Training in progress, step 300, checkpoint

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dabadd77bc689da500c9b7eda73f22b5f0612bb18ac452eb71eaa547de6076b2
-size 69839888
+oid sha256:31c636041d5a39e5ba263480be2d95b989d0b2a2d8aaccd9baba04697672e7f5
+size 139602808

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8ddcaf4b49e6ae3a3e6bfed2eabe642fbe9475c2d5427243583e2339ea0754d1
-size 139961199
+oid sha256:438476638b8c74ff254de1792962b47cc5edb1e5249ff9000a99fafef66f4e6f
+size 279487343

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d45dcb26e076badfa1f9b7021eb697d18eae744725aec55db9cf2e55eb4872b0
+oid sha256:ad2fe0b97216cb1e43c2595ce462f1258a2fe337debcb3a721d3057202db3412
 size 14645

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:624edc83d795c4b4c219a679dcf9a8467dd8fc40fd38a2415b413d6849e9b256
+oid sha256:c439a886e19d13052bfa2021fe22ade0a0a85b83baaf6eb8e834e0feb73b8be6
 size 1465

last-checkpoint/trainer_state.json

@@ -2,9 +2,9 @@
   "best_global_step": 175,
   "best_metric": 0.0012933596735820174,
   "best_model_checkpoint": "./c2c_gemma4_e4b_qlora_checkpoint/checkpoint-175",
-  "epoch": 2.5,
+  "epoch": 3.0,
   "eval_steps": 25,
-  "global_step": 250,
+  "global_step": 300,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -318,12 +318,84 @@
       "eval_samples_per_second": 0.718,
       "eval_steps_per_second": 0.718,
       "step": 250
+    },
+    {
+      "entropy": 0.9087011106312275,
+      "epoch": 2.52,
+      "grad_norm": 0.002573611680418253,
+      "learning_rate": 3.920265780730897e-05,
+      "loss": 0.0007128582801669836,
+      "mean_token_accuracy": 1.0,
+      "num_tokens": 3959.0,
+      "step": 252
+    },
+    {
+      "entropy": 0.924984677384297,
+      "epoch": 2.64,
+      "grad_norm": 0.0005084075382910669,
+      "learning_rate": 3.122923588039867e-05,
+      "loss": 0.0006440305539096395,
+      "mean_token_accuracy": 1.0,
+      "num_tokens": 26050.0,
+      "step": 264
+    },
+    {
+      "epoch": 2.75,
+      "eval_entropy": 0.9374615728855134,
+      "eval_loss": 0.001335766282863915,
+      "eval_mean_token_accuracy": 0.9996986410021782,
+      "eval_num_tokens": 45328.0,
+      "eval_runtime": 252.969,
+      "eval_samples_per_second": 0.791,
+      "eval_steps_per_second": 0.791,
+      "step": 275
+    },
+    {
+      "entropy": 0.9492521900683641,
+      "epoch": 2.76,
+      "grad_norm": 0.0005572825903072953,
+      "learning_rate": 2.3255813953488374e-05,
+      "loss": 0.0005182567983865738,
+      "mean_token_accuracy": 1.0,
+      "num_tokens": 46997.0,
+      "step": 276
+    },
+    {
+      "entropy": 0.9195458479225636,
+      "epoch": 2.88,
+      "grad_norm": 0.0006922565517015755,
+      "learning_rate": 1.5282392026578074e-05,
+      "loss": 0.0005983434772739807,
+      "mean_token_accuracy": 1.0,
+      "num_tokens": 69490.0,
+      "step": 288
+    },
+    {
+      "entropy": 0.9362974322090546,
+      "epoch": 3.0,
+      "grad_norm": 0.0005240127211436629,
+      "learning_rate": 7.308970099667775e-06,
+      "loss": 0.0005150201808040341,
+      "mean_token_accuracy": 1.0,
+      "num_tokens": 90696.0,
+      "step": 300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.9336179578304291,
+      "eval_loss": 0.0013242117129266262,
+      "eval_mean_token_accuracy": 0.9996986410021782,
+      "eval_num_tokens": 90696.0,
+      "eval_runtime": 252.73,
+      "eval_samples_per_second": 0.791,
+      "eval_steps_per_second": 0.791,
+      "step": 300
     }
   ],
   "logging_steps": 12,
-  "max_steps": 300,
+  "max_steps": 310,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 3,
+  "num_train_epochs": 4,
   "save_steps": 25,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -337,7 +409,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.20860153697168e+16,
+  "total_flos": 1.452736382565024e+16,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d120c92ba91ddf4ca0253a099ee22d2291712c5f091a4855f1d947cb6f6a6059
+oid sha256:7ce060488f05343f0437ad90e26b0a95fbf802432a19bd58f1082d6c190ec4c8
 size 5713