End of training

README.md

@@ -3,6 +3,8 @@ library_name: transformers
 license: llama3.1
 base_model: mlfoundations-dev/llama3-1_8b_oh_v3.1_wo_camel_ai_math
 tags:
+- llama-factory
+- full
 - trl
 - dpo
 - llama-factory
@@ -17,7 +19,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # simpo-oh_v3.1_wo_camel_ai_math
 
-This model is a fine-tuned version of [mlfoundations-dev/llama3-1_8b_oh_v3.1_wo_camel_ai_math](https://huggingface.co/mlfoundations-dev/llama3-1_8b_oh_v3.1_wo_camel_ai_math) on an unknown dataset.
+This model is a fine-tuned version of [mlfoundations-dev/llama3-1_8b_oh_v3.1_wo_camel_ai_math](https://huggingface.co/mlfoundations-dev/llama3-1_8b_oh_v3.1_wo_camel_ai_math) on the mlfoundations-dev/gemma2-ultrafeedback-armorm dataset.
 It achieves the following results on the evaluation set:
 - Loss: 2.3920
 - Rewards/chosen: -43.0742

all_results.json

@@ -0,0 +1,20 @@
+{
+    "epoch": 0.9997172745264349,
+    "eval_logits/chosen": -1.1595183610916138,
+    "eval_logits/rejected": -1.1711369752883911,
+    "eval_logps/chosen": -4.307417392730713,
+    "eval_logps/rejected": -5.051736831665039,
+    "eval_loss": 2.3919546604156494,
+    "eval_rewards/accuracies": 0.8038102388381958,
+    "eval_rewards/chosen": -43.07417297363281,
+    "eval_rewards/margins": 7.443192958831787,
+    "eval_rewards/rejected": -50.517372131347656,
+    "eval_runtime": 100.3048,
+    "eval_samples_per_second": 29.699,
+    "eval_steps_per_second": 1.864,
+    "total_flos": 134366991482880.0,
+    "train_loss": 3.293350306571339,
+    "train_runtime": 7617.916,
+    "train_samples_per_second": 7.429,
+    "train_steps_per_second": 0.058
+}

eval_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 0.9997172745264349,
+    "eval_logits/chosen": -1.1595183610916138,
+    "eval_logits/rejected": -1.1711369752883911,
+    "eval_logps/chosen": -4.307417392730713,
+    "eval_logps/rejected": -5.051736831665039,
+    "eval_loss": 2.3919546604156494,
+    "eval_rewards/accuracies": 0.8038102388381958,
+    "eval_rewards/chosen": -43.07417297363281,
+    "eval_rewards/margins": 7.443192958831787,
+    "eval_rewards/rejected": -50.517372131347656,
+    "eval_runtime": 100.3048,
+    "eval_samples_per_second": 29.699,
+    "eval_steps_per_second": 1.864
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 0.9997172745264349,
+    "total_flos": 134366991482880.0,
+    "train_loss": 3.293350306571339,
+    "train_runtime": 7617.916,
+    "train_samples_per_second": 7.429,
+    "train_steps_per_second": 0.058
+}