End of training

README.md

@@ -3,10 +3,16 @@ library_name: transformers
 license: apache-2.0
 base_model: alignment-handbook/zephyr-7b-sft-full
 tags:
+- alignment-handbook
+- trl
+- dpo
+- generated_from_trainer
 - trl
 - dpo
 - alignment-handbook
 - generated_from_trainer
+datasets:
+- data/zephyr_uf_rlced_conifer_ref
 model-index:
 - name: uf-rlced-conifer-zephyr-7b-group-dpo-no-clip
   results: []
@@ -17,7 +23,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # uf-rlced-conifer-zephyr-7b-group-dpo-no-clip
 
-This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
+This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the data/zephyr_uf_rlced_conifer_ref dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.2786
+- Rewards/chosen: -1.9498
+- Rewards/rejected: -4.7561
+- Rewards/accuracies: 0.8806
+- Rewards/margins: 2.8063
+- Logps/rejected: -885.7299
+- Logps/chosen: -579.5609
+- Logits/rejected: 0.9913
+- Logits/chosen: -0.2160
 
 ## Model description
 

all_results.json

@@ -1,18 +1,18 @@
 {
     "epoch": 0.9993060374739764,
-    "eval_logits/chosen": -2.7333760261535645,
-    "eval_logits/rejected": -2.687427043914795,
-    "eval_logps/chosen": -425.93402099609375,
-    "eval_logps/rejected": -314.981201171875,
-    "eval_loss": 0.6931471824645996,
-    "eval_rewards/accuracies": 0.0,
-    "eval_rewards/chosen": 0.0,
-    "eval_rewards/margins": 0.0,
-    "eval_rewards/rejected": 0.0,
-    "eval_runtime": 6.0564,
-    "eval_samples": 100,
-    "eval_samples_per_second": 16.512,
-    "eval_steps_per_second": 0.33,
+    "eval_logits/chosen": -0.21596238017082214,
+    "eval_logits/rejected": 0.9912689924240112,
+    "eval_logps/chosen": -579.5609130859375,
+    "eval_logps/rejected": -885.7299194335938,
+    "eval_loss": 0.2786034047603607,
+    "eval_rewards/accuracies": 0.8806390762329102,
+    "eval_rewards/chosen": -1.94976806640625,
+    "eval_rewards/margins": 2.806307792663574,
+    "eval_rewards/rejected": -4.756075859069824,
+    "eval_runtime": 387.8423,
+    "eval_samples": 8491,
+    "eval_samples_per_second": 21.893,
+    "eval_steps_per_second": 0.343,
     "total_flos": 0.0,
     "train_loss": 0.1315754380491045,
     "train_runtime": 20812.0317,

config.json

@@ -22,6 +22,6 @@
   "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": -2.7333760261535645,
-    "eval_logits/rejected": -2.687427043914795,
-    "eval_logps/chosen": -425.93402099609375,
-    "eval_logps/rejected": -314.981201171875,
-    "eval_loss": 0.6931471824645996,
-    "eval_rewards/accuracies": 0.0,
-    "eval_rewards/chosen": 0.0,
-    "eval_rewards/margins": 0.0,
-    "eval_rewards/rejected": 0.0,
-    "eval_runtime": 6.0564,
-    "eval_samples": 100,
-    "eval_samples_per_second": 16.512,
-    "eval_steps_per_second": 0.33
+    "epoch": 0.9993060374739764,
+    "eval_logits/chosen": -0.21596238017082214,
+    "eval_logits/rejected": 0.9912689924240112,
+    "eval_logps/chosen": -579.5609130859375,
+    "eval_logps/rejected": -885.7299194335938,
+    "eval_loss": 0.2786034047603607,
+    "eval_rewards/accuracies": 0.8806390762329102,
+    "eval_rewards/chosen": -1.94976806640625,
+    "eval_rewards/margins": 2.806307792663574,
+    "eval_rewards/rejected": -4.756075859069824,
+    "eval_runtime": 387.8423,
+    "eval_samples": 8491,
+    "eval_samples_per_second": 21.893,
+    "eval_steps_per_second": 0.343
 }