End of training

README.md

@@ -1,9 +1,15 @@
 ---
 base_model: princeton-nlp/Llama-3-Base-8B-SFT
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: llama3-8b-mypo3_sim-full-beta12.5-lr4e-7
   results: []
@@ -14,17 +20,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama3-8b-mypo3_sim-full-beta12.5-lr4e-7
 
-This model is a fine-tuned version of [princeton-nlp/Llama-3-Base-8B-SFT](https://huggingface.co/princeton-nlp/Llama-3-Base-8B-SFT) on an unknown dataset.
+This model is a fine-tuned version of [princeton-nlp/Llama-3-Base-8B-SFT](https://huggingface.co/princeton-nlp/Llama-3-Base-8B-SFT) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.3769
-- Rewards/chosen: 0.0610
-- Rewards/rejected: -0.3706
-- Rewards/accuracies: 0.7460
-- Rewards/margins: 0.4315
-- Logps/rejected: -1.5191
-- Logps/chosen: -1.2663
-- Logits/rejected: -1.1038
-- Logits/chosen: -1.0760
+- Loss: 1.3762
+- Rewards/chosen: 0.0655
+- Rewards/rejected: -0.3701
+- Rewards/accuracies: 0.7560
+- Rewards/margins: 0.4356
+- Logps/rejected: -1.5190
+- Logps/chosen: -1.2659
+- Logits/rejected: -1.1037
+- Logits/chosen: -1.0759
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 0.9997382884061764,
+    "eval_logits/chosen": -1.0759395360946655,
+    "eval_logits/rejected": -1.1037203073501587,
+    "eval_logps/chosen": -1.2659075260162354,
+    "eval_logps/rejected": -1.5190186500549316,
+    "eval_loss": 1.376198172569275,
+    "eval_rewards/accuracies": 0.7559523582458496,
+    "eval_rewards/chosen": 0.06548410654067993,
+    "eval_rewards/margins": 0.435606449842453,
+    "eval_rewards/rejected": -0.37012240290641785,
+    "eval_runtime": 262.4629,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 7.62,
+    "eval_steps_per_second": 0.24,
     "total_flos": 0.0,
     "train_loss": 1.393932879782472,
     "train_runtime": 26759.5867,

config.json

@@ -24,6 +24,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.43.1",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 128256
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 0.9997382884061764,
+    "eval_logits/chosen": -1.0759395360946655,
+    "eval_logits/rejected": -1.1037203073501587,
+    "eval_logps/chosen": -1.2659075260162354,
+    "eval_logps/rejected": -1.5190186500549316,
+    "eval_loss": 1.376198172569275,
+    "eval_rewards/accuracies": 0.7559523582458496,
+    "eval_rewards/chosen": 0.06548410654067993,
+    "eval_rewards/margins": 0.435606449842453,
+    "eval_rewards/rejected": -0.37012240290641785,
+    "eval_runtime": 262.4629,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 7.62,
+    "eval_steps_per_second": 0.24
+}