Instructions to use hyper-accel/tiny-random-kimi-k2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hyper-accel/tiny-random-kimi-k2 with Transformers:

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

pipe = pipeline("text-generation", model="hyper-accel/tiny-random-kimi-k2", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("hyper-accel/tiny-random-kimi-k2", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use hyper-accel/tiny-random-kimi-k2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hyper-accel/tiny-random-kimi-k2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hyper-accel/tiny-random-kimi-k2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/hyper-accel/tiny-random-kimi-k2

SGLang

How to use hyper-accel/tiny-random-kimi-k2 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 "hyper-accel/tiny-random-kimi-k2" \
    --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": "hyper-accel/tiny-random-kimi-k2",
		"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 "hyper-accel/tiny-random-kimi-k2" \
        --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": "hyper-accel/tiny-random-kimi-k2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use hyper-accel/tiny-random-kimi-k2 with Docker Model Runner:
```
docker model run hf.co/hyper-accel/tiny-random-kimi-k2
```

tiny-random-kimi-k2 / config.json

HyperAccel

Upload tiny-random kimi_k2 model

576780f verified about 2 months ago

raw

history blame contribute delete

2.02 kB

	{
	"_attn_implementation_autoset": false,
	"add_cross_attention": false,
	"architectures": [
	"KimiK2ForCausalLM"
	],
	"attention_bias": false,
	"attention_dropout": 0.0,
	"auto_map": {
	"AutoConfig": "configuration_kimi_k2.KimiK2Config",
	"AutoModelForCausalLM": "modeling_kimi_k2.KimiK2ForCausalLM"
	},
	"aux_loss_alpha": 0.001,
	"bos_token_id": 163584,
	"cross_attention_hidden_size": null,
	"decoder_start_token_id": null,
	"dtype": "float32",
	"eos_token_id": 163585,
	"finetuning_task": null,
	"first_k_dense_replace": 1,
	"head_dim": 16,
	"hidden_act": "silu",
	"hidden_size": 512,
	"initializer_range": 0.02,
	"intermediate_size": 256,
	"is_decoder": false,
	"kv_lora_rank": 32,
	"max_position_embeddings": 131072,
	"model_type": "kimi_k2",
	"moe_intermediate_size": 256,
	"moe_layer_freq": 1,
	"n_group": 1,
	"n_routed_experts": 4,
	"n_shared_experts": 1,
	"norm_topk_prob": true,
	"num_attention_heads": 8,
	"num_experts_per_tok": 2,
	"num_hidden_layers": 2,
	"num_key_value_heads": 8,
	"num_nextn_predict_layers": 0,
	"pad_token_id": null,
	"prefix": null,
	"pretraining_tp": 1,
	"pruned_heads": {},
	"q_lora_rank": 64,
	"qk_head_dim": 48,
	"qk_nope_head_dim": 32,
	"qk_rope_head_dim": 16,
	"rms_norm_eps": 1e-06,
	"rope_interleave": true,
	"rope_parameters": {
	"beta_fast": 1.0,
	"beta_slow": 1.0,
	"factor": 32.0,
	"mscale": 1.0,
	"mscale_all_dim": 1.0,
	"original_max_position_embeddings": 4096,
	"rope_theta": 50000.0,
	"rope_type": "yarn",
	"type": "yarn"
	},
	"routed_scaling_factor": 2.827,
	"scoring_func": "sigmoid",
	"sep_token_id": null,
	"seq_aux": true,
	"task_specific_params": null,
	"tf_legacy_loss": false,
	"tie_encoder_decoder": false,
	"tie_word_embeddings": false,
	"tokenizer_class": null,
	"topk_group": 1,
	"topk_method": "noaux_tc",
	"torchscript": false,
	"transformers_version": "5.3.0",
	"use_bfloat16": false,
	"use_cache": true,
	"v_head_dim": 32,
	"vocab_size": 163840
	}