Model save

README.md

@@ -16,6 +16,8 @@ should probably proofread and complete it, then remove this comment. -->
 # Phi-3-mini-4k-instruct-finetuned
 
 This model was trained from scratch on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 8.6678
 
 ## Model description
 

all_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 1.0,
-    "eval_loss": 12.382317543029785,
-    "eval_runtime": 3.6879,
+    "eval_loss": 8.667840957641602,
+    "eval_runtime": 12.9661,
     "eval_samples": 145,
-    "eval_samples_per_second": 39.318,
-    "eval_steps_per_second": 19.795,
+    "eval_samples_per_second": 11.183,
+    "eval_steps_per_second": 5.63,
     "total_flos": 8592765192732672.0,
     "train_loss": 10.40311779153758,
     "train_runtime": 167.4524,

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "eval_loss": 12.382317543029785,
-    "eval_runtime": 3.6879,
+    "eval_loss": 8.667840957641602,
+    "eval_runtime": 12.9661,
     "eval_samples": 145,
-    "eval_samples_per_second": 39.318,
-    "eval_steps_per_second": 19.795
+    "eval_samples_per_second": 11.183,
+    "eval_steps_per_second": 5.63
 }

tokenizer_config.json

@@ -122,7 +122,7 @@
   "legacy": false,
   "model_max_length": 2048,
   "pad_token": "<unk>",
-  "padding_side": "right",
+  "padding_side": "left",
   "sp_model_kwargs": {},
   "tokenizer_class": "LlamaTokenizer",
   "unk_token": "<unk>",