Repository Index

Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.

• kv-cache
• llm-inference
• local-llm
• quantization

Demo of high performance AI model (pmml export) evaluator api base on Vert.x, Supports dynamic routing configuration for multiple pmml models via JSON

• pmml
• ai
• java-pmml
• verticle