_rate": 2e-05,
-      "loss": 4.3407,
+      "epoch": 8.385744234800839,
+      "grad_norm": 2.451873779296875,
+      "learning_rate": 1.9997500000000003e-05,
+      "loss": 5.2182,
       "step": 8000
     },
     {
-      "epoch": 12.578616352201259,
-      "eval_loss": 4.4404096603393555,
-      "eval_runtime": 0.6445,
-      "eval_samples_per_second": 1413.55,
-      "eval_steps_per_second": 88.444,
+      "epoch": 8.385744234800839,
+      "eval_loss": 5.278811454772949,
+      "eval_runtime": 0.9809,
+      "eval_samples_per_second": 1401.823,
+      "eval_steps_per_second": 87.678,
       "step": 8000
     },
     {
-      "epoch": 15.723270440251572,
-      "eval_loss": 4.286661624908447,
-      "eval_runtime": 0.6446,
-      "eval_samples_per_second": 1413.35,
-      "eval_steps_per_second": 88.431,
+      "epoch": 10.482180293501049,
+      "eval_loss": 5.099233150482178,
+      "eval_runtime": 0.9694,
+      "eval_samples_per_second": 1418.396,
+      "eval_steps_per_second": 88.714,
       "step": 10000
     },
     {
-      "epoch": 18.867924528301888,
-      "grad_norm": 2.381523609161377,
+      "epoch": 12.578616352201259,
+      "grad_norm": 2.490121841430664,
       "learning_rate": 2.9995e-05,
-      "loss": 3.8725,
+      "loss": 4.7379,
       "step": 12000
     },
     {
-      "epoch": 18.867924528301888,
-      "eval_loss": 4.171908855438232,
-      "eval_runtime": 0.6441,
-      "eval_samples_per_second": 1414.321,
-      "eval_steps_per_second": 88.492,
+      "epoch": 12.578616352201259,
+      "eval_loss": 4.970981597900391,
+      "eval_runtime": 0.9729,
+      "eval_samples_per_second": 1413.251,
+      "eval_steps_per_second": 88.392,
       "step": 12000
     },
     {
-      "epoch": 22.0125786163522,
-      "eval_loss": 4.096552848815918,
-      "eval_runtime": 0.6506,
-      "eval_samples_per_second": 1400.196,
-      "eval_steps_per_second": 87.608,
+      "epoch": 14.675052410901468,
+      "eval_loss": 4.875853061676025,
+      "eval_runtime": 0.9902,
+      "eval_samples_per_second": 1388.55,
+      "eval_steps_per_second": 86.848,
       "step": 14000
     },
     {
-      "epoch": 25.157232704402517,
-      "grad_norm": 2.4479799270629883,
-      "learning_rate": 3.999e-05,
-      "loss": 3.5953,
+      "epoch": 16.771488469601678,
+      "grad_norm": 2.485079765319824,
+      "learning_rate": 3.99925e-05,
+      "loss": 4.4249,
       "step": 16000
     },
     {
-      "epoch": 25.157232704402517,
-      "eval_loss": 4.033241271972656,
-      "eval_runtime": 0.6465,
-      "eval_samples_per_second": 1409.06,
-      "eval_steps_per_second": 88.163,
+      "epoch": 16.771488469601678,
+      "eval_loss": 4.8005266189575195,
+      "eval_runtime": 0.9709,
+      "eval_samples_per_second": 1416.144,
+      "eval_steps_per_second": 88.573,
       "step": 16000
     },
     {
-      "epoch": 28.30188679245283,
-      "eval_loss": 3.9956283569335938,
-      "eval_runtime": 0.6443,
-      "eval_samples_per_second": 1413.837,
-      "eval_steps_per_second": 88.462,
+      "epoch": 18.867924528301888,
+      "eval_loss": 4.743557453155518,
+      "eval_runtime": 0.9764,
+      "eval_samples_per_second": 1408.167,
+      "eval_steps_per_second": 88.074,
       "step": 18000
     },
     {
-      "epoch": 31.446540880503143,
-      "grad_norm": 2.266655921936035,
-      "learning_rate": 4.9985e-05,
-      "loss": 3.3822,
+      "epoch": 20.964360587002098,
+      "grad_norm": 2.311079978942871,
+      "learning_rate": 4.99875e-05,
+      "loss": 4.1842,
       "step": 20000
     },
     {
-      "epoch": 31.446540880503143,
-      "eval_loss": 3.970871686935425,
-      "eval_runtime": 0.6424,
-      "eval_samples_per_second": 1418.161,
-      "eval_steps_per_second": 88.732,
+      "epoch": 20.964360587002098,
+      "eval_loss": 4.692162990570068,
+      "eval_runtime": 0.9728,
+      "eval_samples_per_second": 1413.396,
+      "eval_steps_per_second": 88.401,
       "step": 20000
     }
   ],
   "logging_steps": 4000,
   "max_steps": 100000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 158,
+  "num_train_epochs": 105,
   "save_steps": 4000,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -141,7 +141,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5170824201633792.0,
+  "total_flos": 5171324163735552.0,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null

checkpoint-20000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e6d9a18dff8e8501ed8e1286b71e34a5ee0478c42b015fd65b8981e86e58a8c2
+oid sha256:7e7557b0dc8ef07c9db255e1167d98819525b043deb4934f580e21d716914111
 size 5368