Instructions to use moonshotai/Kimi-K2-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use moonshotai/Kimi-K2-Instruct with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="moonshotai/Kimi-K2-Instruct", trust_remote_code=True) messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("moonshotai/Kimi-K2-Instruct", trust_remote_code=True, dtype="auto") - Inference
- HuggingChat
- Notebooks
- Google Colab
- Kaggle
- Local Apps
- vLLM
How to use moonshotai/Kimi-K2-Instruct with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "moonshotai/Kimi-K2-Instruct" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "moonshotai/Kimi-K2-Instruct", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/moonshotai/Kimi-K2-Instruct
- SGLang
How to use moonshotai/Kimi-K2-Instruct with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "moonshotai/Kimi-K2-Instruct" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "moonshotai/Kimi-K2-Instruct", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "moonshotai/Kimi-K2-Instruct" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "moonshotai/Kimi-K2-Instruct", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use moonshotai/Kimi-K2-Instruct with Docker Model Runner:
docker model run hf.co/moonshotai/Kimi-K2-Instruct
Update tokenizer_config.json
Support tool call, temp 0 tau2-airline went from 14% to 50%. Old template was ignoring all the previous tool calls, breaking all multi step tool call conversations.
thanks.. are u using sglang or vllm to host? i tested the template, multistep tool call in sglang still not working
Thank you so much. We forgot to put that in chat template.
In addition, role=tool message has a tool_call_id attribute that should be encoded to the content, the format is
"<|im_system|>tool<|im_middle|>" +
"## Return of {{tool_call_id}}\n" + "{{content}}" +
"<|im_end|>"
Can you update this part?
BTW, did you decipher this template from raw tokens?
@bchenfireworks thank you very much !
Also, I would like to suggest the following updates:
(1) use {{ tool_call['function']['arguments'] | tojson}} instead, to make sure arguments are legal json.
(2) add more - in control statements to eliminate blanks, so we can get a compact string.
Thanks! Let update the template following your suggestion.
Thanks! Let update the template following your suggestion.
Since the template string it too loooooong to review, could you pls also provide us an example (chat history and the rendered string), so we can verify it locally. Thank you very much, really.