Issues

• Replace deprecated torch_dtype with dtype in llava_hf.py

  ModelTC/LightCompress#465 · Filter by author xyf5432 opened on Aug 28, 2026

• 量化qwen3-8b

  ModelTC/LightCompress#452 · Filter by author Jim2016713 opened on Mar 4, 2026

• 量化qwen3-8b

  ModelTC/LightCompress#449 · Filter by author Jim2016713 opened on Mar 3, 2026

• GPTQ per-channel 量化配置

  ModelTC/LightCompress#442 · Filter by author laobadao opened on Nov 4, 2025

• 支持qwen3-vl

  ModelTC/LightCompress#440 · Filter by author ZanePoe opened on Oct 10, 2025
  2

• quant_analysis.py未适配最新代码，import失败

  ModelTC/LightCompress#439 · Filter by author sweetlove12 opened on Sep 4, 2025

• 使用Qwen2.5-VL时模型属性不对应的问题

  ModelTC/LightCompress#437 · Filter by author EustiaAstraea opened on Aug 20, 2025
  1

• Qwen2.5VL 3B calibration using awq takes too long

  ModelTC/LightCompress#433 · Filter by author kritohyh opened on Aug 11, 2025
  3

• bug on awq v2

  ModelTC/LightCompress#432 · Filter by author ChenBinfighting1 opened on Aug 4, 2025

• How to extract input tensors for GEMM after pruning and quantization?

  ModelTC/LightCompress#430 · Filter by author RealJustinNi opened on Aug 1, 2025

• Can llmc support Whisper model quantization?

  ModelTC/LightCompress#419 · Filter by author Ai-ZL opened on Jul 18, 2025

• support gemma3 quantization

  ModelTC/LightCompress#417 · Filter by author MrDoghead opened on Jul 17, 2025

• Why quarot algo R2 rotate needed online_rotate?

  ModelTC/LightCompress#392 · Filter by author 54limiao opened on Jun 19, 2025
  2

• 关于导出vllm的问题

  ModelTC/LightCompress#387 · Filter by author djm012 opened on Jun 4, 2025
  1

• get_float_qparams 忽略量化粒度？

  ModelTC/LightCompress#384 · Filter by author guanchenl opened on May 25, 2025
  2

• SpinQuant merge

  ModelTC/LightCompress#368 · Filter by author Marouan-git opened on Apr 23, 2025
  5

• [W416 15:12:21.131066630 socket.cpp:933] [c10d] The server socket on [::ffff:36.xxx.xxx.13]:38405 has timed out, will retry.

  ModelTC/LightCompress#365 · Filter by author dzy1128 opened on Apr 16, 2025

• 自定义校准数据集文档与代码对不上

  ModelTC/LightCompress#364 · Filter by author xinyubai1209 opened on Apr 14, 2025
  3

• Static Activation Quantization and Mixed-Precision Quantization Incompatibility

  ModelTC/LightCompress#363 · Filter by author aptsunny opened on Apr 11, 2025
  6

• Does llmc support Qwen2.5-VL quantization?

  ModelTC/LightCompress#356 · Filter by author lookinmyeyestellmewhy opened on Apr 9, 2025
  1

• Maybe a bug in "allow_padding"

  ModelTC/LightCompress#354 · Filter by author pengyao96 opened on Apr 8, 2025

• How to get the calibration dataset for Multimodal-LLMs

  ModelTC/LightCompress#353 · Filter by author shixu312349410 opened on Apr 5, 2025

• Two Issues / Bugs with NaiveQuantKVCache Implementation

  ModelTC/LightCompress#350 · Filter by author sasha-hailo opened on Mar 31, 2025
  1

• Saving/Loading Fake Quant weights

  ModelTC/LightCompress#349 · Filter by author allzero-kwon opened on Mar 31, 2025
  1

• OOM Error when Running AWQ + OmniQuant Combination (Step 2: OmniQuant) Despite Using Multiple GPUs

  ModelTC/LightCompress#347 · Filter by author Barryshen1 opened on Mar 28, 2025