Training in progress, step 12000, checkpoint

checkpoint-12000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5bd12d8362daaf60f5abd5118192a668eeac0c72e420f62d044e521215afe9c3
+oid sha256:543084a3442ec1c0a0d39378ccab4487b4a5fea4cae7517bab6b8bb006701d4b
 size 51007160

checkpoint-12000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5aba2580f0360fc17ab651ae2944db201138491993c780470373e7a794197c0e
+oid sha256:1244bcdf6e37c78740e607dc4a4c925b59a3395202e0339c6802a6ccc31c26e4
 size 102078202

checkpoint-12000/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a03f2243b9ad6f78a986344cde84785337e9924ed8d0fa0226f60c36e8fd90bd
+oid sha256:8478f09044f69f1d3fcf5c4c89cff15a9ac066c7a6f21e25a4822fb4b98a5e39
 size 14244

checkpoint-12000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 4.171908855438232,
-  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/fr_clm/childes_42/checkpoint-12000",
-  "epoch": 18.867924528301888,
+  "best_metric": 4.970981597900391,
+  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/de_clm/childes_42/checkpoint-12000",
+  "epoch": 12.578616352201259,
   "eval_steps": 2000,
   "global_step": 12000,
   "is_hyper_param_search": false,
@@ -9,79 +9,79 @@
   "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
     }
   ],
   "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": {
@@ -95,7 +95,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3102504217214976.0,
+  "total_flos": 3102794094231552.0,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null

checkpoint-12000/training_args.bin

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