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-full-0913
   results: []
@@ -16,18 +22,18 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama3.1-cpo-full-0913
 
-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: 1.5947
-- Rewards/chosen: -15.5964
-- Rewards/rejected: -16.3155
+- Loss: 1.5934
+- Rewards/chosen: -15.4936
+- Rewards/rejected: -16.2190
 - Rewards/accuracies: 0.6261
-- Rewards/margins: 0.7192
-- Logps/rejected: -163.1553
-- Logps/chosen: -155.9637
-- Logits/rejected: -0.4910
-- Logits/chosen: -0.5144
-- Nll Loss: 0.4262
+- Rewards/margins: 0.7255
+- Logps/rejected: -162.1901
+- Logps/chosen: -154.9355
+- Logits/rejected: -0.4926
+- Logits/chosen: -0.5160
+- Nll Loss: 0.4228
 
 ## Model description
 

all_results.json

@@ -1,5 +1,19 @@
 {
     "epoch": 0.9985553308292401,
+    "eval_logits/chosen": -0.5159785151481628,
+    "eval_logits/rejected": -0.4926067888736725,
+    "eval_logps/chosen": -154.935546875,
+    "eval_logps/rejected": -162.19007873535156,
+    "eval_loss": 1.5933648347854614,
+    "eval_nll_loss": 0.42284756898880005,
+    "eval_rewards/accuracies": 0.626086950302124,
+    "eval_rewards/chosen": -15.493555068969727,
+    "eval_rewards/margins": 0.7254539132118225,
+    "eval_rewards/rejected": -16.219009399414062,
+    "eval_runtime": 73.3207,
+    "eval_samples": 1826,
+    "eval_samples_per_second": 24.904,
+    "eval_steps_per_second": 1.568,
     "total_flos": 0.0,
     "train_loss": 1.7731637126869626,
     "train_runtime": 10231.9294,

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": 0.9985553308292401,
+    "eval_logits/chosen": -0.5159785151481628,
+    "eval_logits/rejected": -0.4926067888736725,
+    "eval_logps/chosen": -154.935546875,
+    "eval_logps/rejected": -162.19007873535156,
+    "eval_loss": 1.5933648347854614,
+    "eval_nll_loss": 0.42284756898880005,
+    "eval_rewards/accuracies": 0.626086950302124,
+    "eval_rewards/chosen": -15.493555068969727,
+    "eval_rewards/margins": 0.7254539132118225,
+    "eval_rewards/rejected": -16.219009399414062,
+    "eval_runtime": 73.3207,
+    "eval_samples": 1826,
+    "eval_samples_per_second": 24.904,
+    "eval_steps_per_second": 1.568
+}

runs/Sep13_20-12-38_jjb_prism_dev2/events.out.tfevents.1726237144.jjb_prism_dev2.78070.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1570fa634f69dcd6bb8d4dbfb099a4b93ac806654bd8198eaeccc548d3a8c84
+size 880