BUG

[Dobby114]

Hello, thank you for sharing your code generously, I want to know what are the keywords_gpt_total.json and keywords_gpt_total.json . And I always get accept the following error when I run the code ' python cli.py train with model=no_gt_sos fix_gpt_epoch=5 feature_names="['video', 'images']" '

ERROR - lsmdc - Failed after 0:04:15! Traceback (most recent calls WITHOUT Sacred internals): File "/root/TAPM-0/tapm-master/code/exp.py", line 48, in captured_function return lazy_capture(*args, kwargs) File "/root/TAPM-0/tapm-master/code/exp.py", line 40, in call return self.lazy_fn(*args, *kwargs) File "/root/TAPM-0/tapm-master/code/cli.py", line 48, in train all_args = prepare() File "/root/TAPM-0/tapm-master/code/cli.py", line 31, in prepare model, tokenizer, ckpt, datasets, epoch = prepare_model() File "/root/TAPM-0/tapm-master/code/exp.py", line 48, in captured_function return lazy_capture(args, kwargs) File "/root/TAPM-0/tapm-master/code/exp.py", line 40, in call return self.lazy_fn(*args, kwargs) File "/root/TAPM-0/tapm-master/code/cli.py", line 25, in prepare_model return get_model_ckpt(model_name) File "/root/TAPM-0/tapm-master/code/exp.py", line 48, in captured_function return lazy_capture(*args, *kwargs) File "/root/TAPM-0/tapm-master/code/exp.py", line 40, in call return self.lazy_fn(args, kwargs) File "/root/TAPM-0/tapm-master/code/ckpt.py", line 63, in get_model_ckpt datasets = get_datasets(data_path, pretrain_path) File "/root/TAPM-0/tapm-master/code/exp.py", line 48, in captured_function return lazy_capture(*args, kwargs) File "/root/TAPM-0/tapm-master/code/exp.py", line 40, in call return self.lazy_fn(*args, *kwargs) File "/root/TAPM-0/tapm-master/code/data/dataloader.py", line 71, in get_datasets datasets[k] = Dataset(p) File "/root/TAPM-0/tapm-master/code/data/dataloader.py", line 17, in init self.task, self.path = load_tasks(data_path) File "/root/TAPM-0/tapm-master/code/exp.py", line 48, in captured_function return lazy_capture(args, kwargs) File "/root/TAPM-0/tapm-master/code/exp.py", line 40, in call return self.lazy_fn(*args, kwargs) File "/root/TAPM-0/tapm-master/code/data/task_loaders.py", line 30, in load_tasks data, global_data = f(path) File "/root/TAPM-0/tapm-master/code/exp.py", line 48, in captured_function return lazy_capture(*args, *kwargs) File "/root/TAPM-0/tapm-master/code/exp.py", line 40, in call return self.lazy_fn(args, kwargs) File "/root/TAPM-0/tapm-master/code/data/task_loaders.py", line 39, in load_task1 return load_task1_group_with_features_and_keyword(path) File "/root/TAPM-0/tapm-master/code/data/task_loaders.py", line 158, in load_task1_group_with_features_and_keyword total = load_keyword_only(path, name='keywords_gpt_total.json') File "/root/TAPM-0/tapm-master/code/exp.py", line 48, in captured_function return lazy_capture(*args, *kwargs) File "/root/TAPM-0/tapm-master/code/exp.py", line 40, in call return self.lazy_fn(args, **kwargs) File "/root/TAPM-0/tapm-master/code/data/task_loaders.py", line 378, in load_keyword_only assert len(paths) > 0, f"no keyword candidate for {name}"
AssertionError: no keyword candidate for keywords_gpt_total.json

[JiwanChung]

First of all, thank you for notifying the issue. Keyword is now a deprecated feature, so in theory the code should run fine without it. I suspect a bug in the code, and currently inspecting the cause. Thanks.

[JiwanChung]

Update: Simply changing 'keyword_name' field in config.py to None fixed this issue. I also uploaded a new commit of this fix.