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: 1.5939
-- Rewards/chosen: 0.0537
-- Rewards/rejected: 0.0702
-- Rewards/accuracies: 0.5191
-- Rewards/margins: -0.0166
-- Logps/rejected: -35.8787
-- Logps/chosen: -32.3761
-- Logits/rejected: 98.9855
-- Logits/chosen: 98.9917
+- Loss: 1.7667
+- Rewards/chosen: 0.0466
+- Rewards/rejected: 0.0526
+- Rewards/accuracies: 0.4880
+- Rewards/margins: -0.0060
+- Logps/rejected: -35.9008
+- Logps/chosen: -32.3849
+- Logits/rejected: 98.9812
+- Logits/chosen: 98.9874
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": 98.98737335205078,
+    "eval_logits/rejected": 98.98121643066406,
+    "eval_logps/chosen": -32.38493728637695,
+    "eval_logps/rejected": -35.900760650634766,
+    "eval_loss": 1.7667114734649658,
+    "eval_rewards/accuracies": 0.48795682191848755,
+    "eval_rewards/chosen": 0.04659241810441017,
+    "eval_rewards/margins": -0.006021121051162481,
+    "eval_rewards/rejected": 0.05261354520916939,
+    "eval_runtime": 103.7641,
+    "eval_samples": 343,
+    "eval_samples_per_second": 3.306,
+    "eval_steps_per_second": 0.414,
     "train_loss": 2.2952296467570514,
     "train_runtime": 2556.4368,
     "train_samples": 3079,

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": 98.66954803466797,
-    "eval_logits/rejected": 98.6399917602539,
-    "eval_logps/chosen": -32.599971771240234,
-    "eval_logps/rejected": -36.293174743652344,
-    "eval_loss": 25.394121170043945,
-    "eval_rewards/accuracies": 0.550664484500885,
-    "eval_rewards/chosen": -0.015679799020290375,
-    "eval_rewards/margins": 0.016984686255455017,
-    "eval_rewards/rejected": -0.03266448527574539,
-    "eval_runtime": 103.7202,
+    "eval_logits/chosen": 98.98737335205078,
+    "eval_logits/rejected": 98.98121643066406,
+    "eval_logps/chosen": -32.38493728637695,
+    "eval_logps/rejected": -35.900760650634766,
+    "eval_loss": 1.7667114734649658,
+    "eval_rewards/accuracies": 0.48795682191848755,
+    "eval_rewards/chosen": 0.04659241810441017,
+    "eval_rewards/margins": -0.006021121051162481,
+    "eval_rewards/rejected": 0.05261354520916939,
+    "eval_runtime": 103.7641,
     "eval_samples": 343,
-    "eval_samples_per_second": 3.307,
-    "eval_steps_per_second": 0.415
+    "eval_samples_per_second": 3.306,
+    "eval_steps_per_second": 0.414
 }