Search results

Add a differentiable CPU backend for KDA research and prototyping

Status: Open.

#26 In MoonshotAI/FlashKDA;

· xy200303 opened on Jul 29, 2026

Add Pythonic API wrapper with automatic workspace management

Status: Open.

#23 In MoonshotAI/FlashKDA;

· haix716 opened on Jul 28, 2026

Add random seed to bench_fwd.py for reproducible benchmarks

Status: Open.

#22 In MoonshotAI/FlashKDA;

· haix716 opened on Jul 28, 2026

Add standalone usage example and inference-only note

Status: Open.

#21 In MoonshotAI/FlashKDA;

· haix716 opened on Jul 28, 2026

Centralize CHUNK and D constants as single source of truth

Status: Open.

#20 In MoonshotAI/FlashKDA;

· haix716 opened on Jul 28, 2026

Surface CUTLASS initialization and CUDA launch-configuration failures

Status: Open.

#15 In MoonshotAI/FlashKDA;

· morluto opened on Jul 27, 2026

Partial output tiles serialize stores through one lane

Status: Open.

#14 In MoonshotAI/FlashKDA;

· morluto opened on Jul 27, 2026

[Bug] test_fwd failed with CUDA 13

Status: Open.

#12 In MoonshotAI/FlashKDA;

· qianyp18 opened on Jul 17, 2026

Publish FlashKDA kernels on Hugging Face Kernels?

Status: Open.

#11 In MoonshotAI/FlashKDA;

· NielsRogge opened on Jul 15, 2026

[Feature] Intra-Card Context Parallelism to improve B=1 long-sequence performance

Status: Open.

#8 In MoonshotAI/FlashKDA;

· yyq0210 opened on May 29, 2026

[Question] Roadmap for supporting additional K/V head dimensions and backward kernels

Status: Open.

#4 In MoonshotAI/FlashKDA;

· dmm19941210 opened on Apr 24, 2026

gate mode diff

Status: Open.

#3 In MoonshotAI/FlashKDA;

· theNiemand opened on Apr 23, 2026