xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-13 13:46:51 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Remove unused encoder-decoder error strings (#25374)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Cyrus Leung 2025-09-22 19:04:32 +08:00 committed by yewentao256
parent cbba9bd0b0
commit ddf4e1f56f
2 changed files with 0 additions and 63 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/attention/backends/utils.py
View File

@ -18,11 +18,6 @@ from vllm.utils import async_tensor_h2d, make_tensor_with_pad
logger = init_logger(__name__) logger = init_logger(__name__)
# Error string(s) for encoder/decoder
# unsupported attention scenarios
STR_NOT_IMPL_ENC_DEC_ROCM_HIP = ("ROCm/HIP is not currently supported "
"with encoder/decoder models.")
PAD_SLOT_ID = -1 PAD_SLOT_ID = -1
# Switch to numpy implementation of compute_slot_mapping # Switch to numpy implementation of compute_slot_mapping

58
vllm/utils/__init__.py
View File

@ -88,64 +88,6 @@ DEFAULT_MAX_NUM_BATCHED_TOKENS = 2048
POOLING_MODEL_MAX_NUM_BATCHED_TOKENS = 32768 POOLING_MODEL_MAX_NUM_BATCHED_TOKENS = 32768
MULTIMODAL_MODEL_MAX_NUM_BATCHED_TOKENS = 5120 MULTIMODAL_MODEL_MAX_NUM_BATCHED_TOKENS = 5120
# Exception strings for non-implemented encoder/decoder scenarios
# Reminder: Please update docs/features/compatibility_matrix.md
# If the feature combo become valid
STR_NOT_IMPL_ENC_DEC_SWA = \
"Sliding window attention for encoder/decoder models " + \
"is not currently supported."
STR_NOT_IMPL_ENC_DEC_PREFIX_CACHE = \
"Prefix caching for encoder/decoder models " + \
"is not currently supported."
STR_NOT_IMPL_ENC_DEC_CHUNKED_PREFILL = \
"Chunked prefill for encoder/decoder models " + \
"is not currently supported."
STR_NOT_IMPL_ENC_DEC_LOGIT_SOFTCAP = (
"Models with logits_soft_cap "
"require FlashInfer backend, which is "
"currently not supported for encoder/decoder "
"models.")
STR_NOT_IMPL_ENC_DEC_LORA = ("LoRA is not currently "
"supported with encoder/decoder "
"models.")
STR_NOT_IMPL_ENC_DEC_PP = ("Pipeline parallelism is not "
"currently supported with "
"encoder/decoder models.")
STR_NOT_IMPL_ENC_DEC_MM = ("Multimodal is not currently "
"supported with encoder/decoder "
"models.")
STR_NOT_IMPL_ENC_DEC_SPEC_DEC = ("Speculative decoding is not "
"currently supported with encoder/"
"decoder models.")
STR_NOT_IMPL_ENC_DEC_BACKEND = ("XFormers and Flash-Attention are the only "
"backends currently supported with encoder/"
"decoder models.")
# Efficiently import all enc/dec error strings
# rather than having to import all of the above
STR_NOT_IMPL_ENC_DEC_ERR_STRS = {
"STR_NOT_IMPL_ENC_DEC_SWA": STR_NOT_IMPL_ENC_DEC_SWA,
"STR_NOT_IMPL_ENC_DEC_PREFIX_CACHE": STR_NOT_IMPL_ENC_DEC_PREFIX_CACHE,
"STR_NOT_IMPL_ENC_DEC_CHUNKED_PREFILL":
STR_NOT_IMPL_ENC_DEC_CHUNKED_PREFILL,
"STR_NOT_IMPL_ENC_DEC_LOGIT_SOFTCAP": STR_NOT_IMPL_ENC_DEC_LOGIT_SOFTCAP,
"STR_NOT_IMPL_ENC_DEC_LORA": STR_NOT_IMPL_ENC_DEC_LORA,
"STR_NOT_IMPL_ENC_DEC_PP": STR_NOT_IMPL_ENC_DEC_PP,
"STR_NOT_IMPL_ENC_DEC_MM": STR_NOT_IMPL_ENC_DEC_MM,
"STR_NOT_IMPL_ENC_DEC_SPEC_DEC": STR_NOT_IMPL_ENC_DEC_SPEC_DEC,
"STR_NOT_IMPL_ENC_DEC_BACKEND": STR_NOT_IMPL_ENC_DEC_BACKEND,
}
# Constants related to forcing the attention backend selection # Constants related to forcing the attention backend selection
# String name of register which may be set in order to # String name of register which may be set in order to