End of training

README.md

@@ -1,9 +1,15 @@
 ---
 base_model: meta-llama/Llama-2-7b-hf
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: llama2-7b-dpo-full-wo-kqa_silver_wogold-ep3
   results: []
@@ -14,17 +20,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama2-7b-dpo-full-wo-kqa_silver_wogold-ep3
 
-This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on an unknown dataset.
+This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.6597
-- Rewards/chosen: 0.0473
-- Rewards/rejected: -0.0269
-- Rewards/accuracies: 0.8044
-- Rewards/margins: 0.0742
-- Logps/rejected: -1010.1399
-- Logps/chosen: -408.6627
-- Logits/rejected: -0.5930
-- Logits/chosen: 0.4742
+- Loss: 0.6588
+- Rewards/chosen: 0.0476
+- Rewards/rejected: -0.0291
+- Rewards/accuracies: 0.7912
+- Rewards/margins: 0.0767
+- Logps/rejected: -1010.3619
+- Logps/chosen: -408.6368
+- Logits/rejected: -0.5925
+- Logits/chosen: 0.4749
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": 0.47494596242904663,
+    "eval_logits/rejected": -0.5924796462059021,
+    "eval_logps/chosen": -408.63677978515625,
+    "eval_logps/rejected": -1010.3619384765625,
+    "eval_loss": 0.6588297486305237,
+    "eval_rewards/accuracies": 0.7911764979362488,
+    "eval_rewards/chosen": 0.047554861754179,
+    "eval_rewards/margins": 0.07667416334152222,
+    "eval_rewards/rejected": -0.029119295999407768,
+    "eval_runtime": 302.3402,
+    "eval_samples": 2712,
+    "eval_samples_per_second": 8.97,
+    "eval_steps_per_second": 0.281,
     "train_loss": 0.6376642821327089,
     "train_runtime": 2213.5067,
     "train_samples": 8087,

config.json

@@ -23,6 +23,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.39.0.dev0",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.0,
+    "eval_logits/chosen": 0.47494596242904663,
+    "eval_logits/rejected": -0.5924796462059021,
+    "eval_logps/chosen": -408.63677978515625,
+    "eval_logps/rejected": -1010.3619384765625,
+    "eval_loss": 0.6588297486305237,
+    "eval_rewards/accuracies": 0.7911764979362488,
+    "eval_rewards/chosen": 0.047554861754179,
+    "eval_rewards/margins": 0.07667416334152222,
+    "eval_rewards/rejected": -0.029119295999407768,
+    "eval_runtime": 302.3402,
+    "eval_samples": 2712,
+    "eval_samples_per_second": 8.97,
+    "eval_steps_per_second": 0.281
+}