[Misc] Upgrade to Python 3.9 typing for additional directories (#14492)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-03-08 17:35:50 +00:00
committed by GitHub
parent 0d5e73d30e
commit 9513290032
12 changed files with 78 additions and 78 deletions
+27 -27
View File
@@ -1,9 +1,9 @@
# SPDX-License-Identifier: Apache-2.0
from collections.abc import Iterable
from dataclasses import dataclass
from functools import cached_property
from typing import (TYPE_CHECKING, Any, Dict, Generic, Iterable, List, Literal,
Optional, Tuple, Union, cast)
from typing import TYPE_CHECKING, Any, Generic, Literal, Optional, Union, cast
import torch
from typing_extensions import NotRequired, TypedDict, TypeVar, assert_never
@@ -26,7 +26,7 @@ class TextPrompt(TypedDict):
if the model supports it.
"""
mm_processor_kwargs: NotRequired[Dict[str, Any]]
mm_processor_kwargs: NotRequired[dict[str, Any]]
"""
Optional multi-modal processor kwargs to be forwarded to the
multimodal input mapper & processor. Note that if multiple modalities
@@ -38,10 +38,10 @@ class TextPrompt(TypedDict):
class TokensPrompt(TypedDict):
"""Schema for a tokenized prompt."""
prompt_token_ids: List[int]
prompt_token_ids: list[int]
"""A list of token IDs to pass to the model."""
token_type_ids: NotRequired[List[int]]
token_type_ids: NotRequired[list[int]]
"""A list of token type IDs to pass to the cross encoder model."""
multi_modal_data: NotRequired["MultiModalDataDict"]
@@ -50,7 +50,7 @@ class TokensPrompt(TypedDict):
if the model supports it.
"""
mm_processor_kwargs: NotRequired[Dict[str, Any]]
mm_processor_kwargs: NotRequired[dict[str, Any]]
"""
Optional multi-modal processor kwargs to be forwarded to the
multimodal input mapper & processor. Note that if multiple modalities
@@ -115,7 +115,7 @@ class ExplicitEncoderDecoderPrompt(TypedDict, Generic[_T1_co, _T2_co]):
decoder_prompt: Optional[_T2_co]
mm_processor_kwargs: NotRequired[Dict[str, Any]]
mm_processor_kwargs: NotRequired[dict[str, Any]]
PromptType = Union[SingletonPrompt, ExplicitEncoderDecoderPrompt]
@@ -136,10 +136,10 @@ class TokenInputs(TypedDict):
type: Literal["token"]
"""The type of inputs."""
prompt_token_ids: List[int]
prompt_token_ids: list[int]
"""The token IDs of the prompt."""
token_type_ids: NotRequired[List[int]]
token_type_ids: NotRequired[list[int]]
"""The token type IDs of the prompt."""
prompt: NotRequired[str]
@@ -164,12 +164,12 @@ class TokenInputs(TypedDict):
Placeholder ranges for the multi-modal data.
"""
multi_modal_hashes: NotRequired[List[str]]
multi_modal_hashes: NotRequired[list[str]]
"""
The hashes of the multi-modal data.
"""
mm_processor_kwargs: NotRequired[Dict[str, Any]]
mm_processor_kwargs: NotRequired[dict[str, Any]]
"""
Optional multi-modal processor kwargs to be forwarded to the
multimodal input mapper & processor. Note that if multiple modalities
@@ -179,14 +179,14 @@ class TokenInputs(TypedDict):
def token_inputs(
prompt_token_ids: List[int],
token_type_ids: Optional[List[int]] = None,
prompt_token_ids: list[int],
token_type_ids: Optional[list[int]] = None,
prompt: Optional[str] = None,
multi_modal_data: Optional["MultiModalDataDict"] = None,
multi_modal_inputs: Optional["MultiModalKwargs"] = None,
multi_modal_hashes: Optional[List[str]] = None,
multi_modal_hashes: Optional[list[str]] = None,
multi_modal_placeholders: Optional["MultiModalPlaceholderDict"] = None,
mm_processor_kwargs: Optional[Dict[str, Any]] = None,
mm_processor_kwargs: Optional[dict[str, Any]] = None,
) -> TokenInputs:
"""Construct :class:`TokenInputs` from optional values."""
inputs = TokenInputs(type="token", prompt_token_ids=prompt_token_ids)
@@ -255,7 +255,7 @@ class SingletonInputsAdapter:
assert_never(inputs) # type: ignore[arg-type]
@cached_property
def prompt_token_ids(self) -> List[int]:
def prompt_token_ids(self) -> list[int]:
inputs = self.inputs
if inputs["type"] == "token" or inputs["type"] == "multimodal":
@@ -264,7 +264,7 @@ class SingletonInputsAdapter:
assert_never(inputs) # type: ignore[arg-type]
@cached_property
def token_type_ids(self) -> List[int]:
def token_type_ids(self) -> list[int]:
inputs = self.inputs
if inputs["type"] == "token" or inputs["type"] == "multimodal":
@@ -294,7 +294,7 @@ class SingletonInputsAdapter:
assert_never(inputs) # type: ignore[arg-type]
@cached_property
def multi_modal_inputs(self) -> Union[Dict, "MultiModalKwargs"]:
def multi_modal_inputs(self) -> Union[dict, "MultiModalKwargs"]:
inputs = self.inputs
if inputs["type"] == "token":
@@ -306,7 +306,7 @@ class SingletonInputsAdapter:
assert_never(inputs) # type: ignore[arg-type]
@cached_property
def multi_modal_hashes(self) -> List[str]:
def multi_modal_hashes(self) -> list[str]:
inputs = self.inputs
if inputs["type"] == "token":
@@ -331,7 +331,7 @@ class SingletonInputsAdapter:
assert_never(inputs) # type: ignore[arg-type]
@cached_property
def mm_processor_kwargs(self) -> Dict[str, Any]:
def mm_processor_kwargs(self) -> dict[str, Any]:
inputs = self.inputs
if inputs["type"] == "token":
@@ -355,7 +355,7 @@ _T2 = TypeVar("_T2", bound=SingletonPrompt, default=SingletonPrompt)
def build_explicit_enc_dec_prompt(
encoder_prompt: _T1,
decoder_prompt: Optional[_T2],
mm_processor_kwargs: Optional[Dict[str, Any]] = None,
mm_processor_kwargs: Optional[dict[str, Any]] = None,
) -> ExplicitEncoderDecoderPrompt[_T1, _T2]:
if mm_processor_kwargs is None:
mm_processor_kwargs = {}
@@ -368,9 +368,9 @@ def build_explicit_enc_dec_prompt(
def zip_enc_dec_prompts(
enc_prompts: Iterable[_T1],
dec_prompts: Iterable[Optional[_T2]],
mm_processor_kwargs: Optional[Union[Iterable[Dict[str, Any]],
Dict[str, Any]]] = None,
) -> List[ExplicitEncoderDecoderPrompt[_T1, _T2]]:
mm_processor_kwargs: Optional[Union[Iterable[dict[str, Any]],
dict[str, Any]]] = None,
) -> list[ExplicitEncoderDecoderPrompt[_T1, _T2]]:
"""
Zip encoder and decoder prompts together into a list of
:class:`ExplicitEncoderDecoderPrompt` instances.
@@ -380,12 +380,12 @@ def zip_enc_dec_prompts(
provided, it will be zipped with the encoder/decoder prompts.
"""
if mm_processor_kwargs is None:
mm_processor_kwargs = cast(Dict[str, Any], {})
mm_processor_kwargs = cast(dict[str, Any], {})
if isinstance(mm_processor_kwargs, dict):
return [
build_explicit_enc_dec_prompt(
encoder_prompt, decoder_prompt,
cast(Dict[str, Any], mm_processor_kwargs))
cast(dict[str, Any], mm_processor_kwargs))
for (encoder_prompt,
decoder_prompt) in zip(enc_prompts, dec_prompts)
]
@@ -399,7 +399,7 @@ def zip_enc_dec_prompts(
def to_enc_dec_tuple_list(
enc_dec_prompts: Iterable[ExplicitEncoderDecoderPrompt[_T1, _T2]],
) -> List[Tuple[_T1, Optional[_T2]]]:
) -> list[tuple[_T1, Optional[_T2]]]:
return [(enc_dec_prompt["encoder_prompt"],
enc_dec_prompt["decoder_prompt"])
for enc_dec_prompt in enc_dec_prompts]
+9 -8
View File
@@ -1,6 +1,7 @@
# SPDX-License-Identifier: Apache-2.0
from typing import List, Literal, Sequence, TypedDict, Union, cast, overload
from collections.abc import Sequence
from typing import Literal, TypedDict, Union, cast, overload
from typing_extensions import TypeIs
@@ -17,24 +18,24 @@ class ParsedText(TypedDict):
class ParsedTokens(TypedDict):
content: List[int]
content: list[int]
is_tokens: Literal[True]
@overload
def parse_and_batch_prompt(
prompt: Union[str, List[str]]) -> Sequence[ParsedText]:
prompt: Union[str, list[str]]) -> Sequence[ParsedText]:
...
@overload
def parse_and_batch_prompt(
prompt: Union[List[int], List[List[int]]]) -> Sequence[ParsedTokens]:
prompt: Union[list[int], list[list[int]]]) -> Sequence[ParsedTokens]:
...
def parse_and_batch_prompt(
prompt: Union[str, List[str], List[int], List[List[int]]],
prompt: Union[str, list[str], list[int], list[list[int]]],
) -> Union[Sequence[ParsedText], Sequence[ParsedTokens]]:
if isinstance(prompt, str):
# case 1: a string
@@ -46,16 +47,16 @@ def parse_and_batch_prompt(
if is_list_of(prompt, str):
# case 2: array of strings
prompt = cast(List[str], prompt)
prompt = cast(list[str], prompt)
return [
ParsedText(content=elem, is_tokens=False) for elem in prompt
]
if is_list_of(prompt, int):
# case 3: array of tokens
prompt = cast(List[int], prompt)
prompt = cast(list[int], prompt)
return [ParsedTokens(content=prompt, is_tokens=True)]
if is_list_of(prompt, list):
prompt = cast(List[List[int]], prompt)
prompt = cast(list[list[int]], prompt)
if len(prompt[0]) == 0:
raise ValueError("please provide at least one prompt")
+12 -11
View File
@@ -1,7 +1,8 @@
# SPDX-License-Identifier: Apache-2.0
import asyncio
from typing import List, Mapping, Optional, Tuple, Union, cast
from collections.abc import Mapping
from typing import Optional, Union, cast
from typing_extensions import assert_never
@@ -92,7 +93,7 @@ class InputPreprocessor:
return dec_start_token_id
def _get_default_enc_dec_decoder_prompt(self) -> List[int]:
def _get_default_enc_dec_decoder_prompt(self) -> list[int]:
'''
Specifically for encoder/decoder models:
generate a default decoder prompt for when
@@ -130,8 +131,8 @@ class InputPreprocessor:
def _prepare_decoder_input_ids_for_generation(
self,
decoder_input_ids: Optional[List[int]],
) -> List[int]:
decoder_input_ids: Optional[list[int]],
) -> list[int]:
"""
Prepares `decoder_input_ids` for generation with encoder-decoder models.
@@ -168,9 +169,9 @@ class InputPreprocessor:
def _apply_prompt_adapter(
self,
prompt_token_ids: List[int],
prompt_token_ids: list[int],
prompt_adapter_request: Optional[PromptAdapterRequest],
) -> List[int]:
) -> list[int]:
if prompt_adapter_request:
prompt_token_ids = (
[0] * prompt_adapter_request.prompt_adapter_num_virtual_tokens
@@ -183,7 +184,7 @@ class InputPreprocessor:
prompt: str,
request_id: str,
lora_request: Optional[LoRARequest],
) -> List[int]:
) -> list[int]:
"""
Apply the model's tokenizer to a text prompt, returning the
corresponding token IDs.
@@ -211,7 +212,7 @@ class InputPreprocessor:
prompt: str,
request_id: str,
lora_request: Optional[LoRARequest],
) -> List[int]:
) -> list[int]:
"""Async version of :meth:`_tokenize_prompt`."""
tokenizer = self.get_tokenizer_group()
add_special_tokens = None
@@ -250,7 +251,7 @@ class InputPreprocessor:
def _process_multimodal(
self,
prompt: Union[str, List[int]],
prompt: Union[str, list[int]],
mm_data: MultiModalDataDict,
mm_processor_kwargs: Optional[Mapping[str, object]],
lora_request: Optional[LoRARequest],
@@ -280,7 +281,7 @@ class InputPreprocessor:
async def _process_multimodal_async(
self,
prompt: Union[str, List[int]],
prompt: Union[str, list[int]],
mm_data: MultiModalDataDict,
mm_processor_kwargs: Optional[Mapping[str, object]],
lora_request: Optional[LoRARequest],
@@ -511,7 +512,7 @@ class InputPreprocessor:
self,
inputs: SingletonInputs,
decoder_inputs_to_override: Optional[SingletonInputs] = None,
) -> Tuple[SingletonInputs, SingletonInputs]:
) -> tuple[SingletonInputs, SingletonInputs]:
"""
For encoder/decoder models only:
Separate Encoder/Decoder inputs from a MultiModalEncDecInputs
+3 -2
View File
@@ -2,9 +2,10 @@
import functools
from collections import UserDict
from collections.abc import Mapping
from dataclasses import dataclass
from typing import (TYPE_CHECKING, Any, Callable, Mapping, NamedTuple,
Optional, Protocol, Union)
from typing import (TYPE_CHECKING, Any, Callable, NamedTuple, Optional,
Protocol, Union)
from torch import nn
from transformers import BatchFeature, PretrainedConfig, ProcessorMixin