Manish-KT/Fine_tune_Llama_2

Fine-Tuned Llama 2 Model

Model Description

This model is a fine-tuned version of Llama 2, trained on a dataset of diverse prompts and scenarios. The model has been designed to generate responses based on various tasks described in the prompt column of the dataset. The fine-tuning process aims to improve the model's performance in handling specific tasks across multiple domains, such as software development, SEO, and Linux commands.

Dataset Information

The dataset used for fine-tuning this model consists of two primary columns:

1. act: The role or scenario that the model is asked to act upon. For example:

   • "An Ethereum Developer"
   • "SEO Prompt"
   • "Linux Terminal"

2. prompt: The detailed task or scenario description related to the act. This provides context and specific instructions that the model needs to follow. Example prompts:

   • "Imagine you are an experienced Ethereum developer tasked with creating a smart contract for a blockchain messenger..."
   • "Using WebPilot, create an outline for an article that will be 2,000 words on the keyword 'Best SEO prompts'..."
   • "I want you to act as a linux terminal. I will type commands and you will reply with what the terminal should show..."

The dataset includes a wide range of scenarios aimed at helping the model generalize across technical and creative tasks.

Dataset Samples

Act	Prompt
Ethereum Developer	Imagine you are an experienced Ethereum developer tasked with creating a smart contract for a blockchain...
SEO Prompt	Using WebPilot, create an outline for an article that will be 2,000 words on the keyword 'Best SEO prompts'...
Linux Terminal	I want you to act as a linux terminal. I will type commands and you will reply with what the terminal should show...

Output Example

The model has been fine-tuned to generate detailed, contextually relevant responses based on the prompts provided. Here’s an example of how the model might respond to a sample prompt:

Input:

Act: Linux Terminal

Output:

Prompt: I want you to act as a Linux terminal. I will type commands, and you will reply with what the terminal should show. Execute the command ls.

In this scenario, the model understands that it should act as a Linux terminal and simulate the result of running the ls command.

Another Example:

Input:

Act: Ethereum Developer

Output:

Prompt: Imagine you are an experienced Ethereum developer tasked with creating a smart contract for a blockchain messenger...

In this example, the model generates Solidity code based on the prompt, addressing the requirements for a blockchain messenger.

How to Use the Model

This model can be loaded and used through the Hugging Face Hub:

from transformers import AutoModelForCausalLM, AutoTokenizer

# Load the model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("Manish-KT/Fine_tune_Llama_2")
model = AutoModelForCausalLM.from_pretrained("Manish-KT/Fine_tune_Llama_2")

# Encode the prompt
inputs = tokenizer("I want you to act as a linux terminal.", return_tensors="pt")

# Generate the response
outputs = model.generate(inputs["input_ids"], max_length=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Acknowledgements

Special thanks to the creators of the dataset fka/awesome-chatgpt-prompts, which provided the rich prompts and diverse scenarios used in fine-tuning this model.

License

This model is open-sourced and can be used for both commercial and non-commercial purposes. Please ensure that you attribute the original dataset and respect any usage policies.

license: mit