amd/Kimi-K2-Thinking-W4A8

Model Overview

• Model Architecture: Kimi-K2-Thinking
  • Input: Text
  • Output: Text
• Supported Hardware Microarchitecture: AMD MI300/MI355
• ROCm: 7.0
• PyTorch: 2.8.0
• Transformers: 4.53.0
• Operating System(s): Linux
• Inference Engine: vLLM
• Model Optimizer: AMD-Quark (V0.10)
  • Weight quantization: INT4 Per-Channel & FP8E4M3, Static
  • Activation quantization: FP8E4M3, Dynamic

This model was built with moonshotai Kimi-K2-Thinking model by applying AMD-Quark for INT4-FP8 quantization.

Model Quantization

The model was quantized from moonshotai/Kimi-K2-Thinking using AMD-Quark.

Deployment

This model can be deployed efficiently using the vLLM backends.

Evaluation

The model was evaluated on GSM8K benchmarks using the lm-evaluation-harness framework.

Accuracy

Benchmark	Kimi-K2-Thinking	Kimi-K2-Thinking-W4A8(this model)	Recovery
GSM8K	93.93	93.4	99.4%

Reproduction

The results of GSM8K were obtained using lm-evaluation-harness and latest vLLM.

Launch vLLM

MODEL_DIR=/data/amd/Kimi-K2-Thinking-W4A8
VLLM_ATTENTION_BACKEND="TRITON_MLA" VLLM_ROCM_USE_AITER=1 VLLM_ROCM_USE_AITER_FUSION_SHARED_EXPERTS=0 VLLM_ROCM_USE_AITER_FP4BMM=0 vllm serve $MODEL_DIR \
    --port 8001 \
    --trust-remote-code \
    --gpu-memory-utilization 0.9 \
    --tensor-parallel-size 8 \
    --load-format "fastsafetensors"

GSM8K evaluation

MODEL_ARGS="model=/data/amd/Kimi-K2-Thinking-W4A8,base_url=http://localhost:8001/v1/completions,num_concurrent=999999,timeout=999999,tokenized_requests=False,max_length=38768,temperature=0.6,top_p=0.95,add_bos_token=True,seed=$SEED,trust_remote_code=True"
lm_eval \
    --model local-completions \
    --model_args $MODEL_ARGS \
    --tasks gsm8k \
    --num_fewshot 8 \
    --batch_size auto

License

Modifications Copyright(c) 2025 Advanced Micro Devices, Inc. All rights reserved.