Model save

README.md

@@ -27,7 +27,7 @@ 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/qe6p1l0a) 
+[<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/si3pg5xv) 
 
 
 This model was trained with SFT.

all_results.json

@@ -1,8 +1,8 @@
 {
-    "total_flos": 967966479089664.0,
-    "train_loss": 1.0463442883413772,
-    "train_runtime": 10114.1342,
-    "train_samples": 47071,
-    "train_samples_per_second": 13.962,
-    "train_steps_per_second": 0.055
+    "total_flos": 94118804062208.0,
+    "train_loss": 1.2374288660864676,
+    "train_runtime": 1146.56,
+    "train_samples": 3923,
+    "train_samples_per_second": 6.843,
+    "train_steps_per_second": 0.108
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "total_flos": 967966479089664.0,
-    "train_loss": 1.0463442883413772,
-    "train_runtime": 10114.1342,
-    "train_samples": 47071,
-    "train_samples_per_second": 13.962,
-    "train_steps_per_second": 0.055
+    "total_flos": 94118804062208.0,
+    "train_loss": 1.2374288660864676,
+    "train_runtime": 1146.56,
+    "train_samples": 3923,
+    "train_samples_per_second": 6.843,
+    "train_steps_per_second": 0.108
 }