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: 0.7744
-- Rewards/chosen: -0.0855
-- Rewards/rejected: -0.2061
-- Rewards/accuracies: 0.5428
-- Rewards/margins: 0.1206
-- Logps/rejected: -36.2242
-- Logps/chosen: -32.5501
-- Logits/rejected: 98.7517
-- Logits/chosen: 98.7722
+- Loss: 0.7822
+- Rewards/chosen: -0.1005
+- Rewards/rejected: -0.2047
+- Rewards/accuracies: 0.5249
+- Rewards/margins: 0.1042
+- Logps/rejected: -36.2224
+- Logps/chosen: -32.5688
+- Logits/rejected: 98.7467
+- Logits/chosen: 98.7645
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": 98.7645034790039,
+    "eval_logits/rejected": 98.74665832519531,
+    "eval_logps/chosen": -32.56877517700195,
+    "eval_logps/rejected": -36.22236251831055,
+    "eval_loss": 0.7821906208992004,
+    "eval_rewards/accuracies": 0.52491694688797,
+    "eval_rewards/chosen": -0.10047715157270432,
+    "eval_rewards/margins": 0.1041894406080246,
+    "eval_rewards/rejected": -0.20466658473014832,
+    "eval_runtime": 103.6205,
+    "eval_samples": 343,
+    "eval_samples_per_second": 3.31,
+    "eval_steps_per_second": 0.415,
     "train_loss": 0.6050039254225694,
     "train_runtime": 2556.17,
     "train_samples": 3079,

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": 97.9656753540039,
-    "eval_logits/rejected": 97.93604278564453,
-    "eval_logps/chosen": -33.003868103027344,
-    "eval_logps/rejected": -36.95783615112305,
-    "eval_loss": 0.6761856079101562,
-    "eval_rewards/accuracies": 0.5888704061508179,
-    "eval_rewards/chosen": -0.05606912076473236,
-    "eval_rewards/margins": 0.04306147247552872,
-    "eval_rewards/rejected": -0.09913058578968048,
-    "eval_runtime": 103.6924,
+    "eval_logits/chosen": 98.7645034790039,
+    "eval_logits/rejected": 98.74665832519531,
+    "eval_logps/chosen": -32.56877517700195,
+    "eval_logps/rejected": -36.22236251831055,
+    "eval_loss": 0.7821906208992004,
+    "eval_rewards/accuracies": 0.52491694688797,
+    "eval_rewards/chosen": -0.10047715157270432,
+    "eval_rewards/margins": 0.1041894406080246,
+    "eval_rewards/rejected": -0.20466658473014832,
+    "eval_runtime": 103.6205,
     "eval_samples": 343,
-    "eval_samples_per_second": 3.308,
+    "eval_samples_per_second": 3.31,
     "eval_steps_per_second": 0.415
 }