Optimum Intel 利用 Intel Extension for Pytorch, (IPEX) 优化，在 Intel 架构上加速 Hugging Face 流水线

可以通过由 LlamaIndex 封装的 OptimumIntelLLM 实体在本地运行 Optimum Intel 模型：

在下面一行中，我们安装本演示所需的软件包：

%pip install llama-index-llms-optimum-intel

设置完成后，我们来试试：

如果你在 Colab 中打开这个 Notebook，可能需要先安装 LlamaIndex 🦙。

!pip install llama-index

from llama_index.llms.optimum_intel import OptimumIntelLLM

def messages_to_prompt(messages):
    prompt = ""
    for message in messages:
        if message.role == "system":
            prompt += f"<|system|>\n{message.content}</s>\n"
        elif message.role == "user":
            prompt += f"<|user|>\n{message.content}</s>\n"
        elif message.role == "assistant":
            prompt += f"<|assistant|>\n{message.content}</s>\n"

    # ensure we start with a system prompt, insert blank if needed
    if not prompt.startswith("<|system|>\n"):
        prompt = "<|system|>\n</s>\n" + prompt

    # add final assistant prompt
    prompt = prompt + "<|assistant|>\n"

    return prompt

def completion_to_prompt(completion):
    return f"<|system|>\n</s>\n<|user|>\n{completion}</s>\n<|assistant|>\n"

模型加载

可以通过 OptimumIntelLLM 方法指定模型参数来加载模型。

oi_llm = OptimumIntelLLM(
    model_name="Intel/neural-chat-7b-v3-3",
    tokenizer_name="Intel/neural-chat-7b-v3-3",
    context_window=3900,
    max_new_tokens=256,
    generate_kwargs={"temperature": 0.7, "top_k": 50, "top_p": 0.95},
    messages_to_prompt=messages_to_prompt,
    completion_to_prompt=completion_to_prompt,
    device_map="cpu",
)

response = oi_llm.complete("What is the meaning of life?")
print(str(response))

流式

使用 stream_complete 端点

response = oi_llm.stream_complete("Who is Mother Teresa?")
for r in response:
    print(r.delta, end="")

使用 stream_chat 端点

from llama_index.core.llms import ChatMessage

messages = [
    ChatMessage(
        role="system",
        content="You are an American chef in a small restaurant in New Orleans",
    ),
    ChatMessage(role="user", content="What is your dish of the day?"),
]
resp = oi_llm.stream_chat(messages)

for r in resp:
    print(r.delta, end="")