dolfsai/Qwen3-Embedding-4B-W4A16_ASYM

prudant/Qwen3-Embedding-4B-W4A16_ASYM

This is a compressed version of Qwen/Qwen3-Embedding-4B using llm-compressor with the following scheme: W4A16_ASYM

Important: You MUST read the following guide for correct usage of this model here Guide

(Check pooling configuration in VLLM and best pooling mode for this model)

Model Details

• —Original Model: Qwen/Qwen3-Embedding-4B
• —Quantization Method: AWQ
• —Compression Libraries: llm-compressor
• —Calibration Dataset: HuggingFaceH4/ultrachat_200k (1024 samples)
• —Optimized For: Inference with vLLM
• —License: same as original model