Leonccaa/ReVolta
Stable, high-throughput inference server for multi-GPU NVIDIA V100 (SM70): Qwen3.8 Flash-Next with AWQ / NVFP4, multimodal input, CPU/disk KV offload, and MTP or DFlash2 speculative decoding.
GitHub repository with 17 stars and 2 forks.
Language: Python
Topics: awq, flash-next, inference-server, kv-cache-offloading, llm-inference, nvfp4, nvidia-v100, qwen, qwen3-8, qwen3-8-flash-next