End of training

README.md

@@ -3,9 +3,15 @@ library_name: transformers
 license: llama3.1
 base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
 tags:
+- alignment-handbook
 - trl
 - cpo
 - generated_from_trainer
+- trl
+- cpo
+- generated_from_trainer
+datasets:
+- princeton-nlp/llama3-ultrafeedback
 model-index:
 - name: llama3.1-cpo_j-full-0911
   results: []
@@ -16,7 +22,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama3.1-cpo_j-full-0911
 
-This model is a fine-tuned version of [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) on an unknown dataset.
+This model is a fine-tuned version of [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) on the princeton-nlp/llama3-ultrafeedback dataset.
 It achieves the following results on the evaluation set:
 - Loss: 1.4373
 - Rewards/chosen: -14.1493

all_results.json

@@ -1,5 +1,19 @@
 {
     "epoch": 4.9927766541462,
+    "eval_logits/chosen": -0.14757364988327026,
+    "eval_logits/rejected": -0.11364421248435974,
+    "eval_logps/chosen": -141.49264526367188,
+    "eval_logps/rejected": -155.7095184326172,
+    "eval_loss": 1.4373149871826172,
+    "eval_nll_loss": 0.17254449427127838,
+    "eval_rewards/accuracies": 0.654347836971283,
+    "eval_rewards/chosen": -14.149263381958008,
+    "eval_rewards/margins": 1.4216874837875366,
+    "eval_rewards/rejected": -15.570951461791992,
+    "eval_runtime": 76.603,
+    "eval_samples": 1826,
+    "eval_samples_per_second": 23.837,
+    "eval_steps_per_second": 1.501,
     "total_flos": 0.0,
     "train_loss": 0.5995175864961412,
     "train_runtime": 46944.6998,

config.json

@@ -34,6 +34,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.44.2",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 128256
 }

eval_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 4.9927766541462,
+    "eval_logits/chosen": -0.14757364988327026,
+    "eval_logits/rejected": -0.11364421248435974,
+    "eval_logps/chosen": -141.49264526367188,
+    "eval_logps/rejected": -155.7095184326172,
+    "eval_loss": 1.4373149871826172,
+    "eval_nll_loss": 0.17254449427127838,
+    "eval_rewards/accuracies": 0.654347836971283,
+    "eval_rewards/chosen": -14.149263381958008,
+    "eval_rewards/margins": 1.4216874837875366,
+    "eval_rewards/rejected": -15.570951461791992,
+    "eval_runtime": 76.603,
+    "eval_samples": 1826,
+    "eval_samples_per_second": 23.837,
+    "eval_steps_per_second": 1.501
+}

runs/Sep11_13-48-23_jjb_prism_dev2/events.out.tfevents.1726077898.jjb_prism_dev2.69219.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:201374446b81e619a03bbc646e1d837aef2ab12c5ad674a9763f6d19954a873a
+size 880