End of training

README.md

@@ -3,10 +3,17 @@ 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
 - alignment-handbook
 - generated_from_trainer
+datasets:
+- data/ui_math_ref
+- data/ui_coding_ref
 model-index:
 - name: tinyllama-1.1b-chat-v1.0-ui-math-coding-group-dpo
   results: []
@@ -17,7 +24,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # tinyllama-1.1b-chat-v1.0-ui-math-coding-group-dpo
 
-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 None 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_ref and the data/ui_coding_ref datasets.
+It achieves the following results on the evaluation set:
+- Loss: 0.4193
+- Rewards/chosen: -0.6057
+- Rewards/rejected: -1.6972
+- Rewards/accuracies: 0.7188
+- Rewards/margins: 1.0915
+- Logps/rejected: -538.1440
+- Logps/chosen: -448.5587
+- Logits/rejected: -2.6494
+- Logits/chosen: -2.6509
 
 ## Model description
 

all_results.json

@@ -1,18 +1,18 @@
 {
     "epoch": 0.9963369963369964,
-    "eval_logits/chosen": -2.156226873397827,
-    "eval_logits/rejected": -2.1579225063323975,
-    "eval_logps/chosen": -308.5898132324219,
-    "eval_logps/rejected": -291.931640625,
-    "eval_loss": 0.6931471228599548,
-    "eval_rewards/accuracies": 0.0,
-    "eval_rewards/chosen": 0.0,
-    "eval_rewards/margins": 0.0,
-    "eval_rewards/rejected": 0.0,
-    "eval_runtime": 1.5116,
-    "eval_samples": 100,
-    "eval_samples_per_second": 66.156,
-    "eval_steps_per_second": 0.662,
+    "eval_logits/chosen": -2.650862693786621,
+    "eval_logits/rejected": -2.649400234222412,
+    "eval_logps/chosen": -448.55865478515625,
+    "eval_logps/rejected": -538.1439819335938,
+    "eval_loss": 0.41933417320251465,
+    "eval_rewards/accuracies": 0.71875,
+    "eval_rewards/chosen": -0.6056715250015259,
+    "eval_rewards/margins": 1.091530680656433,
+    "eval_rewards/rejected": -1.697202205657959,
+    "eval_runtime": 2.9829,
+    "eval_samples": 200,
+    "eval_samples_per_second": 67.048,
+    "eval_steps_per_second": 0.67,
     "total_flos": 0.0,
     "train_loss": 0.09771968247186319,
     "train_runtime": 5407.8058,

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

@@ -1,16 +1,16 @@
 {
-    "epoch": 1.0,
-    "eval_logits/chosen": -2.156226873397827,
-    "eval_logits/rejected": -2.1579225063323975,
-    "eval_logps/chosen": -308.5898132324219,
-    "eval_logps/rejected": -291.931640625,
-    "eval_loss": 0.6931471228599548,
-    "eval_rewards/accuracies": 0.0,
-    "eval_rewards/chosen": 0.0,
-    "eval_rewards/margins": 0.0,
-    "eval_rewards/rejected": 0.0,
-    "eval_runtime": 1.5116,
-    "eval_samples": 100,
-    "eval_samples_per_second": 66.156,
-    "eval_steps_per_second": 0.662
+    "epoch": 0.9963369963369964,
+    "eval_logits/chosen": -2.650862693786621,
+    "eval_logits/rejected": -2.649400234222412,
+    "eval_logps/chosen": -448.55865478515625,
+    "eval_logps/rejected": -538.1439819335938,
+    "eval_loss": 0.41933417320251465,
+    "eval_rewards/accuracies": 0.71875,
+    "eval_rewards/chosen": -0.6056715250015259,
+    "eval_rewards/margins": 1.091530680656433,
+    "eval_rewards/rejected": -1.697202205657959,
+    "eval_runtime": 2.9829,
+    "eval_samples": 200,
+    "eval_samples_per_second": 67.048,
+    "eval_steps_per_second": 0.67
 }