End of training

README.md

@@ -2,9 +2,15 @@
 license: other
 base_model: HuggingFaceH4/zephyr-7b-gemma-sft-v0.1
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- argilla/dpo-mix-7k
 model-index:
 - name: zephyr-7b-gemma-dpo
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-gemma-dpo
 
-This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-gemma-sft-v0.1](https://huggingface.co/HuggingFaceH4/zephyr-7b-gemma-sft-v0.1) on the None dataset.
+This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-gemma-sft-v0.1](https://huggingface.co/HuggingFaceH4/zephyr-7b-gemma-sft-v0.1) on the argilla/dpo-mix-7k dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4705
-- Rewards/chosen: -2.8703
-- Rewards/rejected: -4.4285
-- Rewards/accuracies: 0.7188
-- Rewards/margins: 1.5583
-- Logps/rejected: -448.1896
-- Logps/chosen: -419.4271
-- Logits/rejected: 88.1427
-- Logits/chosen: 94.0592
+- Loss: 0.4682
+- Rewards/chosen: -2.8383
+- Rewards/rejected: -4.4417
+- Rewards/accuracies: 0.7396
+- Rewards/margins: 1.6034
+- Logps/rejected: -448.4524
+- Logps/chosen: -418.7875
+- Logits/rejected: 88.1175
+- Logits/chosen: 94.0612
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.97,
+    "eval_logits/chosen": 94.06119537353516,
+    "eval_logits/rejected": 88.1175308227539,
+    "eval_logps/chosen": -418.7874755859375,
+    "eval_logps/rejected": -448.4524230957031,
+    "eval_loss": 0.46823105216026306,
+    "eval_rewards/accuracies": 0.7395833134651184,
+    "eval_rewards/chosen": -2.838282823562622,
+    "eval_rewards/margins": 1.6033902168273926,
+    "eval_rewards/rejected": -4.4416728019714355,
+    "eval_runtime": 57.2277,
+    "eval_samples": 750,
+    "eval_samples_per_second": 13.106,
+    "eval_steps_per_second": 0.419,
     "train_loss": 0.3924295770434233,
     "train_runtime": 3265.4264,
     "train_samples": 6750,

config.json

@@ -23,6 +23,6 @@
   "rope_theta": 10000.0,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.39.0.dev0",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 256000
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.97,
+    "eval_logits/chosen": 94.06119537353516,
+    "eval_logits/rejected": 88.1175308227539,
+    "eval_logps/chosen": -418.7874755859375,
+    "eval_logps/rejected": -448.4524230957031,
+    "eval_loss": 0.46823105216026306,
+    "eval_rewards/accuracies": 0.7395833134651184,
+    "eval_rewards/chosen": -2.838282823562622,
+    "eval_rewards/margins": 1.6033902168273926,
+    "eval_rewards/rejected": -4.4416728019714355,
+    "eval_runtime": 57.2277,
+    "eval_samples": 750,
+    "eval_samples_per_second": 13.106,
+    "eval_steps_per_second": 0.419
+}

runs/Mar28_19-45-13_batch-block1-0100/events.out.tfevents.1711684610.batch-block1-0100.2604988.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb93784ce00b098f78fef7f4a0e80a87baf31c59631ea43262843dad636595c7
+size 815