Spaces:
Running
Running
File size: 1,740 Bytes
30ffb9e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 |
#%%
import os, time, io, zipfile
from preprocessing import FileIO
import shutil
import modal
from llama_index.finetuning import EmbeddingQAFinetuneDataset
from dotenv import load_dotenv, find_dotenv
env = load_dotenv(find_dotenv('env'), override=True)
#%%
training_path = 'data/training_data_300.json'
valid_path = 'data/validation_data_100.json'
training_set = EmbeddingQAFinetuneDataset.from_json(training_path)
valid_set = EmbeddingQAFinetuneDataset.from_json(valid_path)
def finetune(model='all-mpnet-base-v2', savemodel=False, outpath='.'):
""" Finetunes a model on Modal GPU A100.
The model is saved in /root/models on a Modal volume
and can be stored locally.
Args:
model (str): the Sentence Transformer model name
savemodel (bool, optional): whether to save the model or not.
Returns:
path of the saved model (when saved)
"""
f = modal.Function.lookup("vector-search-project", "finetune")
model = model.replace('/','')
if 'sentence-transformers' not in model:
model = f"sentence-transformers/{model}"
fullpath = os.path.join(outpath, f"finetuned-{model}-300")
if os.path.exists(fullpath):
msg = "Model already exists!"
print(msg)
return msg
start = time.perf_counter()
finetuned_model = f.remote(training_path, valid_path, model_id=model)
end = time.perf_counter() - start
print(f"Finetuning with GPU lasted {end:.2f} seconds")
if savemodel:
with open(fullpath, 'wb') as file:
# Write the contents of the BytesIO object to a new file
file.write(finetuned_model.getbuffer())
print(f"Model saved in {fullpath}")
return fullpath
|