Search results

[CuTe, FWD, SM100]: ValueError("mask_block_cnt and mask_block_idx must be provided for block sparsity.")

Status: Open.

#1979 In Dao-AILab/flash-attention;

· complexfilter opened on Oct 31, 2025

✅ Prebuilt wheel for FlashAttention 2.8.0.post2 (CUDA 12.1 / NVIDIA L40S / PyTorch 2.9 / Python 3.10)

Status: Open.

#1978 In Dao-AILab/flash-attention;

· h1312200313122003-code opened on Oct 31, 2025

Make alibi slopes a trainable parameter?

Status: Open.

#1977 In Dao-AILab/flash-attention;

· penguinshin opened on Oct 31, 2025

[Cute,Bwd,Sm100] ValueError: invalid mode element for input of rank 3, got mode=[1, 3, 2, 0]

Status: Open.

#1976 In Dao-AILab/flash-attention;

· cassanof opened on Oct 30, 2025

Failed docker build Lightx2v Dockerfile -- VLM+FLashAttention issue

Status: Open.

#1975 In Dao-AILab/flash-attention;

· christopher5106 opened on Oct 30, 2025

FlashAttention3 forward producing NaN output when NaN exist in parts of input data that it should not be reading

Status: Open.

#1974 In Dao-AILab/flash-attention;

· tdoublep opened on Oct 30, 2025

Learnable sink gradient support in flash_attn.cute

Status: Open.

#1973 In Dao-AILab/flash-attention;

· jootanehorror opened on Oct 30, 2025

Support for compute capability 12.1 (sm121) - NVIDIA GB10 GPU

Status: Open.

#1969 In Dao-AILab/flash-attention;

· swatson1000000 opened on Oct 28, 2025

CUDA error: an illegal memory access was encountered

Status: Open.

#1968 In Dao-AILab/flash-attention;

· zhangboSJTU opened on Oct 28, 2025

Prebuilt wheels for PyTorch 2.9

Status: Open.

#1967 In Dao-AILab/flash-attention;

· lgeiger opened on Oct 27, 2025

CUDA Error: no kernel image is available for execution on the device flash-attention/csrc/layer_norm/ln_fwd_kernels.cuh 236

Status: Open.

#1966 In Dao-AILab/flash-attention;

· janelu9 opened on Oct 27, 2025

head dims mismatch for Qwen 3 VL (30ba3, 8b, etc..)

Status: Open.

#1965 In Dao-AILab/flash-attention;

· bhaktatejas922 opened on Oct 27, 2025