End of training

README.md

@@ -15,11 +15,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 # smol_llama-220M-GQA-fineweb-edu-10BT-mincols-vN
 
-This model is a fine-tuned version of [BEE-spoke-data/smol_llama-220M-GQA](https://huggingface.co/BEE-spoke-data/smol_llama-220M-GQA) on an unknown dataset.
+This model is a fine-tuned version of [BEE-spoke-data/smol_llama-220M-GQA](https://huggingface.co/BEE-spoke-data/smol_llama-220M-GQA) on the BEE-spoke-data/fineweb-edu-10BT-mincols dataset.
 It achieves the following results on the evaluation set:
 - Loss: 2.7416
-- Accuracy: 0.4559
-- Num Input Tokens Seen: 10695475200
+- Accuracy: 0.4560
+- Num Input Tokens Seen: 10810818560
 
 ## Model description
 

all_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 0.9999939379610938,
+    "eval_accuracy": 0.4560332193453835,
+    "eval_loss": 2.741572141647339,
+    "eval_runtime": 5.7613,
+    "eval_samples": 300,
+    "eval_samples_per_second": 52.072,
+    "eval_steps_per_second": 6.596,
+    "num_input_tokens_seen": 10810818560,
+    "perplexity": 15.511351979678839,
+    "total_flos": 1.199767969182253e+19,
+    "train_loss": 2.7848671950609445,
+    "train_runtime": 243040.2706,
+    "train_samples": 5278746,
+    "train_samples_per_second": 21.72,
+    "train_steps_per_second": 0.085
+}

eval_results.json

@@ -0,0 +1,11 @@
+{
+    "epoch": 0.9999939379610938,
+    "eval_accuracy": 0.4560332193453835,
+    "eval_loss": 2.741572141647339,
+    "eval_runtime": 5.7613,
+    "eval_samples": 300,
+    "eval_samples_per_second": 52.072,
+    "eval_steps_per_second": 6.596,
+    "num_input_tokens_seen": 10810818560,
+    "perplexity": 15.511351979678839
+}

train_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 0.9999939379610938,
+    "num_input_tokens_seen": 10810818560,
+    "total_flos": 1.199767969182253e+19,
+    "train_loss": 2.7848671950609445,
+    "train_runtime": 243040.2706,
+    "train_samples": 5278746,
+    "train_samples_per_second": 21.72,
+    "train_steps_per_second": 0.085
+}