Skip to content

GitLab

"tools/pre_commit/update-dockerfile-graph.sh" did not exist on "ce8d6b75fc0586045df75ee1568a5b5f9957251b"
Unverified Commit 02c9afa2 authored by Roger Wang's avatar Roger Wang Committed by GitHub
Browse files

Revert "[Misc][Bugfix] Disable guided decoding for mistral tokenizer" (#8593)

parent 3118f633
Show whitespace changes
Inline Side-by-side
Showing with 0 additions and 23 deletions
vllm/model_executor/guided_decoding/__init__.py
View file @ 02c9afa2
......@@ -6,7 +6,6 @@ from vllm.entrypoints.openai.protocol import (
from vllm.model_executor.guided_decoding.guided_fields import (
GuidedDecodingRequest)
from vllm.sampling_params import LogitsProcessor
from vllm.transformers_utils.tokenizer import MistralTokenizer
async def get_guided_decoding_logits_processor(
......@@ -16,23 +15,12 @@ async def get_guided_decoding_logits_processor(
request = _adapt_request_for_tool_use(request)
if guided_decoding_backend == 'outlines':
if isinstance(tokenizer, MistralTokenizer):
raise NotImplementedError(
"Guided decoding with 'outlines' is currently not supported "
"for Mistral tokenizer. Please consider contributing to the "
"'outlines' project if you are interested in this feature.")
# NOTE: lazy import outlines to avoid https://github.com/vllm-project/vllm/issues/4193
from vllm.model_executor.guided_decoding.outlines_decoding import ( # noqa
get_outlines_guided_decoding_logits_processor)
return await get_outlines_guided_decoding_logits_processor(
request, tokenizer)
if guided_decoding_backend == 'lm-format-enforcer':
if isinstance(tokenizer, MistralTokenizer):
raise NotImplementedError(
"Guided decoding with 'lm-format-enforcer' is currently not "
"supported for Mistral tokenizer. Please consider contributing "
"to the 'lm-format-enforcer' project if you are interested "
"in this feature.")
from vllm.model_executor.guided_decoding.lm_format_enforcer_decoding import ( # noqa
get_lm_format_enforcer_guided_decoding_logits_processor)
return await get_lm_format_enforcer_guided_decoding_logits_processor(
......@@ -49,23 +37,12 @@ def get_local_guided_decoding_logits_processor(
# request = _adapt_request_for_tool_use(request)
if guided_decoding_backend == 'outlines':
if isinstance(tokenizer, MistralTokenizer):
raise NotImplementedError(
"Guided decoding with 'outlines' is currently not supported "
"for Mistral tokenizer. Please consider contributing to the "
"'outlines' project if you are interested in this feature.")
# NOTE: lazy import outlines to avoid https://github.com/vllm-project/vllm/issues/4193
from vllm.model_executor.guided_decoding.outlines_decoding import ( # noqa
get_local_outlines_guided_decoding_logits_processor)
return get_local_outlines_guided_decoding_logits_processor(
guided_options, tokenizer)
if guided_decoding_backend == 'lm-format-enforcer':
if isinstance(tokenizer, MistralTokenizer):
raise NotImplementedError(
"Guided decoding with 'lm-format-enforcer' is currently not "
"supported for Mistral tokenizer. Please consider contributing "
"to the 'lm-format-enforcer' project if you are interested "
"in this feature.")
from vllm.model_executor.guided_decoding.lm_format_enforcer_decoding import ( # noqa
get_local_lm_format_enforcer_guided_decoding_logits_processor)
return get_local_lm_format_enforcer_guided_decoding_logits_processor(
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment