vrr/kvcache-ai-ktransformers
2
0
Fork 0
mirror of https://github.com/kvcache-ai/ktransformers.git synced 2025-09-09 22:05:30 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Solve torch.backends.cuda.sdp_kernel() is deprecated.

Browse source
This commit is contained in:
MuWinds 2025-02-15 12:41:51 +08:00 • committed by GitHub
parent 1548c99234
commit f74c2d1d17
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

3
ktransformers/server/backend/interfaces/transformers.py
View file

@ -13,6 +13,7 @@ from transformers import (
from ktransformers.server.config.config import Config
from ktransformers.server.schemas.base import ObjectID
from ktransformers.server.utils.multi_timer import Profiler
from torch.nn.attention import SDPBackend
import torch
import sys, os
from ..base import ThreadContext, BackendInterfaceBase
@ -292,7 +293,7 @@ class TransformersInterface(BackendInterfaceBase):
def generate(self):
self.profiler.set_counter("decode", 0)
for _ in range(1, self.args.max_new_tokens):
with torch.backends.cuda.sdp_kernel(enable_flash=False, enable_mem_efficient=False, enable_math=True):
with torch.nn.attention.sdpa_kernel(backends=[SDPBackend.FLASH_ATTENTION, SDPBackend.MATH, SDPBackend.EFFICIENT_ATTENTION]):
next_token = self.decode_one_tokens()
self.profiler.inc("decode")
if next_token == self.tokenizer.eos_token_id: