Model save

README.md

@@ -26,7 +26,7 @@ print(output["generated_text"])
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/mingzeli/PreThink_MemAgent/runs/c5tij9g9) 
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/mingzeli/PreThink_MemAgent/runs/6pz32z57) 
 
 
 This model was trained with SFT.

all_results.json

@@ -1,8 +1,13 @@
 {
+    "eval_loss": 0.5437560081481934,
+    "eval_runtime": 2.2423,
+    "eval_samples": 92,
+    "eval_samples_per_second": 41.029,
+    "eval_steps_per_second": 5.352,
     "total_flos": 21958899171328.0,
-    "train_loss": 0.744699491904332,
-    "train_runtime": 317.873,
+    "train_loss": 0.0,
+    "train_runtime": 1.4645,
     "train_samples": 821,
-    "train_samples_per_second": 7.748,
-    "train_steps_per_second": 0.123
+    "train_samples_per_second": 1681.768,
+    "train_steps_per_second": 26.63
 }

eval_results.json

@@ -0,0 +1,7 @@
+{
+    "eval_loss": 0.5437560081481934,
+    "eval_runtime": 2.2423,
+    "eval_samples": 92,
+    "eval_samples_per_second": 41.029,
+    "eval_steps_per_second": 5.352
+}

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 21958899171328.0,
-    "train_loss": 0.744699491904332,
-    "train_runtime": 317.873,
+    "train_loss": 0.0,
+    "train_runtime": 1.4645,
     "train_samples": 821,
-    "train_samples_per_second": 7.748,
-    "train_steps_per_second": 0.123
+    "train_samples_per_second": 1681.768,
+    "train_steps_per_second": 26.63
 }

trainer_state.json

@@ -352,10 +352,10 @@
       "epoch": 3.0,
       "step": 39,
       "total_flos": 21958899171328.0,
-      "train_loss": 0.744699491904332,
-      "train_runtime": 317.873,
-      "train_samples_per_second": 7.748,
-      "train_steps_per_second": 0.123
+      "train_loss": 0.0,
+      "train_runtime": 1.4645,
+      "train_samples_per_second": 1681.768,
+      "train_steps_per_second": 26.63
     }
   ],
   "logging_steps": 1,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9bd5b9b81772425f2ca2d35ad7d8a3b7ce1df0e42aa7fc718f442a5c6f551fe3
+oid sha256:7abbf6d93c48bb7a1e3846a182c4aed4be2d9d92eb50729f609e3a287e235b19
 size 7889