End of training

README.md

@@ -1,11 +1,13 @@
 ---
-license: apache-2.0
 library_name: peft
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
 base_model: norallm/normistral-7b-warm
+datasets:
+- hugodk-sch/aftonposten_title_prefs
 model-index:
 - name: ap-normistral-7b-align-scan
   results: []
@@ -16,17 +18,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # ap-normistral-7b-align-scan
 
-This model is a fine-tuned version of [norallm/normistral-7b-warm](https://huggingface.co/norallm/normistral-7b-warm) on the None dataset.
+This model is a fine-tuned version of [data/ap-normistral-7b-sft-qlora](https://huggingface.co/data/ap-normistral-7b-sft-qlora) on the hugodk-sch/aftonposten_title_prefs dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4994
-- Rewards/chosen: 0.0186
-- Rewards/rejected: 0.0126
-- Rewards/accuracies: 0.5129
-- Rewards/margins: 0.0060
-- Logps/rejected: -35.8407
-- Logps/chosen: -32.2572
-- Logits/rejected: 98.2712
-- Logits/chosen: 98.2791
+- Loss: 0.4992
+- Rewards/chosen: 0.0239
+- Rewards/rejected: 0.0174
+- Rewards/accuracies: 0.4892
+- Rewards/margins: 0.0064
+- Logps/rejected: -35.7921
+- Logps/chosen: -32.2044
+- Logits/rejected: 98.2668
+- Logits/chosen: 98.2727
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": 98.27266693115234,
+    "eval_logits/rejected": 98.26683044433594,
+    "eval_logps/chosen": -32.204383850097656,
+    "eval_logps/rejected": -35.79214859008789,
+    "eval_loss": 0.49918678402900696,
+    "eval_rewards/accuracies": 0.489202618598938,
+    "eval_rewards/chosen": 0.023878853768110275,
+    "eval_rewards/margins": 0.00644117034971714,
+    "eval_rewards/rejected": 0.017437683418393135,
+    "eval_runtime": 103.7538,
+    "eval_samples": 343,
+    "eval_samples_per_second": 3.306,
+    "eval_steps_per_second": 0.414,
     "train_loss": 0.47652822160101554,
     "train_runtime": 2558.0485,
     "train_samples": 3079,

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": 98.34632873535156,
-    "eval_logits/rejected": 98.3492202758789,
-    "eval_logps/chosen": -32.23118591308594,
-    "eval_logps/rejected": -35.818275451660156,
-    "eval_loss": 0.4955456852912903,
-    "eval_rewards/accuracies": 0.5282392501831055,
-    "eval_rewards/chosen": 0.08479735255241394,
-    "eval_rewards/margins": 0.025497542694211006,
-    "eval_rewards/rejected": 0.059299811720848083,
-    "eval_runtime": 103.8037,
+    "eval_logits/chosen": 98.27266693115234,
+    "eval_logits/rejected": 98.26683044433594,
+    "eval_logps/chosen": -32.204383850097656,
+    "eval_logps/rejected": -35.79214859008789,
+    "eval_loss": 0.49918678402900696,
+    "eval_rewards/accuracies": 0.489202618598938,
+    "eval_rewards/chosen": 0.023878853768110275,
+    "eval_rewards/margins": 0.00644117034971714,
+    "eval_rewards/rejected": 0.017437683418393135,
+    "eval_runtime": 103.7538,
     "eval_samples": 343,
-    "eval_samples_per_second": 3.304,
+    "eval_samples_per_second": 3.306,
     "eval_steps_per_second": 0.414
 }