Michael Feil 71cd6d5533 Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688) ...

* update flash-attn v1

* restore: hdim224

* add 224 flash_fwd_template

* remove whitespace

2024-12-31 09:32:22 +01:00

..

alibi.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

block_info.h

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

dropout.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

error.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

flash_api.cu

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

flash_fwd_hdim32_bf16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim32_bf16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim32_fp16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim32_fp16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim64_bf16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim64_bf16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim64_fp16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim64_fp16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim96_bf16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim96_bf16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim96_fp16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim96_fp16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim128_bf16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim128_bf16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim128_fp16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim128_fp16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim160_bf16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim160_bf16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim160_fp16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim160_fp16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim192_bf16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim192_bf16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim192_fp16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim192_fp16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim224_bf16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim224_bf16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim224_fp16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim224_fp16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim256_bf16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim256_bf16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim256_fp16_causal_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_hdim256_fp16_sm80.cu

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_kernel.h

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash_fwd_launch_template.h

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

flash.h

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

hardware_info.h

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

kernel_helpers.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

kernel_traits_sm90.h

chore: update flash attention kernels (#1518)

2024-01-05 18:28:55 +01:00

kernel_traits.h

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00

kernels.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

mask.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

philox.cuh

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

rotary.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

softmax.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

static_switch.h

Update the flash attn kernels. (#2333)

2024-07-15 20:37:36 +02:00

utils.h

Flash-Attn upgrade / SoftCap Candle-FlashAttn [1/n] (#2688)

2024-12-31 09:32:22 +01:00