End of training

README.md

@@ -18,9 +18,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # train_math_qa_42_1760637607
 
-This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) on an unknown dataset.
+This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) on the math_qa dataset.
 It achieves the following results on the evaluation set:
-- Loss: 3.8375
+- Loss: 0.6375
 - Num Input Tokens Seen: 77902976
 
 ## Model description

all_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 20.0,
+    "eval_loss": 0.6375064253807068,
+    "eval_runtime": 51.23,
+    "eval_samples_per_second": 58.247,
+    "eval_steps_per_second": 14.562,
+    "num_input_tokens_seen": 77902976,
+    "total_flos": 3.514797523668566e+18,
+    "train_loss": 0.18788277705373982,
+    "train_runtime": 32107.5081,
+    "train_samples_per_second": 16.727,
+    "train_steps_per_second": 4.182
+}

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 20.0,
+    "eval_loss": 0.6375064253807068,
+    "eval_runtime": 51.23,
+    "eval_samples_per_second": 58.247,
+    "eval_steps_per_second": 14.562,
+    "num_input_tokens_seen": 77902976
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 20.0,
+    "num_input_tokens_seen": 77902976,
+    "total_flos": 3.514797523668566e+18,
+    "train_loss": 0.18788277705373982,
+    "train_runtime": 32107.5081,
+    "train_samples_per_second": 16.727,
+    "train_steps_per_second": 4.182
+}