End of training

README.md

@@ -3,9 +3,15 @@ library_name: transformers
 license: other
 base_model: trl-lib/qwen1.5-0.5b-sft
 tags:
+- alignment-handbook
 - trl
 - simpo
 - generated_from_trainer
+- trl
+- simpo
+- generated_from_trainer
+datasets:
+- yakazimir/ultrafeedback_binarized
 model-index:
 - name: qwen_l21_entropy_0_01
   results: []
@@ -16,15 +22,15 @@ should probably proofread and complete it, then remove this comment. -->
 
 # qwen_l21_entropy_0_01
 
-This model is a fine-tuned version of [trl-lib/qwen1.5-0.5b-sft](https://huggingface.co/trl-lib/qwen1.5-0.5b-sft) on an unknown dataset.
+This model is a fine-tuned version of [trl-lib/qwen1.5-0.5b-sft](https://huggingface.co/trl-lib/qwen1.5-0.5b-sft) on the yakazimir/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.6901
 - Sft Loss: 2.1331
 - Rewards/chosen: -2.1707
-- Rewards/rejected: -3.2271
+- Rewards/rejected: -3.2270
 - Rewards/accuracies: 0.6914
 - Rewards/margins: 1.0563
-- Logps/rejected: -3.2271
+- Logps/rejected: -3.2270
 - Logps/chosen: -2.1707
 - Logits/rejected: 0.2151
 - Logits/chosen: 0.1185

all_results.json

@@ -1,5 +1,19 @@
 {
     "epoch": 2.999297541394882,
+    "eval_logits/chosen": 0.11851656436920166,
+    "eval_logits/rejected": 0.21514640748500824,
+    "eval_logps/chosen": -2.170732259750366,
+    "eval_logps/rejected": -3.2270374298095703,
+    "eval_loss": 0.690127432346344,
+    "eval_rewards/accuracies": 0.6913946866989136,
+    "eval_rewards/chosen": -2.170732259750366,
+    "eval_rewards/margins": 1.0563050508499146,
+    "eval_rewards/rejected": -3.2270374298095703,
+    "eval_runtime": 43.2304,
+    "eval_samples": 1345,
+    "eval_samples_per_second": 31.112,
+    "eval_sft_loss": 2.133072853088379,
+    "eval_steps_per_second": 7.795,
     "total_flos": 0.0,
     "train_loss": 0.6914210148065283,
     "train_runtime": 33926.4655,

config.json

@@ -22,7 +22,7 @@
   "tie_word_embeddings": true,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.44.2",
-  "use_cache": false,
+  "use_cache": true,
   "use_sliding_window": false,
   "vocab_size": 151648
 }

eval_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 2.999297541394882,
+    "eval_logits/chosen": 0.11851656436920166,
+    "eval_logits/rejected": 0.21514640748500824,
+    "eval_logps/chosen": -2.170732259750366,
+    "eval_logps/rejected": -3.2270374298095703,
+    "eval_loss": 0.690127432346344,
+    "eval_rewards/accuracies": 0.6913946866989136,
+    "eval_rewards/chosen": -2.170732259750366,
+    "eval_rewards/margins": 1.0563050508499146,
+    "eval_rewards/rejected": -3.2270374298095703,
+    "eval_runtime": 43.2304,
+    "eval_samples": 1345,
+    "eval_samples_per_second": 31.112,
+    "eval_sft_loss": 2.133072853088379,
+    "eval_steps_per_second": 7.795
+}