[mypy] Enable following imports for some directories (#6681)

This commit is contained in:
Cyrus Leung
2024-07-31 10:38:03 +08:00
committed by GitHub
parent c32ab8be1a
commit da1f7cc12a
18 changed files with 185 additions and 143 deletions
@@ -1,16 +1,14 @@
from typing import List, Optional
from transformers import PreTrainedTokenizer
from vllm.config import TokenizerPoolConfig
from vllm.lora.request import LoRARequest
from vllm.transformers_utils.tokenizer import (get_lora_tokenizer,
get_lora_tokenizer_async,
get_tokenizer)
from vllm.transformers_utils.tokenizer_group.base_tokenizer_group import (
BaseTokenizerGroup)
from vllm.utils import LRUCache
from .base_tokenizer_group import AnyTokenizer, BaseTokenizerGroup
class TokenizerGroup(BaseTokenizerGroup):
"""A group of tokenizers that can be used for LoRA adapters."""
@@ -22,8 +20,8 @@ class TokenizerGroup(BaseTokenizerGroup):
self.enable_lora = enable_lora
self.max_input_length = max_input_length
self.tokenizer = get_tokenizer(self.tokenizer_id, **tokenizer_config)
self.lora_tokenizers = LRUCache[PreTrainedTokenizer](
capacity=max_num_seqs) if enable_lora else None
self.lora_tokenizers = LRUCache[AnyTokenizer](
capacity=max_num_seqs if enable_lora else 0)
@classmethod
def from_config(cls, tokenizer_pool_config: Optional[TokenizerPoolConfig],
@@ -41,7 +39,7 @@ class TokenizerGroup(BaseTokenizerGroup):
return self.max_input_length
def _raise_if_input_too_long(self,
encoded_tokens: List[str],
encoded_tokens: List[int],
lora_request: Optional[LoRARequest] = None):
input_length = len(encoded_tokens)
if lora_request:
@@ -72,9 +70,9 @@ class TokenizerGroup(BaseTokenizerGroup):
return ret
def get_lora_tokenizer(
self,
lora_request: Optional[LoRARequest] = None
) -> "PreTrainedTokenizer":
self,
lora_request: Optional[LoRARequest] = None,
) -> AnyTokenizer:
if not lora_request or not self.enable_lora:
return self.tokenizer
if lora_request.lora_int_id not in self.lora_tokenizers:
@@ -83,12 +81,12 @@ class TokenizerGroup(BaseTokenizerGroup):
self.lora_tokenizers.put(lora_request.lora_int_id, tokenizer)
return tokenizer
else:
return self.lora_tokenizers.get(lora_request.lora_int_id)
return self.lora_tokenizers[lora_request.lora_int_id]
async def get_lora_tokenizer_async(
self,
lora_request: Optional[LoRARequest] = None
) -> "PreTrainedTokenizer":
self,
lora_request: Optional[LoRARequest] = None,
) -> AnyTokenizer:
if not lora_request or not self.enable_lora:
return self.tokenizer
if lora_request.lora_int_id not in self.lora_tokenizers:
@@ -97,4 +95,4 @@ class TokenizerGroup(BaseTokenizerGroup):
self.lora_tokenizers.put(lora_request.lora_int_id, tokenizer)
return tokenizer
else:
return self.lora_tokenizers.get(lora_request.lora_int_id)
return self.lora_tokenizers[lora_request.lora_int_id]