Upload 4 files

adapter_config.json

@@ -15,17 +15,17 @@
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 32,
+  "r": 16,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "up_proj",
-    "k_proj",
-    "q_proj",
-    "gate_proj",
     "down_proj",
+    "gate_proj",
+    "o_proj",
     "v_proj",
-    "o_proj"
+    "k_proj",
+    "q_proj",
+    "up_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_rslora": false

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a9ce050e4e4496519431daaf68b3ed0aebfbb4002facbbaf7ae1da64ad590a53
-size 156926880
+oid sha256:7f81d394abe35a4ee6127ee7258b849a9a14cc2614dd80c4370b456ce23b0bda
+size 78480072

trainer_state.json

@@ -0,0 +1,186 @@
+{
+  "best_metric": 0.6502026915550232,
+  "best_model_checkpoint": "gemma2_on_korean_summary/checkpoint-220",
+  "epoch": 2.8947368421052633,
+  "eval_steps": 20,
+  "global_step": 220,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9722719192504883,
+      "learning_rate": 2e-05,
+      "loss": 1.1522,
+      "step": 20
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.033677101135254,
+      "eval_runtime": 209.4976,
+      "eval_samples_per_second": 0.955,
+      "eval_steps_per_second": 0.477,
+      "step": 20
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.436105489730835,
+      "learning_rate": 4e-05,
+      "loss": 0.9522,
+      "step": 40
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 0.8600569367408752,
+      "eval_runtime": 206.2633,
+      "eval_samples_per_second": 0.97,
+      "eval_steps_per_second": 0.485,
+      "step": 40
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6157077550888062,
+      "learning_rate": 4.933333333333334e-05,
+      "loss": 0.8405,
+      "step": 60
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.7875047326087952,
+      "eval_runtime": 207.822,
+      "eval_samples_per_second": 0.962,
+      "eval_steps_per_second": 0.481,
+      "step": 60
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.4430443048477173,
+      "learning_rate": 4.8e-05,
+      "loss": 0.7687,
+      "step": 80
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 0.7403695583343506,
+      "eval_runtime": 210.2054,
+      "eval_samples_per_second": 0.951,
+      "eval_steps_per_second": 0.476,
+      "step": 80
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.640892505645752,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.6901,
+      "step": 100
+    },
+    {
+      "epoch": 1.32,
+      "eval_loss": 0.7110423445701599,
+      "eval_runtime": 206.6708,
+      "eval_samples_per_second": 0.968,
+      "eval_steps_per_second": 0.484,
+      "step": 100
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.6705243587493896,
+      "learning_rate": 4.5333333333333335e-05,
+      "loss": 0.6588,
+      "step": 120
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 0.6947363018989563,
+      "eval_runtime": 208.6184,
+      "eval_samples_per_second": 0.959,
+      "eval_steps_per_second": 0.479,
+      "step": 120
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.7401701211929321,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.6698,
+      "step": 140
+    },
+    {
+      "epoch": 1.84,
+      "eval_loss": 0.6739406585693359,
+      "eval_runtime": 207.3092,
+      "eval_samples_per_second": 0.965,
+      "eval_steps_per_second": 0.482,
+      "step": 140
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.804783582687378,
+      "learning_rate": 4.266666666666667e-05,
+      "loss": 0.6075,
+      "step": 160
+    },
+    {
+      "epoch": 2.11,
+      "eval_loss": 0.6695161461830139,
+      "eval_runtime": 206.5895,
+      "eval_samples_per_second": 0.968,
+      "eval_steps_per_second": 0.484,
+      "step": 160
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.2475624084472656,
+      "learning_rate": 4.133333333333333e-05,
+      "loss": 0.5221,
+      "step": 180
+    },
+    {
+      "epoch": 2.37,
+      "eval_loss": 0.6713693141937256,
+      "eval_runtime": 209.2432,
+      "eval_samples_per_second": 0.956,
+      "eval_steps_per_second": 0.478,
+      "step": 180
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 2.268982172012329,
+      "learning_rate": 4e-05,
+      "loss": 0.4865,
+      "step": 200
+    },
+    {
+      "epoch": 2.63,
+      "eval_loss": 0.6660403609275818,
+      "eval_runtime": 205.9824,
+      "eval_samples_per_second": 0.971,
+      "eval_steps_per_second": 0.485,
+      "step": 200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.3548617362976074,
+      "learning_rate": 3.866666666666667e-05,
+      "loss": 0.5239,
+      "step": 220
+    },
+    {
+      "epoch": 2.89,
+      "eval_loss": 0.6502026915550232,
+      "eval_runtime": 206.4589,
+      "eval_samples_per_second": 0.969,
+      "eval_steps_per_second": 0.484,
+      "step": 220
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 800,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 11,
+  "save_steps": 20,
+  "total_flos": 1.917650992083763e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}