Update README.md

README.md

@@ -160,7 +160,7 @@ Finally, we can train the model using the ORPOTrainer, which acts as a wrapper.
 
 
 ```python
-
+# I saved the dataset in my local directory! but you may not
 dataset_name = "/data/bio-eng-llm/llm_repo/mlabonne/OrpoLlama-3-8B"
 
 dataset = load_dataset(dataset_name, split="all")
@@ -242,13 +242,8 @@ Training the model on these 1,000 samples and 20 epochs took about 22 hours on a
 
 ## Test the model
 
-# -*- coding: utf-8 -*-
-"""
-Created on Wed Jul  3 15:57:22 2024
-
-@author: Ali forootani
-"""
 
+### Required packages
 ```bash
 pip install -U transformers datasets accelerate peft trl bitsandbytes wandb
 pip install -qqq flash-attn
@@ -290,15 +285,6 @@ from trl import ORPOConfig, ORPOTrainer, setup_chat_format
 
 
 
-
-"""
-https://huggingface.co/blog/mlabonne/orpo-llama-3
-
-mlabonne/orpo-dpo-mix-40k
-
-https://huggingface.co/datasets/mlabonne/orpo-dpo-mix-40k/tree/main
-"""
-
 if torch.cuda.get_device_capability()[0] >= 128:
     
     attn_implementation = "flash_attention_2"
@@ -326,7 +312,7 @@ def setting_directory(depth):
         sys.path.append(os.path.dirname(root_dir))
     return root_dir
 
-
+# I loaded the base model form local directory but you may load it directy from huggingface
 model_path = "/data/bio-eng-llm/llm_repo/mlabonne/OrpoLlama-3-8B"
 
 
@@ -375,6 +361,7 @@ model, tokenizer = setup_chat_format(model, tokenizer)
 root_dir = setting_directory(0)
 epochs = 20
 
+# I loaded the fine tuned model from my local directory but you may have it somewhere elese
 new_model_path = root_dir + f"models/fine_tuned_models/OrpoLlama-3-8B_{epochs}e_qa_qa"
 
 
@@ -412,4 +399,10 @@ tokenizer.push_to_hub(repo_name, use_auth_token=True)
 
 
 
+https://huggingface.co/blog/mlabonne/orpo-llama-3
+mlabonne/orpo-dpo-mix-40k
+https://huggingface.co/datasets/mlabonne/orpo-dpo-mix-40k/tree/main
+
+
+
 [More Information Needed]