riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-19 12:38:23 +06:00
Code Issues Actions 21 Packages Projects Releases Wiki Activity

Add resume checkpoint support to ClearML callback #37502

Browse Source
This commit is contained in:
Minjun Zhu 2025-04-20 02:37:34 -04:00
parent 861917173f
commit 82087b5722
1 changed files with 11 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
tests/test_clearml_resume.py
View File

@ -1,15 +1,17 @@
import os import os
import torch
from clearml import Task
from datasets import load_dataset
from transformers import ( from transformers import (
AutoModelForSequenceClassification, AutoModelForSequenceClassification,
AutoTokenizer, AutoTokenizer,
TrainingArguments,
Trainer,
DataCollatorWithPadding, DataCollatorWithPadding,
Trainer,
TrainingArguments,
) )
from datasets import load_dataset
from clearml import Task
import torch
import shutil
# Set environment variables # Set environment variables
os.environ["CLEARML_PROJECT"] = "Test Project" os.environ["CLEARML_PROJECT"] = "Test Project"
@ -17,11 +19,7 @@ os.environ["CLEARML_TASK"] = "Test Task"
os.environ["CLEARML_LOG_MODEL"] = "TRUE" os.environ["CLEARML_LOG_MODEL"] = "TRUE"
# Initialize ClearML task # Initialize ClearML task
task = Task.init( task = Task.init(project_name="Test Project", task_name="Test Task", reuse_last_task_id=False)
project_name="Test Project",
task_name="Test Task",
reuse_last_task_id=False
)
# Load model and tokenizer # Load model and tokenizer
model_name = "bert-base-uncased" model_name = "bert-base-uncased"
@ -95,5 +93,5 @@ trainer.train()
initial_params = {name: param.data.clone() for name, param in model.named_parameters()} initial_params = {name: param.data.clone() for name, param in model.named_parameters()}
trainer.train() trainer.train()
for name, param in model.named_parameters(): for name, param in model.named_parameters():
if 'weight' in name: if "weight" in name:
diff = torch.abs(param.data - initial_params[name]).mean().item() diff = torch.abs(param.data - initial_params[name]).mean().item()