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: 1.0329
-- Rewards/chosen: -0.0989
-- Rewards/rejected: -0.1202
-- Rewards/accuracies: 0.5457
-- Rewards/margins: 0.0213
-- Logps/rejected: -36.1669
-- Logps/chosen: -32.6081
-- Logits/rejected: 98.7340
-- Logits/chosen: 98.7641
+- Loss: 0.9997
+- Rewards/chosen: -0.0473
+- Rewards/rejected: -0.0990
+- Rewards/accuracies: 0.5395
+- Rewards/margins: 0.0517
+- Logps/rejected: -36.1316
+- Logps/chosen: -32.5220
+- Logits/rejected: 98.7129
+- Logits/chosen: 98.7435
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": 98.74346160888672,
+    "eval_logits/rejected": 98.71293640136719,
+    "eval_logps/chosen": -32.52199172973633,
+    "eval_logps/rejected": -36.13157272338867,
+    "eval_loss": 0.9996863603591919,
+    "eval_rewards/accuracies": 0.5394518375396729,
+    "eval_rewards/chosen": -0.04729102551937103,
+    "eval_rewards/margins": 0.05173712968826294,
+    "eval_rewards/rejected": -0.09902816265821457,
+    "eval_runtime": 103.8083,
+    "eval_samples": 343,
+    "eval_samples_per_second": 3.304,
+    "eval_steps_per_second": 0.414,
     "train_loss": 0.746250953302755,
     "train_runtime": 2556.1232,
     "train_samples": 3079,

eval_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": 99.08063507080078,
-    "eval_logits/rejected": 99.07344818115234,
-    "eval_logps/chosen": -32.34668731689453,
-    "eval_logps/rejected": -35.868736267089844,
-    "eval_loss": 1.664738655090332,
-    "eval_rewards/accuracies": 0.510797381401062,
-    "eval_rewards/chosen": 0.08684198558330536,
-    "eval_rewards/margins": -0.0011658848961815238,
-    "eval_rewards/rejected": 0.08800788223743439,
-    "eval_runtime": 103.8246,
+    "eval_logits/chosen": 98.74346160888672,
+    "eval_logits/rejected": 98.71293640136719,
+    "eval_logps/chosen": -32.52199172973633,
+    "eval_logps/rejected": -36.13157272338867,
+    "eval_loss": 0.9996863603591919,
+    "eval_rewards/accuracies": 0.5394518375396729,
+    "eval_rewards/chosen": -0.04729102551937103,
+    "eval_rewards/margins": 0.05173712968826294,
+    "eval_rewards/rejected": -0.09902816265821457,
+    "eval_runtime": 103.8083,
     "eval_samples": 343,
     "eval_samples_per_second": 3.304,
     "eval_steps_per_second": 0.414