Model save

README.md

@@ -1,7 +1,7 @@
 ---
-base_model: meta-llama/Llama-3.2-3B-Instruct
+base_model: RyanYr/self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter3
 library_name: transformers
-model_name: self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter1
+model_name: self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter4
 tags:
 - generated_from_trainer
 - trl
@@ -9,9 +9,9 @@ tags:
 licence: license
 ---
 
-# Model Card for self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter1
+# Model Card for self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter4
 
-This model is a fine-tuned version of [meta-llama/Llama-3.2-3B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct).
+This model is a fine-tuned version of [RyanYr/self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter3](https://huggingface.co/RyanYr/self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter3).
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start
@@ -20,14 +20,14 @@ It has been trained using [TRL](https://github.com/huggingface/trl).
 from transformers import pipeline
 
 question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
-generator = pipeline("text-generation", model="RyanYr/self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter1", device="cuda")
+generator = pipeline("text-generation", model="RyanYr/self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter4", device="cuda")
 output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
 print(output["generated_text"])
 ```
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/yyr/huggingface/runs/x8n9cjbe)
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/yyr/huggingface/runs/vvmc4ls0)
 
 This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
 

last_checkpoint/config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "meta-llama/Llama-3.2-3B-Instruct",
+  "_name_or_path": "RyanYr/self-correct_Llama-3.2-3B-Instruct_metaMathQA_dpo_iter3",
   "architectures": [
     "LlamaForCausalLM"
   ],

last_checkpoint/model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fbb8867b4e2ca72e9fdffb399559526440aa6eb0dcbc022ad5a6d1c2e47d64b9
+oid sha256:8092f1a017769819bdf32e602d3d4746bcb91200a082f9e2cac39bc859c310bc
 size 4965805240

last_checkpoint/model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d05836db16437fae5f7537a00295bab1948f8c67ea2617f24efde451dd799314
+oid sha256:0071cb39f5515e4f58491354535bc4060eae31da836586c2c4b772e236997441
 size 2247741136

last_checkpoint/tokenizer_config.json

@@ -2065,7 +2065,7 @@
     "input_ids",
     "attention_mask"
   ],
-  "model_max_length": 2048,
+  "model_max_length": 4096,
   "pad_token": "[PAD]",
   "tokenizer_class": "PreTrainedTokenizerFast"
 }

last_checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c85fb0f99bdb174f76af81b90416c4b5fe227dfe7a5893f5335a3d11e9e7534f
+oid sha256:401022172dd61b8f7395e63bb1a93146a023ade30a8e1d61e880ef73ae18ca19
 size 7608