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: 0.9012
-- Rewards/chosen: -0.0642
-- Rewards/rejected: -0.1770
-- Rewards/accuracies: 0.5656
-- Rewards/margins: 0.1128
-- Logps/rejected: -34.9511
-- Logps/chosen: -31.3742
-- Logits/rejected: -2.8184
-- Logits/chosen: -2.8208
+- Loss: 0.8988
+- Rewards/chosen: -0.0761
+- Rewards/rejected: -0.1908
+- Rewards/accuracies: 0.5772
+- Rewards/margins: 0.1147
+- Logps/rejected: -34.9707
+- Logps/chosen: -31.3912
+- Logits/rejected: -2.8188
+- Logits/chosen: -2.8213
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -2.821291208267212,
+    "eval_logits/rejected": -2.818777084350586,
+    "eval_logps/chosen": -31.391199111938477,
+    "eval_logps/rejected": -34.970741271972656,
+    "eval_loss": 0.8988423347473145,
+    "eval_rewards/accuracies": 0.5772424936294556,
+    "eval_rewards/chosen": -0.07612106204032898,
+    "eval_rewards/margins": 0.11468725651502609,
+    "eval_rewards/rejected": -0.19080831110477448,
+    "eval_runtime": 112.7622,
+    "eval_samples": 343,
+    "eval_samples_per_second": 3.042,
+    "eval_steps_per_second": 0.381,
     "train_loss": 0.6684874992866021,
     "train_runtime": 2718.5428,
     "train_samples": 3079,

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": -2.805530071258545,
-    "eval_logits/rejected": -2.802974224090576,
-    "eval_logps/chosen": -31.315858840942383,
-    "eval_logps/rejected": -34.838993072509766,
-    "eval_loss": 6.074533939361572,
-    "eval_rewards/accuracies": 0.5660299062728882,
-    "eval_rewards/chosen": -0.006680913269519806,
-    "eval_rewards/margins": 0.021486198529601097,
-    "eval_rewards/rejected": -0.0281671155244112,
-    "eval_runtime": 112.8101,
+    "eval_logits/chosen": -2.821291208267212,
+    "eval_logits/rejected": -2.818777084350586,
+    "eval_logps/chosen": -31.391199111938477,
+    "eval_logps/rejected": -34.970741271972656,
+    "eval_loss": 0.8988423347473145,
+    "eval_rewards/accuracies": 0.5772424936294556,
+    "eval_rewards/chosen": -0.07612106204032898,
+    "eval_rewards/margins": 0.11468725651502609,
+    "eval_rewards/rejected": -0.19080831110477448,
+    "eval_runtime": 112.7622,
     "eval_samples": 343,
-    "eval_samples_per_second": 3.041,
+    "eval_samples_per_second": 3.042,
     "eval_steps_per_second": 0.381
 }