Views
No views yet
Qwen/Qwen3-32B. Companion to drawais/Qwen3-32B-AWQ-INT4.| Source params | 32B |
| Quantized weights | ~19.7 GB on disk |
| Inference VRAM (incl. KV cache @ 32K context) | ~24 GB |
drawais/Qwen3-32B-AWQ-INT4 scored 100.0% overall on drawais/needle-1M-bench-mvp (50K-token haystack, real arxiv text). Same base, comparable quality expected. A direct HQQ score will land once a vLLM-compatible HQQ pathway is finalized.1from transformers import AutoTokenizer
2from hqq.models.hf.base import AutoHQQHFModel
3
4tok = AutoTokenizer.from_pretrained("drawais/Qwen3-32B-HQQ-INT4")
5model = AutoHQQHFModel.from_quantized("drawais/Qwen3-32B-HQQ-INT4", device="cuda")