github/langchain
1
0
Fork 0
mirror of https://github.com/hwchase17/langchain.git synced 2025-06-04 14:13:29 +00:00
Code Issues Packages Projects Releases Wiki Activity

Harrison/hf lru (#10154)

Browse Source 
Co-authored-by: Pascal Bro <git@pascalbrokmeier.de>
Co-authored-by: Bagatur <baskaryan@gmail.com>
This commit is contained in:
Harrison Chase 2023-09-03 15:39:25 -07:00 committed by GitHub
parent 4765c09703
commit 794ff2dae8
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 10 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
libs/langchain/langchain/schema/language_model.py
View File

@ -1,6 +1,7 @@
from __future__ import annotations from __future__ import annotations
from abc import ABC, abstractmethod from abc import ABC, abstractmethod
from functools import lru_cache
from typing import ( from typing import (
TYPE_CHECKING, TYPE_CHECKING,
Any, Any,
@ -23,10 +24,8 @@ if TYPE_CHECKING:
from langchain.callbacks.manager import Callbacks from langchain.callbacks.manager import Callbacks
def _get_token_ids_default_method(text: str) -> List[int]: @lru_cache(maxsize=None) # Cache the tokenizer
"""Encode the text into token IDs.""" def get_tokenizer() -> Any:
# TODO: this method may not be exact.
# TODO: this method may differ based on model (eg codex).
try: try:
from transformers import GPT2TokenizerFast from transformers import GPT2TokenizerFast
except ImportError: except ImportError:
@ -36,7 +35,13 @@ def _get_token_ids_default_method(text: str) -> List[int]:
"Please install it with `pip install transformers`." "Please install it with `pip install transformers`."
) )
# create a GPT-2 tokenizer instance # create a GPT-2 tokenizer instance
tokenizer = GPT2TokenizerFast.from_pretrained("gpt2") return GPT2TokenizerFast.from_pretrained("gpt2")
def _get_token_ids_default_method(text: str) -> List[int]:
"""Encode the text into token IDs."""
# get the cached tokenizer
tokenizer = get_tokenizer()
# tokenize the text using the GPT-2 tokenizer # tokenize the text using the GPT-2 tokenizer
return tokenizer.encode(text) return tokenizer.encode(text)