End of training

README.md

@@ -1,11 +1,13 @@
 ---
 library_name: peft
 tags:
+- alignment-handbook
 - trl
 - dpo
-- alignment-handbook
 - generated_from_trainer
 base_model: NbAiLab/nb-gpt-j-6B-v2
+datasets:
+- hugodk-sch/aftonposten_title_prefs
 model-index:
 - name: aftonposten-6b-align-scan
   results: []
@@ -16,17 +18,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # aftonposten-6b-align-scan
 
-This model is a fine-tuned version of [NbAiLab/nb-gpt-j-6B-v2](https://huggingface.co/NbAiLab/nb-gpt-j-6B-v2) on an unknown dataset.
+This model is a fine-tuned version of [data/ap-gpt-j-6b-sft-qlora-04-08](https://huggingface.co/data/ap-gpt-j-6b-sft-qlora-04-08) on the hugodk-sch/aftonposten_title_prefs dataset.
 It achieves the following results on the evaluation set:
-- Logits/chosen: -2.2436
-- Logits/rejected: -2.2387
-- Logps/chosen: -34.0238
-- Logps/rejected: -37.5004
-- Loss: 0.4345
-- Rewards/accuracies: 0.4809
-- Rewards/chosen: 0.0086
-- Rewards/margins: -0.0044
-- Rewards/rejected: 0.0130
+- Loss: 0.4422
+- Rewards/chosen: 0.0026
+- Rewards/rejected: 0.0097
+- Rewards/accuracies: 0.4568
+- Rewards/margins: -0.0071
+- Logps/rejected: -37.5045
+- Logps/chosen: -34.0313
+- Logits/rejected: -2.2388
+- Logits/chosen: -2.2437
 
 ## Model description
 

all_results.json

@@ -9,9 +9,9 @@
     "eval_rewards/chosen": 0.002567538060247898,
     "eval_rewards/margins": -0.0071130492724478245,
     "eval_rewards/rejected": 0.009680588729679585,
-    "eval_runtime": 145.7928,
+    "eval_runtime": 145.9043,
     "eval_samples": 343,
-    "eval_samples_per_second": 2.353,
+    "eval_samples_per_second": 2.351,
     "eval_steps_per_second": 0.295,
     "train_loss": 0.08930327675559303,
     "train_runtime": 627.8357,

eval_results.json

@@ -9,8 +9,8 @@
     "eval_rewards/chosen": 0.002567538060247898,
     "eval_rewards/margins": -0.0071130492724478245,
     "eval_rewards/rejected": 0.009680588729679585,
-    "eval_runtime": 145.7928,
+    "eval_runtime": 145.9043,
     "eval_samples": 343,
-    "eval_samples_per_second": 2.353,
+    "eval_samples_per_second": 2.351,
     "eval_steps_per_second": 0.295
 }