Search results

How to convert BF16 safetensors Generative AI models into NVFP4 like Qwen 2512, Wan 2.2, LTX 2, FLUX SRPO or any other model

bugSomething isn't workingSomething isn't working

Status: Open.

#775 In NVIDIA/Model-Optimizer;

· FurkanGozukara opened on Jan 13, 2026

Can the model be trained simultaneously using quantization and sparsity?

questionHelp is is neededHelp is is needed

Status: Open.

#771 In NVIDIA/Model-Optimizer;

· Vieeo opened on Jan 13, 2026

How to convert FP16 or FP8 model files to NVFP4 model files

questionHelp is is neededHelp is is needed

Status: Open.

#770 In NVIDIA/Model-Optimizer;

· Alan-D-Chen opened on Jan 13, 2026

Kimi K2 NVFP4 ModelOpt Calibration Error

bugSomething isn't workingSomething isn't working

Status: Open.

#766 In NVIDIA/Model-Optimizer;

· pdasgup opened on Jan 12, 2026

DeepSeek-V3.2-Exp NVFP4 quantized model fails with flash_mla KV cache dtype incompatibility (TensorRT-LLM 1.2.0rc7)

bugSomething isn't workingSomething isn't working

Status: Open.

#763 In NVIDIA/Model-Optimizer;

· evgeniiperepelkin opened on Jan 12, 2026

Any plans to add support for vlm pruning, like Qwen/Qwen3-VL-8B-Instruct?

feature requestNew feature or requestNew feature or request

Status: Open.

#761 In NVIDIA/Model-Optimizer;

· jackmego opened on Jan 12, 2026

NVFP4 __Dyna layer took most of the inference time

bugSomething isn't workingSomething isn't working

Status: Open.

#751 In NVIDIA/Model-Optimizer;

· CHOcho-quan opened on Jan 8, 2026

NVFP4 Quantization for GLM 4.7 w MTP

bugSomething isn't workingSomething isn't working

Status: Open.

#750 In NVIDIA/Model-Optimizer;

· pdasgup opened on Jan 8, 2026

Need guidance on converting ModelOpt QAT model (ResNet50) to ONNX then TensorRT

questionHelp is is neededHelp is is needed

Status: Open.

#741 In NVIDIA/Model-Optimizer;

· Sam-gsj opened on Jan 7, 2026

When try to prune a Qwen2.5 vl model, I got error: Only rope position embedding is supported, got mrope

feature requestNew feature or requestNew feature or request

torch.pruning

Status: Open.

#739 In NVIDIA/Model-Optimizer;

· jackmego opened on Jan 6, 2026

MTP quantizing doesn't work - Qwen3 Next

bugSomething isn't workingSomething isn't working

model support

torch.quantization

waiting for feedback

Status: Open.

#733 In NVIDIA/Model-Optimizer;

· nutriarch opened on Dec 28, 2025

[Doc - Guidance] Proper MoE quantization

documentationImprovements or additions to documentationImprovements or additions to documentation

feature requestNew feature or requestNew feature or request

torch.quantization

waiting for feedback

Status: Open.

#732 In NVIDIA/Model-Optimizer;

· mratsim opened on Dec 28, 2025