End of training

README.md

@@ -3,9 +3,15 @@ library_name: transformers
 license: apache-2.0
 base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- data/ui_math
 model-index:
 - name: tinyllama-1.1b-chat-v1.0-ui-math-dpo-2
   results: []
@@ -16,7 +22,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # tinyllama-1.1b-chat-v1.0-ui-math-dpo-2
 
-This model is a fine-tuned version of [TinyLlama/TinyLlama-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0) on an unknown dataset.
+This model is a fine-tuned version of [TinyLlama/TinyLlama-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0) on the data/ui_math dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.6219
+- Rewards/chosen: -3.3170
+- Rewards/rejected: -4.6926
+- Rewards/accuracies: 0.6875
+- Rewards/margins: 1.3757
+- Logps/rejected: -721.5773
+- Logps/chosen: -610.5119
+- Logits/rejected: -0.7616
+- Logits/chosen: -0.8420
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.9977349943374858,
+    "eval_logits/chosen": -0.8419910669326782,
+    "eval_logits/rejected": -0.7616477012634277,
+    "eval_logps/chosen": -610.5119018554688,
+    "eval_logps/rejected": -721.5773315429688,
+    "eval_loss": 0.6218873858451843,
+    "eval_rewards/accuracies": 0.6875,
+    "eval_rewards/chosen": -3.3169686794281006,
+    "eval_rewards/margins": 1.3756744861602783,
+    "eval_rewards/rejected": -4.692643165588379,
+    "eval_runtime": 1.4063,
+    "eval_samples": 100,
+    "eval_samples_per_second": 71.106,
+    "eval_steps_per_second": 1.422,
     "total_flos": 0.0,
     "train_loss": 0.32467431276022984,
     "train_runtime": 5857.252,

config.json

@@ -24,6 +24,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.44.1",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.9977349943374858,
+    "eval_logits/chosen": -0.8419910669326782,
+    "eval_logits/rejected": -0.7616477012634277,
+    "eval_logps/chosen": -610.5119018554688,
+    "eval_logps/rejected": -721.5773315429688,
+    "eval_loss": 0.6218873858451843,
+    "eval_rewards/accuracies": 0.6875,
+    "eval_rewards/chosen": -3.3169686794281006,
+    "eval_rewards/margins": 1.3756744861602783,
+    "eval_rewards/rejected": -4.692643165588379,
+    "eval_runtime": 1.4063,
+    "eval_samples": 100,
+    "eval_samples_per_second": 71.106,
+    "eval_steps_per_second": 1.422
+}