All issues

Search results

[Bug] init_quantized_weights / --low_memory_mode exports numerically broken NVFP4 (quantizes meta tensors before weights load)

Status: Open.

#2160 In NVIDIA/Model-Optimizer;

· spped2000 opened on Aug 12, 2026

Can auto_quant calculate the score for kv-cache separately?

questionHelp is is neededHelp is is needed

Status: Open.

#2158 In NVIDIA/Model-Optimizer;

· zcfh opened on Aug 12, 2026

Support MiniMax-H3 Visual VAE quantization in ModelOpt

feature requestNew feature or requestNew feature or request

Status: Open.

#2137 In NVIDIA/Model-Optimizer;

· baonudesifeizhai opened on Aug 10, 2026

fold_weight crashes with AttributeError: 'NoneType' object has no attribute 'data' on Megatron models with tied word embeddings

bugSomething isn't workingSomething isn't working

Status: Open.

#2131 In NVIDIA/Model-Optimizer;

· babyplutokurt opened on Aug 10, 2026

# [ONNX][Autotune] Integrated quantization does not preserve AutoTune Q/DQ placement on ViT

bugSomething isn't workingSomething isn't working

Status: Open.

#2123 In NVIDIA/Model-Optimizer;

· HannisLee opened on Aug 10, 2026

[ONNX PTQ] Support for third-party custom ORT/TRT plugins in static calibration

feature requestNew feature or requestNew feature or request

Status: Open.

#2016 In NVIDIA/Model-Optimizer;

· e-said opened on Jul 24, 2026

Recommended FP8 recipe for Blackwell (B200)? Per-tensor FP8_DEFAULT_CFG underperforms BF16 at prefill; NVFP4 much faster. Also: exported FP8 checkpoint has no KV q/k/v_scale

bugSomething isn't workingSomething isn't working

Status: Open.

#2015 In NVIDIA/Model-Optimizer;

· Francesco-Carlucci opened on Jul 24, 2026

NVFP4 support for Qwen3_5MoeExperts (fused MoE): quantizer registration + export serializer

Status: Open.

#2011 In NVIDIA/Model-Optimizer;

· frischeDaten opened on Jul 23, 2026

[QAD] Request for exact training configurations used in arXiv:2601.20088

Status: Open.

#2002 In NVIDIA/Model-Optimizer;

· Sekri0 opened on Jul 21, 2026

Public preset quant_cfg changed from dict to list in 0.44 (breaking, undocumented?)

Status: Open.

#2001 In NVIDIA/Model-Optimizer;

· sayakpaul opened on Jul 21, 2026

Nvidia Modelopt structured 2:4 weight sparsity showing no speed improvement compared to dense model

bugSomething isn't workingSomething isn't working

Status: Open.

#1974 In NVIDIA/Model-Optimizer;

· Pavan6136 opened on Jul 14, 2026

Quantizing error for Qwen3.6 35B A3B

bugSomething isn't workingSomething isn't working

Status: Open.

#1933 In NVIDIA/Model-Optimizer;

· kristianpaul opened on Jul 7, 2026