End of training

README.md

@@ -1,10 +1,13 @@
 ---
 library_name: peft
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
 base_model: NorLLM-AI/NorMistral-7B
+datasets:
+- hugodk-sch/aftonposten_title_prefs
 model-index:
 - name: norllm-ai-normistral-7b-align-scan
   results: []
@@ -15,17 +18,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # norllm-ai-normistral-7b-align-scan
 
-This model is a fine-tuned version of [NorLLM-AI/NorMistral-7B](https://huggingface.co/NorLLM-AI/NorMistral-7B) on the None dataset.
+This model is a fine-tuned version of [data/norllm-ai-normistral-7b-sft-qlora](https://huggingface.co/data/norllm-ai-normistral-7b-sft-qlora) on the hugodk-sch/aftonposten_title_prefs dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.6904
-- Rewards/chosen: 0.0126
-- Rewards/rejected: -0.0056
-- Rewards/accuracies: 0.5573
-- Rewards/margins: 0.0181
-- Logps/rejected: -34.7121
-- Logps/chosen: -31.2510
-- Logits/rejected: -2.8041
-- Logits/chosen: -2.8065
+- Loss: 1.6120
+- Rewards/chosen: 0.0138
+- Rewards/rejected: -0.0163
+- Rewards/accuracies: 0.5689
+- Rewards/margins: 0.0302
+- Logps/rejected: -34.7390
+- Logps/chosen: -31.2479
+- Logits/rejected: -2.8036
+- Logits/chosen: -2.8061
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -2.806074619293213,
+    "eval_logits/rejected": -2.803609848022461,
+    "eval_logps/chosen": -31.247922897338867,
+    "eval_logps/rejected": -34.73902893066406,
+    "eval_loss": 1.6120386123657227,
+    "eval_rewards/accuracies": 0.5689368844032288,
+    "eval_rewards/chosen": 0.01381184346973896,
+    "eval_rewards/margins": 0.03016132488846779,
+    "eval_rewards/rejected": -0.016349485144019127,
+    "eval_runtime": 112.7318,
+    "eval_samples": 343,
+    "eval_samples_per_second": 3.043,
+    "eval_steps_per_second": 0.381,
     "train_loss": 1.7377146052075672,
     "train_runtime": 2719.6001,
     "train_samples": 3079,

eval_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": -2.808791399002075,
-    "eval_logits/rejected": -2.8062777519226074,
-    "eval_logps/chosen": -31.203502655029297,
-    "eval_logps/rejected": -34.66593933105469,
-    "eval_loss": 0.629190981388092,
-    "eval_rewards/accuracies": 0.5166113376617432,
-    "eval_rewards/chosen": 0.05526338145136833,
-    "eval_rewards/margins": 0.03271109238266945,
-    "eval_rewards/rejected": 0.022552290931344032,
-    "eval_runtime": 112.7274,
+    "eval_logits/chosen": -2.806074619293213,
+    "eval_logits/rejected": -2.803609848022461,
+    "eval_logps/chosen": -31.247922897338867,
+    "eval_logps/rejected": -34.73902893066406,
+    "eval_loss": 1.6120386123657227,
+    "eval_rewards/accuracies": 0.5689368844032288,
+    "eval_rewards/chosen": 0.01381184346973896,
+    "eval_rewards/margins": 0.03016132488846779,
+    "eval_rewards/rejected": -0.016349485144019127,
+    "eval_runtime": 112.7318,
     "eval_samples": 343,
     "eval_samples_per_second": 3.043,
     "eval_steps_per_second": 0.381