End of training

README.md

@@ -2,9 +2,15 @@
 license: apache-2.0
 base_model: mistralai/Mistral-7B-Instruct-v0.2
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- Dahoas/full-hh-rlhf
 model-index:
 - name: Mistral-7B-Instruct-v0.2-DPO
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # Mistral-7B-Instruct-v0.2-DPO
 
-This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the None dataset.
+This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the Dahoas/full-hh-rlhf dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.5782
-- Rewards/chosen: -0.2135
-- Rewards/rejected: -0.7018
-- Rewards/accuracies: 0.6920
-- Rewards/margins: 0.4884
-- Logps/rejected: -296.4236
-- Logps/chosen: -255.7232
-- Logits/rejected: -2.4987
-- Logits/chosen: -2.5475
+- Rewards/chosen: -0.2120
+- Rewards/rejected: -0.7002
+- Rewards/accuracies: 0.6926
+- Rewards/margins: 0.4883
+- Logps/rejected: -296.2612
+- Logps/chosen: -255.5737
+- Logits/rejected: -2.4985
+- Logits/chosen: -2.5472
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -2.5471837520599365,
+    "eval_logits/rejected": -2.4984941482543945,
+    "eval_logps/chosen": -255.5736846923828,
+    "eval_logps/rejected": -296.26116943359375,
+    "eval_loss": 0.5782328248023987,
+    "eval_rewards/accuracies": 0.6926282048225403,
+    "eval_rewards/chosen": -0.21196967363357544,
+    "eval_rewards/margins": 0.4882549047470093,
+    "eval_rewards/rejected": -0.7002245783805847,
+    "eval_runtime": 862.1588,
+    "eval_samples": 12451,
+    "eval_samples_per_second": 14.442,
+    "eval_steps_per_second": 0.452,
     "train_loss": 0.5964619764459262,
     "train_runtime": 39064.845,
     "train_samples": 112052,

config.json

@@ -21,6 +21,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.39.0.dev0",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.0,
+    "eval_logits/chosen": -2.5471837520599365,
+    "eval_logits/rejected": -2.4984941482543945,
+    "eval_logps/chosen": -255.5736846923828,
+    "eval_logps/rejected": -296.26116943359375,
+    "eval_loss": 0.5782328248023987,
+    "eval_rewards/accuracies": 0.6926282048225403,
+    "eval_rewards/chosen": -0.21196967363357544,
+    "eval_rewards/margins": 0.4882549047470093,
+    "eval_rewards/rejected": -0.7002245783805847,
+    "eval_runtime": 862.1588,
+    "eval_samples": 12451,
+    "eval_samples_per_second": 14.442,
+    "eval_steps_per_second": 0.452
+}

runs/Apr29_16-59-51_gilbreth-j001.rcac.purdue.edu/events.out.tfevents.1714465224.gilbreth-j001.rcac.purdue.edu.209964.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c1d06a4cc29248fb17569fe28a08ec12198e507c887ba0e7216c69a2ec55ee5
+size 828