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-no-excess
   results: []
@@ -17,7 +23,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # uf-rlced-conifer-zephyr-7b-group-dpo-no-clip-no-excess
 
-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: 0.3975
+- Rewards/chosen: -1.2128
+- Rewards/rejected: -2.4224
+- Rewards/accuracies: 0.8130
+- Rewards/margins: 1.2096
+- Logps/rejected: -652.3658
+- Logps/chosen: -505.8686
+- Logits/rejected: 0.3206
+- Logits/chosen: -0.1209
 
 ## 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": 5.9813,
-    "eval_samples": 100,
-    "eval_samples_per_second": 16.719,
-    "eval_steps_per_second": 0.334,
+    "eval_logits/chosen": -0.12094812095165253,
+    "eval_logits/rejected": 0.3205632269382477,
+    "eval_logps/chosen": -505.86859130859375,
+    "eval_logps/rejected": -652.3658447265625,
+    "eval_loss": 0.3974800109863281,
+    "eval_rewards/accuracies": 0.8129699230194092,
+    "eval_rewards/chosen": -1.2128452062606812,
+    "eval_rewards/margins": 1.2095903158187866,
+    "eval_rewards/rejected": -2.4224352836608887,
+    "eval_runtime": 388.0524,
+    "eval_samples": 8491,
+    "eval_samples_per_second": 21.881,
+    "eval_steps_per_second": 0.343,
     "total_flos": 0.0,
     "train_loss": 0.5182241236997975,
     "train_runtime": 20862.0677,

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": 5.9813,
-    "eval_samples": 100,
-    "eval_samples_per_second": 16.719,
-    "eval_steps_per_second": 0.334
+    "epoch": 0.9993060374739764,
+    "eval_logits/chosen": -0.12094812095165253,
+    "eval_logits/rejected": 0.3205632269382477,
+    "eval_logps/chosen": -505.86859130859375,
+    "eval_logps/rejected": -652.3658447265625,
+    "eval_loss": 0.3974800109863281,
+    "eval_rewards/accuracies": 0.8129699230194092,
+    "eval_rewards/chosen": -1.2128452062606812,
+    "eval_rewards/margins": 1.2095903158187866,
+    "eval_rewards/rejected": -2.4224352836608887,
+    "eval_runtime": 388.0524,
+    "eval_samples": 8491,
+    "eval_samples_per_second": 21.881,
+    "eval_steps_per_second": 0.343
 }