Instructions to use tiny-random/kimi-linear with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tiny-random/kimi-linear with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="tiny-random/kimi-linear", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("tiny-random/kimi-linear", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use tiny-random/kimi-linear with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tiny-random/kimi-linear"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tiny-random/kimi-linear",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tiny-random/kimi-linear

SGLang

How to use tiny-random/kimi-linear 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 "tiny-random/kimi-linear" \
    --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": "tiny-random/kimi-linear",
		"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 "tiny-random/kimi-linear" \
        --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": "tiny-random/kimi-linear",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tiny-random/kimi-linear with Docker Model Runner:
```
docker model run hf.co/tiny-random/kimi-linear
```

kimi-linear

File size: 1,636 Bytes

fecd356

{
  "architectures": [
    "KimiLinearForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "moonshotai/Kimi-Linear-48B-A3B-Instruct--configuration_kimi.KimiLinearConfig",
    "AutoModel": "moonshotai/Kimi-Linear-48B-A3B-Instruct--modeling_kimi.KimiLinearModel",
    "AutoModelForCausalLM": "moonshotai/Kimi-Linear-48B-A3B-Instruct--modeling_kimi.KimiLinearForCausalLM"
  },
  "bos_token_id": 163584,
  "dtype": "bfloat16",
  "eos_token_id": 163586,
  "first_k_dense_replace": 1,
  "head_dim": 32,
  "hidden_act": "silu",
  "hidden_size": 8,
  "initializer_range": 0.02,
  "intermediate_size": 32,
  "kv_lora_rank": 512,
  "linear_attn_config": {
    "full_attn_layers": [
      4
    ],
    "head_dim": 32,
    "kda_layers": [
      1,
      2,
      3
    ],
    "num_heads": 8,
    "short_conv_kernel_size": 4
  },
  "mla_use_nope": true,
  "model_max_length": 1048576,
  "model_type": "kimi_linear",
  "moe_intermediate_size": 32,
  "moe_layer_freq": 1,
  "moe_renormalize": true,
  "moe_router_activation_func": "sigmoid",
  "num_attention_heads": 8,
  "num_expert_group": 1,
  "num_experts": 256,
  "num_experts_per_token": 8,
  "num_hidden_layers": 5,
  "num_key_value_heads": 8,
  "num_nextn_predict_layers": 0,
  "num_shared_experts": 1,
  "pad_token_id": 163839,
  "q_lora_rank": null,
  "qk_nope_head_dim": 128,
  "qk_rope_head_dim": 64,
  "rms_norm_eps": 1e-05,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "routed_scaling_factor": 2.446,
  "tie_word_embeddings": false,
  "topk_group": 1,
  "transformers_version": "4.57.1",
  "use_cache": true,
  "use_grouped_topk": true,
  "v_head_dim": 128,
  "vocab_size": 163840
}