Tinyllama 1.1B Chat V0.6

🚀 TinyLlama-1.1B

The TinyLlama project aims to pretrain a 1.1B Llama model on 3 trillion tokens. With proper optimization, this can be achieved in "just" 90 days using 16 A100 - 40G GPUs 🚀🚀. The training commenced on September 1st, 2023.

TinyLlama adopts the exact same architecture and tokenizer as Llama 2. This enables it to be easily integrated into many open - source projects built upon Llama. Moreover, with only 1.1B parameters, TinyLlama is compact, making it suitable for various applications with limited computation and memory requirements.

🚀 Quick Start

This Model

This is a chat model fine - tuned on TinyLlama/TinyLlama-1.1B-intermediate-step-955k-2T. We follow HF's Zephyr's training recipe. The model was initially fine - tuned on a variant of the UltraChat dataset, which consists of a diverse range of synthetic dialogues generated by ChatGPT. Then, we further aligned the model with 🤗 TRL's DPOTrainer on the openbmb/UltraFeedback dataset, which contains 64k prompts and model completions ranked by GPT - 4.

How to use

You will need transformers>=4.34. Check the TinyLlama GitHub page for more information.

💻 Usage Examples

Basic Usage

# Install transformers from source - only needed for versions <= v4.34
# pip install git+https://github.com/huggingface/transformers.git
# pip install accelerate

import torch
from transformers import pipeline

pipe = pipeline("text-generation", model="TinyLlama/TinyLlama-1.1B-Chat-v0.6", torch_dtype=torch.bfloat16, device_map="auto")

# We use the tokenizer's chat template to format each message - see https://huggingface.co/docs/transformers/main/en/chat_templating
messages = [
    {
        "role": "system",
        "content": "You are a friendly chatbot who always responds in the style of a pirate",
    },
    {"role": "user", "content": "How many helicopters can a human eat in one sitting?"},
]
prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
outputs = pipe(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])
# <|system|>
# You are a friendly chatbot who always responds in the style of a pirate.</s>
# <|user|>
# How many helicopters can a human eat in one sitting?</s>
# <|assistant|>
# ...

📄 License

This project is licensed under the Apache - 2.0 license.

📦 Technical Details