github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-04-28 19:55:29 +00:00
Code Issues Packages Projects Releases Wiki Activity

[Colossal-LLaMA] Fix sft issue for llama2 (#5719)

Browse Source 
* fix minor issue

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
This commit is contained in:
Tong Li 2024-05-15 10:52:11 +08:00 committed by GitHub
parent 43995ee436
commit 913c920ecc
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
applications/Colossal-LLaMA/prepare_sft_dataset.py
View File

@ -10,7 +10,7 @@ import math
import os import os
from multiprocessing import cpu_count from multiprocessing import cpu_count
from colossal_llama.dataset.conversation import default_conversation from colossal_llama.dataset.conversation import LLaMA2_Conv
from colossal_llama.dataset.spliced_and_tokenized_dataset import supervised_tokenize_sft from colossal_llama.dataset.spliced_and_tokenized_dataset import supervised_tokenize_sft
from datasets import dataset_dict, load_dataset from datasets import dataset_dict, load_dataset
from transformers import AddedToken, AutoTokenizer from transformers import AddedToken, AutoTokenizer
@ -78,6 +78,7 @@ def main():
# Fix </s> split issue: https://github.com/huggingface/transformers/issues/23833 # Fix </s> split issue: https://github.com/huggingface/transformers/issues/23833
if args.llama_version == 2: if args.llama_version == 2:
tokenizer.add_tokens(AddedToken("</s>", normalized=False, special=True), special_tokens=True) tokenizer.add_tokens(AddedToken("</s>", normalized=False, special=True), special_tokens=True)
default_conversation = LLaMA2_Conv
tokenizer.add_bos_token = False tokenizer.add_bos_token = False
tokenizer.add_eos_token = False tokenizer.add_eos_token = False