Mirrors/modelscope
1
0
Fork 0
mirror of https://github.com/modelscope/modelscope.git synced 2025-12-24 20:19:22 +01:00
Code Issues Packages Projects Releases Wiki Activity

mgeo fix finetune for rerank test case and reduce UT time

Browse Source 
* reduce UT time 
* fix finetune for rerank test case

Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/11563740
This commit is contained in:
ada.drx
2023-02-07 02:55:33 +00:00
committed by wenmeng.zwm
parent 1d2b7e8634
commit 7298bd2bb4
2 changed files with 11 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
modelscope/trainers/multi_modal/mgeo_ranking_trainer.py
View File

@@ -290,6 +290,8 @@ class MGeoRankingTrainer(NlpEpochBasedTrainer):
label_list.extend(label_ids)
logits_list.extend(logits)
qid_list.extend(qids)
if _step + 1 > self._eval_iters_per_epoch:
break
logger.info('Inference time = {:.2f}s, [{:.4f} ms / sample] '.format(
total_spent_time, total_spent_time * 1000 / total_samples))

19
tests/trainers/test_finetune_mgeo.py
View File

@@ -50,7 +50,7 @@ class TestFinetuneMGeo(unittest.TestCase):
results_files = os.listdir(self.tmp_dir)
self.assertIn(f'{trainer.timestamp}.log.json', results_files)
@unittest.skipUnless(test_level() >= 4, 'skip test in current test level')
@unittest.skipUnless(test_level() >= 0, 'skip test in current test level')
def test_finetune_geotes_rerank(self):
def cfg_modify_fn(cfg):
@@ -82,7 +82,8 @@ class TestFinetuneMGeo(unittest.TestCase):
cfg.train.dataloader.batch_size_per_gpu = 3
cfg.train.dataloader.workers_per_gpu = 16
cfg.evaluation.dataloader.workers_per_gpu = 16
cfg.train.train_iters_per_epoch = 10
cfg.evaluation.val_iters_per_epoch = 10
cfg['evaluation']['metrics'] = 'mrr@1'
cfg.train.max_epochs = 1
cfg.model['neg_sample'] = neg_sample
@@ -139,14 +140,8 @@ class TestFinetuneMGeo(unittest.TestCase):
split='validation',
namespace='damo')
dataset = MsDataset.load(
'json',
data_files={
'train': [train_dataset['train'] + '/train.json'],
'test': [dev_dataset['validation'] + '/dev.json']
})
train_ds = dataset['train'].to_hf_dataset()
dev_ds = dataset['test'].to_hf_dataset()
train_ds = train_dataset['train']
dev_ds = dev_dataset['validation']
model_id = 'damo/mgeo_backbone_chinese_base'
self.finetune(
@@ -170,6 +165,8 @@ class TestFinetuneMGeo(unittest.TestCase):
cfg.evaluation.dataloader.batch_size_per_gpu = 64
cfg.train.optimizer.lr = 2e-5
cfg.train.max_epochs = 1
cfg.train.train_iters_per_epoch = 10
cfg.evaluation.val_iters_per_epoch = 10
cfg['dataset'] = {
'train': {
@@ -236,6 +233,8 @@ class TestFinetuneMGeo(unittest.TestCase):
}
cfg.train.max_epochs = 1
cfg.train.dataloader.batch_size_per_gpu = 32
cfg.train.train_iters_per_epoch = 10
cfg.evaluation.val_iters_per_epoch = 10
cfg.train.optimizer.lr = 3e-5
cfg.train.hooks = [{
'type': 'CheckpointHook',