huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-04-22 15:32:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

cleaning

Browse Source
This commit is contained in:
fxmarty 2024-04-19 11:57:16 +00:00
parent 1b4c8b4b3e
commit ec5343ec5e
2 changed files with 5 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
server/text_generation_server/utils/flash_attn.py
View File

@ -28,10 +28,10 @@ ROCM_USE_FLASH_ATTN_V2_TRITON = False
if IS_ROCM_SYSTEM:
if os.getenv("ROCM_USE_FLASH_ATTN_V2_TRITON", "").lower() == "true":
ROCM_USE_FLASH_ATTN_V2_TRITON = True
logger.info("ROCm: using Flash Attention 2 Triton implementaion.")
logger.info("ROCm: using Flash Attention 2 Triton implementation.")
else:
ROCM_USE_FLASH_ATTN_V2_CK = True
logger.info("ROCm: using Flash Attention 2 Composable Kernel implementaion.")
logger.info("ROCm: using Flash Attention 2 Composable Kernel implementation.")
try:
try:

5
server/text_generation_server/utils/layers.py
View File

@ -1291,5 +1291,6 @@ try:
freqs = torch.outer(t, self.inv_freq.to(device=t.device))
self._cos_cached = (torch.cos(freqs) * self.mscale).to(dtype)
self._sin_cached = (torch.sin(freqs) * self.mscale).to(dtype)
except ImportError as e:
raise e
except ImportError:
pass