mirror of
https://github.com/huggingface/text-generation-inference.git
synced 2025-04-23 16:02:10 +00:00
e6d3eb5d5d
- Fix some type hints, in particular base tokenizer class - Make use of `tensor.new_zero/empty` methods - Simplify env var string parsing in launcher
25 lines
659 B
Python
25 lines
659 B
Python
import torch
|
|
|
|
from abc import ABC, abstractmethod
|
|
from typing import List, Tuple, Optional, TypeVar, Type
|
|
from transformers import PreTrainedTokenizerBase
|
|
|
|
from text_generation.models.types import Batch, GeneratedText
|
|
|
|
B = TypeVar("B", bound=Batch)
|
|
|
|
|
|
class Model(ABC):
|
|
def __init__(self, tokenizer: PreTrainedTokenizerBase, device: torch.device):
|
|
self.tokenizer = tokenizer
|
|
self.device = device
|
|
|
|
@property
|
|
@abstractmethod
|
|
def batch_type(self) -> Type[B]:
|
|
raise NotImplementedError
|
|
|
|
@abstractmethod
|
|
def generate_token(self, batch: B) -> Tuple[List[GeneratedText], Optional[B]]:
|
|
raise NotImplementedError
|