Skip to content

Instantly share code, notes, and snippets.

View sungkim11's full-sized avatar

Sung Kim sungkim11

View GitHub Profile
@Helw150
Helw150 / parallel_t5.py
Last active May 10, 2023 14:52
Flan T5 Parallel Usage
from transformers import AutoTokenizer, T5ForConditionalGeneration
# Model Init
n_gpu = 8
tokenizer = AutoTokenizer.from_pretrained("google/flan-ul2")
model = T5ForConditionalGeneration.from_pretrained("google/flan-ul2")
heads_per_gpu = len(model.encoder.block) // n_gpu
device_map = {
gpu: list(
range(