Gahow Wang
445e491123
third_party/vllm/ now tracked in git for direct patch management.
Based on vLLM v0.18.1 release with one patch applied:
vllm/v1/core/sched/scheduler.py:
Replace fatal assert with graceful skip when KV transfer callback
arrives for an already-aborted request during PD disaggregated serving.
Future vLLM modifications should be made directly in third_party/vllm/
and committed normally. The patches/ directory is kept as documentation
of what changed from upstream.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
18 lines
606 B
Python
18 lines
606 B
Python
# SPDX-License-Identifier: Apache-2.0
|
|
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
|
|
|
import torch
|
|
|
|
# Reference default values of atol and rtol are from
|
|
# https://github.com/pytorch/pytorch/blob/6d96beb6bec24d73ee3f080bac54d2104068f675/test/test_transformers.py#L67
|
|
default_atol = {torch.float16: 1e-3, torch.bfloat16: 1e-3, torch.float: 1e-5}
|
|
default_rtol = {torch.float16: 1e-3, torch.bfloat16: 1.6e-2, torch.float: 1.3e-6}
|
|
|
|
|
|
def get_default_atol(output) -> float:
|
|
return default_atol[output.dtype]
|
|
|
|
|
|
def get_default_rtol(output) -> float:
|
|
return default_rtol[output.dtype]
|