Openchat V2

🚀 OpenChat: Advancing Open-source Language Models with Imperfect Data

OpenChat v2 family is inspired by offline reinforcement learning, offering conditional behavior cloning (OpenChat-v2) and weighted behavior cloning (OpenChat-v2-w). It aims to enhance the performance of open - source language models using imperfect data.

✨ Features

• OpenChat-v2-w: Trained on ~80k cleaned ShareGPT data with conditioning and weighted loss, based on LLaMA - 13B with a context length of 2048.
  • It achieves a 50.9% win - rate over ChatGPT on MT - bench.
  • It achieves a 79.4% win - rate over ChatGPT on Vicuna - bench.
  • It achieves an 87.1% win - rate over text - davinci - 003 on AlpacaEval.
• OpenChat-v2: Trained on ~80k cleaned ShareGPT data with only conditioning, based on LLaMA - 13B with a context length of 2048.
  • It achieves a 48.1% win - rate over ChatGPT on MT - bench.
  • It achieves an 80.6% win - rate over ChatGPT on Vicuna - bench.
  • It achieves an 85.0% win - rate over text - davinci - 003 on AlpacaEval.

📦 Installation

No specific installation steps are provided in the original README. If you want to use the model, you can refer to the following related sections for more information.

💻 Usage Examples

Basic Usage

The conversation template involves concatenating tokens, and cannot be expressed in plain - text. Besides base model vocabulary, an end - of - turn token <|end_of_turn|> is added. Here is an example of single - round conversation template:

def tokenize_single_input(tokenizer, prompt):
    # OpenChat V2
    human_prefix = "User:"
    prefix    = "Assistant GPT4:"
    eot_token = "<|end_of_turn|>"
    bos_token = "<s>"

    def _tokenize(text):
        return tokenizer.convert_tokens_to_ids(tokenizer._tokenize(text))

    def _tokenize_special(special_name):
        return tokenizer.convert_tokens_to_ids(special_name)
    
    return [_tokenize_special(bos_token)] + _tokenize(human_prefix) + _tokenize(prompt) + [_tokenize_special(eot_token)] + \
           _tokenize(prefix)

Advanced Usage

To explore conditional language models, you can also set prefix = "Assistant GPT3:" to mimic ChatGPT behavior (this may cause performance degradation).

Hint: In BPE, tokenize(A) + tokenize(B) does not always equals to tokenize(A + B)

📚 Documentation

Code and Inference Server

We provide the full source code, including an inference server compatible with the "ChatCompletions" API, in the OpenChat GitHub repository.

Web UI

OpenChat also includes a web UI for a better user experience. See the GitHub repository for instructions.

🔧 Technical Details

Limitations

Foundation Model Limitations

Despite its advanced capabilities, OpenChat is still bound by the limitations inherent in its foundation models. These limitations may impact the model's performance in areas such as:

• Complex reasoning
• Mathematical and arithmetic tasks
• Programming and coding challenges

Hallucination of Non - existent Information

OpenChat may sometimes generate information that does not exist or is not accurate, also known as "hallucination". Users should be aware of this possibility and verify any critical information obtained from the model.

📄 License

The license of this project is other.