Make png images into jpg for dark mode

This commit is contained in:
Dan Fu 2022-05-28 22:46:49 +01:00
parent 4decc3c166
commit 7025a092d1
5 changed files with 2 additions and 2 deletions

View File

@ -40,14 +40,14 @@ Our graphs show sequence lengths between 128 and 4096 (when standard attention r
#### Speedup
![FlashAttention speedup](images/flashattn_speedup.png)
![FlashAttention speedup](images/flashattn_speedup.jpg)
We generally see 2-4X speedup at sequence lengths between 128 and 4K, and we see more speedup when using dropout and masking, since we fuse the kernels.
At sequence lengths that are popular with language models like 512 and 1K, we see speedups up to 4X when using dropout and masking.
#### Memory
![FlashAttention memory](images/flashattn_memory.png)
![FlashAttention memory](images/flashattn_memory.jpg)
We show memory savings in this graph (note that memory footprint is the same no matter if you use dropout or masking).
Memory savings are proportional to sequence length -- since standard attention has memory quadratic in sequence length, whereas FlashAttention has memory linear in sequence length.

BIN
images/flashattn_memory.jpg Normal file

Binary file not shown.

After

Width:  |  Height:  |  Size: 83 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 64 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 113 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 80 KiB