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: 2.2604
-- Rewards/chosen: 0.0453
-- Rewards/rejected: 0.0322
-- Rewards/accuracies: 0.4880
-- Rewards/margins: 0.0131
-- Logps/rejected: -35.9021
-- Logps/chosen: -32.3526
-- Logits/rejected: 98.9966
-- Logits/chosen: 99.0055
+- Loss: 2.3100
+- Rewards/chosen: 0.0457
+- Rewards/rejected: 0.0544
+- Rewards/accuracies: 0.5216
+- Rewards/margins: -0.0087
+- Logps/rejected: -35.8577
+- Logps/chosen: -32.3518
+- Logits/rejected: 98.9864
+- Logits/chosen: 98.9943
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": 98.9942626953125,
+    "eval_logits/rejected": 98.98635864257812,
+    "eval_logps/chosen": -32.3518180847168,
+    "eval_logps/rejected": -35.857688903808594,
+    "eval_loss": 2.3100497722625732,
+    "eval_rewards/accuracies": 0.5215947031974792,
+    "eval_rewards/chosen": 0.045678745955228806,
+    "eval_rewards/margins": -0.00873997900635004,
+    "eval_rewards/rejected": 0.05441872030496597,
+    "eval_runtime": 103.7492,
+    "eval_samples": 343,
+    "eval_samples_per_second": 3.306,
+    "eval_steps_per_second": 0.414,
     "train_loss": 2.572123973400562,
     "train_runtime": 2556.3668,
     "train_samples": 3079,

eval_results.json

@@ -1,15 +1,15 @@
 {
     "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_logits/chosen": 98.9942626953125,
+    "eval_logits/rejected": 98.98635864257812,
+    "eval_logps/chosen": -32.3518180847168,
+    "eval_logps/rejected": -35.857688903808594,
+    "eval_loss": 2.3100497722625732,
+    "eval_rewards/accuracies": 0.5215947031974792,
+    "eval_rewards/chosen": 0.045678745955228806,
+    "eval_rewards/margins": -0.00873997900635004,
+    "eval_rewards/rejected": 0.05441872030496597,
+    "eval_runtime": 103.7492,
     "eval_samples": 343,
     "eval_samples_per_second": 3.306,
     "eval_steps_per_second": 0.414