--- license: other license_name: license license_link: LICENSE ---

Index-1.9B-Chat-GGUF

此为[Index-1.9B-Chat](https://huggingface.co/IndexTeam/Index-1.9B-Chat)的GGUF版本,适配llama.cpp,同时提供了Ollma的ModelFile适配。 更多细节详见我们的[GitHub](https://github.com/bilibili/Index-1.9B)和[Index-1.9B技术报告](https://github.com/bilibili/Index-1.9B/blob/main/Index-1.9B%20%E6%8A%80%E6%9C%AF%E6%8A%A5%E5%91%8A.pdf) ### LLAMA.CPP ```shell # 安装llama.cpp(https://github.com/ggerganov/llama.cpp) git clone https://github.com/ggerganov/llama.cpp cd llama.cpp make # 安装llama-cpp-python(https://github.com/abetlen/llama-cpp-python) pip install llama-cpp-python ``` llama.cpp终端交互 ```shell ./build/bin/llama-cli -m models/Index-1.9B-Chat/ggml-model-bf16.gguf --color -if ``` **注意!!** llama.cpp并不支持自定义chat_template, 故而需要自己拼接prompt,Index-1.9B的chat_template(示意)为 ```shell # 三个分隔符为 (token_id=0), reserved_0(token_id=3), reserved_1(token_id=4) []sytem_message[reserved_0]user_message[reserved_1]response ``` 使用llama-cpp-python可支持自定义的chat_template(已写入GGUF,可直接使用) ```python from llama_cpp import Llama model_path = "Index-1.9B-Chat-GGUF/ggml-model-Q6_K.gguf" llm = Llama(model_path =model_path, verbose=True) output = llm.create_chat_completion( messages = [ {"role": "system", "content": "你是由哔哩哔哩自主研发的大语言模型,名为“Index”。你能够根据用户传入的信息,帮助用户完成指定的任务,并生成恰当的、符合要求的回复。"}, #{"role": "system", "content": "你需要扮演B站评论区老哥,用评论区阴阳怪气的话术回复,不要说你是AI"}, {"role": "user","content": "篮球和鸡有什么关系"} ] ) print(output) ``` ### OLLAMA - 安装[Ollama](https://github.com/ollama/ollama) ```shell curl -fsSL https://ollama.com/install.sh | sh ``` ```shell # 启动server ollama serve # 适配模型, OllamaModelFile中可修改模型文件和System Message ollama create Index-1.9B-Chat -f Index-1.9B-Chat-GGUF/OllamaModelFile # 启动终端 ollama run Index-1.9B-Chat # web调用,可以动态指定System Message curl http://localhost:11434/api/chat -d '{ "model": "Index-1.9B-Chat", "messages": [ { "role": "system", "content": "你是由哔哩哔哩自主研发的大语言模型,名为“Index”。你能够根据用户传入的信息,帮助用户完成指定的任务,并生成恰当的、符合要求的回复。" }, { "role": "user", "content": "续写 金坷垃" } ] }' ```