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 the None 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:
-- Loss: 0.5566
-- Rewards/chosen: 0.0778
-- Rewards/rejected: 0.0602
-- Rewards/accuracies: 0.5050
-- Rewards/margins: 0.0176
-- Logps/rejected: -37.4306
-- Logps/chosen: -33.9234
-- Logits/rejected: -2.2423
-- Logits/chosen: -2.2472
+- Loss: 0.5772
+- Rewards/chosen: 0.0684
+- Rewards/rejected: 0.0623
+- Rewards/accuracies: 0.5307
+- Rewards/margins: 0.0061
+- Logps/rejected: -37.4276
+- Logps/chosen: -33.9368
+- Logits/rejected: -2.2420
+- Logits/chosen: -2.2469
 
 ## Model description
 

all_results.json

@@ -1,18 +1,18 @@
 {
     "epoch": 4.0,
-    "eval_logits/chosen": -2.236751079559326,
-    "eval_logits/rejected": -2.2318832874298096,
-    "eval_logps/chosen": -34.04561233520508,
-    "eval_logps/rejected": -37.546016693115234,
-    "eval_loss": 0.5836836099624634,
-    "eval_rewards/accuracies": 0.5427741408348083,
-    "eval_rewards/chosen": -0.007742464076727629,
-    "eval_rewards/margins": 0.01283606793731451,
-    "eval_rewards/rejected": -0.02057853527367115,
-    "eval_runtime": 145.5744,
+    "eval_logits/chosen": -2.2468910217285156,
+    "eval_logits/rejected": -2.24200439453125,
+    "eval_logps/chosen": -33.9367790222168,
+    "eval_logps/rejected": -37.42759704589844,
+    "eval_loss": 0.577243447303772,
+    "eval_rewards/accuracies": 0.5307309031486511,
+    "eval_rewards/chosen": 0.06843975186347961,
+    "eval_rewards/margins": 0.0061217062175273895,
+    "eval_rewards/rejected": 0.06231804937124252,
+    "eval_runtime": 145.2917,
     "eval_samples": 343,
-    "eval_samples_per_second": 2.356,
-    "eval_steps_per_second": 0.295,
+    "eval_samples_per_second": 2.361,
+    "eval_steps_per_second": 0.296,
     "train_loss": 0.20226844080856868,
     "train_runtime": 10768.956,
     "train_samples": 3079,

config.json

@@ -18,6 +18,8 @@
   "n_layer": 28,
   "n_positions": 2048,
   "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
     "bnb_4bit_compute_dtype": "float16",
     "bnb_4bit_quant_type": "nf4",
     "bnb_4bit_use_double_quant": false,
@@ -45,7 +47,7 @@
   },
   "tie_word_embeddings": false,
   "tokenizer_class": "GPT2Tokenizer",
-  "transformers_version": "4.37.2",
+  "transformers_version": "4.39.0.dev0",
   "use_cache": true,
   "vocab_size": 50400
 }

eval_results.json

@@ -1,16 +1,16 @@
 {
-    "epoch": 1.0,
-    "eval_logits/chosen": -2.236751079559326,
-    "eval_logits/rejected": -2.2318832874298096,
-    "eval_logps/chosen": -34.04561233520508,
-    "eval_logps/rejected": -37.546016693115234,
-    "eval_loss": 0.5836836099624634,
-    "eval_rewards/accuracies": 0.5427741408348083,
-    "eval_rewards/chosen": -0.007742464076727629,
-    "eval_rewards/margins": 0.01283606793731451,
-    "eval_rewards/rejected": -0.02057853527367115,
-    "eval_runtime": 145.5744,
+    "epoch": 4.0,
+    "eval_logits/chosen": -2.2468910217285156,
+    "eval_logits/rejected": -2.24200439453125,
+    "eval_logps/chosen": -33.9367790222168,
+    "eval_logps/rejected": -37.42759704589844,
+    "eval_loss": 0.577243447303772,
+    "eval_rewards/accuracies": 0.5307309031486511,
+    "eval_rewards/chosen": 0.06843975186347961,
+    "eval_rewards/margins": 0.0061217062175273895,
+    "eval_rewards/rejected": 0.06231804937124252,
+    "eval_runtime": 145.2917,
     "eval_samples": 343,
-    "eval_samples_per_second": 2.356,
-    "eval_steps_per_second": 0.295
+    "eval_samples_per_second": 2.361,
+    "eval_steps_per_second": 0.296
 }