End of training

README.md

@@ -2,9 +2,15 @@
 license: gemma
 base_model: google/gemma-7b
 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: []
@@ -16,17 +22,17 @@ should probably proofread and complete it, then remove this comment. -->
 [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://zebra.wandb.io/cto/distillm/runs/n5v6nn5w)
 # zephyr-7b-gemma-dpo
 
-This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b) on an unknown dataset.
+This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b) on the argilla/dpo-mix-7k dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.8002
-- Rewards/chosen: -0.4660
-- Rewards/rejected: -1.3128
-- Rewards/accuracies: 0.7604
-- Rewards/margins: 0.8468
-- Logps/rejected: -1648.5675
-- Logps/chosen: -1530.4515
-- Logits/rejected: -25.1625
-- Logits/chosen: -18.0007
+- Loss: 0.8036
+- Rewards/chosen: -0.4463
+- Rewards/rejected: -1.2861
+- Rewards/accuracies: 0.7292
+- Rewards/margins: 0.8397
+- Logps/rejected: -1648.0323
+- Logps/chosen: -1530.0571
+- Logits/rejected: -25.1620
+- Logits/chosen: -18.0449
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.971563981042654,
+    "eval_logits/chosen": -18.044893264770508,
+    "eval_logits/rejected": -25.161956787109375,
+    "eval_logps/chosen": -1530.05712890625,
+    "eval_logps/rejected": -1648.0323486328125,
+    "eval_loss": 0.8036046624183655,
+    "eval_rewards/accuracies": 0.7291666865348816,
+    "eval_rewards/chosen": -0.44631412625312805,
+    "eval_rewards/margins": 0.8397437930107117,
+    "eval_rewards/rejected": -1.2860580682754517,
+    "eval_runtime": 37.4541,
+    "eval_samples": 750,
+    "eval_samples_per_second": 20.025,
+    "eval_steps_per_second": 0.641,
     "total_flos": 0.0,
     "train_loss": 0.6178501087885636,
     "train_runtime": 1142.0913,

config.json

@@ -24,6 +24,6 @@
   "rope_theta": 10000.0,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.42.3",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 256000
 }

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.971563981042654,
-    "eval_logits/chosen": 85.37242889404297,
-    "eval_logits/rejected": 79.6775131225586,
-    "eval_logps/chosen": -464.193603515625,
-    "eval_logps/rejected": -498.6763916015625,
-    "eval_loss": 0.4798102080821991,
-    "eval_rewards/accuracies": 0.7083333134651184,
-    "eval_rewards/chosen": -5.0729851722717285,
-    "eval_rewards/margins": 1.7632160186767578,
-    "eval_rewards/rejected": -6.836201190948486,
-    "eval_runtime": 36.7059,
+    "eval_logits/chosen": -18.044893264770508,
+    "eval_logits/rejected": -25.161956787109375,
+    "eval_logps/chosen": -1530.05712890625,
+    "eval_logps/rejected": -1648.0323486328125,
+    "eval_loss": 0.8036046624183655,
+    "eval_rewards/accuracies": 0.7291666865348816,
+    "eval_rewards/chosen": -0.44631412625312805,
+    "eval_rewards/margins": 0.8397437930107117,
+    "eval_rewards/rejected": -1.2860580682754517,
+    "eval_runtime": 37.4541,
     "eval_samples": 750,
-    "eval_samples_per_second": 20.433,
-    "eval_steps_per_second": 0.654
+    "eval_samples_per_second": 20.025,
+    "eval_steps_per_second": 0.641
 }

runs/Jul11_11-42-48_ale-distillm-8-0-0/events.out.tfevents.1720696575.ale-distillm-8-0-0.6459.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0df482984ff3b81a62a5d3c8c5d14de6b66fbe2c0d4eaecead3c1a40370c24a9
+size 815