End of training

README.md

@@ -1,9 +1,15 @@
 ---
 base_model: argsearch/llama-7b-sft-float32
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- Dahoas/full-hh-rlhf
 model-index:
 - name: llama-7b-sft-DPO
   results: []
@@ -14,17 +20,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama-7b-sft-DPO
 
-This model is a fine-tuned version of [argsearch/llama-7b-sft-float32](https://huggingface.co/argsearch/llama-7b-sft-float32) on the None dataset.
+This model is a fine-tuned version of [argsearch/llama-7b-sft-float32](https://huggingface.co/argsearch/llama-7b-sft-float32) on the Dahoas/full-hh-rlhf dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.6525
-- Rewards/chosen: 0.3312
-- Rewards/rejected: 0.1950
-- Rewards/accuracies: 0.6074
-- Rewards/margins: 0.1363
-- Logps/rejected: -633.4115
-- Logps/chosen: -690.5892
-- Logits/rejected: -1.9098
-- Logits/chosen: -1.9645
+- Rewards/chosen: 0.3315
+- Rewards/rejected: 0.1953
+- Rewards/accuracies: 0.6080
+- Rewards/margins: 0.1362
+- Logps/rejected: -633.3815
+- Logps/chosen: -690.5654
+- Logits/rejected: -1.9212
+- Logits/chosen: -1.9766
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -1.97661554813385,
+    "eval_logits/rejected": -1.9212419986724854,
+    "eval_logps/chosen": -690.5653686523438,
+    "eval_logps/rejected": -633.3815307617188,
+    "eval_loss": 0.6524845957756042,
+    "eval_rewards/accuracies": 0.6080127954483032,
+    "eval_rewards/chosen": 0.33147066831588745,
+    "eval_rewards/margins": 0.13620509207248688,
+    "eval_rewards/rejected": 0.19526559114456177,
+    "eval_runtime": 812.3144,
+    "eval_samples": 12451,
+    "eval_samples_per_second": 15.328,
+    "eval_steps_per_second": 0.48,
     "train_loss": 0.6642669952984608,
     "train_runtime": 38056.9659,
     "train_samples": 112052,

config.json

@@ -25,6 +25,6 @@
   "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": -1.97661554813385,
+    "eval_logits/rejected": -1.9212419986724854,
+    "eval_logps/chosen": -690.5653686523438,
+    "eval_logps/rejected": -633.3815307617188,
+    "eval_loss": 0.6524845957756042,
+    "eval_rewards/accuracies": 0.6080127954483032,
+    "eval_rewards/chosen": 0.33147066831588745,
+    "eval_rewards/margins": 0.13620509207248688,
+    "eval_rewards/rejected": 0.19526559114456177,
+    "eval_runtime": 812.3144,
+    "eval_samples": 12451,
+    "eval_samples_per_second": 15.328,
+    "eval_steps_per_second": 0.48
+}

runs/Apr30_13-22-22_gilbreth-j001.rcac.purdue.edu/events.out.tfevents.1714537921.gilbreth-j001.rcac.purdue.edu.163553.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e82a4de46062d6569253e1aa1860ad7cf161bb1933cc15ea2a36262d673b4ee
+size 828