This model is a FP4 (w4a16_fp4) compressed version of DeepSeek's distill of Qwen3 8B.

VLLM Starter:

python3 -m vllm.entrypoints.openai.api_server --host 0.0.0.0 --port 8000 --model miike-ai/DeepSeek-R1-0528-Qwen3-5B-FP4 --enable-auto-tool-choice --tool-call-parser hermes --enable-reasoning --reasoning-parser deepseek_r1

Downloads last month: 14

Safetensors

Model size

5.15B params

Tensor type

F32

BF16

F8_E4M3

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for miike-ai/DeepSeek-R1-0528-Qwen3-8B-FP4

Base model

deepseek-ai/DeepSeek-R1-0528-Qwen3-8B

Quantized

(84)

this model

Collection including miike-ai/DeepSeek-R1-0528-Qwen3-8B-FP4

FP4

Collection

FP4 compressed models • 7 items • Updated Jul 5