nvidia
/

Kimi-K2.5-NVFP4

Text Generation

Model Optimizer

Model card Files Files and versions

Resources

View closed (4)

TemporalMesh Transformer: 29.4 PPL at 48% compute — beats Mamba, new open-source architecture

#9 opened 24 days ago by

Cannot reproduce claimed 96.3 AIME 2025 score

#7 opened 3 months ago by

Update README.md

#3 opened 4 months ago by

Having issues with repetition.

#2 opened 4 months ago by

[Question] gate_proj and up_proj weight_scale_2 mismatch in layer 3 of Kimi-K2.5-NVFP4

#1 opened 5 months ago by