[17:28:39] [Model]: Loading declare-lab/flan-alpaca-large...

[Sosycs]

Thanks for the great work, I am trying to implement the work of this paper on google colab with 166 G disk and T4. but at the training stage for both rationale generation and answer inference I got the output:

2023-09-29 17:27:49.955571: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
args Namespace(data_root='/content/mm-cot/data', output_dir='/content/mm-cot/experiments', model='declare-lab/flan-alpaca-large', options=['A', 'B', 'C', 'D', 'E'], epoch=50, lr=5e-05, bs=2, input_len=512, output_len=512, eval_bs=4, eval_acc=None, train_split='train', val_split='val', test_split='test', use_generate=True, final_eval=False, user_msg='rationale', img_type='vit', eval_le=None, test_le=None, evaluate_dir=None, caption_file='data/instruct_captions.json', use_caption=True, prompt_format='QCM-E', seed=42)
====Input Arguments====
{
  "data_root": "/content/mm-cot/data",
  "output_dir": "/content/mm-cot/experiments",
  "model": "declare-lab/flan-alpaca-large",
  "options": [
    "A",
    "B",
    "C",
    "D",
    "E"
  ],
  "epoch": 50,
  "lr": 5e-05,
  "bs": 2,
  "input_len": 512,
  "output_len": 512,
  "eval_bs": 4,
  "eval_acc": null,
  "train_split": "train",
  "val_split": "val",
  "test_split": "test",
  "use_generate": true,
  "final_eval": false,
  "user_msg": "rationale",
  "img_type": "vit",
  "eval_le": null,
  "test_le": null,
  "evaluate_dir": null,
  "caption_file": "data/instruct_captions.json",
  "use_caption": true,
  "prompt_format": "QCM-E",
  "seed": 42
}
img_features size:  torch.Size([11208, 145, 1024])
number of train problems: 12726

number of val problems: 4241

number of test problems: 4241

[17:28:39] [Model]: Loading declare-lab/flan-alpaca-large... 

and the cell stop and the expermint folder is empty. can anyone explain what is the problem for me? (I am still a new learner in the field)

[cooelf]

Hi, did you try to conduct a unit test to see if it is possible to load a pre-trained model using huggingface?

My guess is that the memory is not enough for loading the model.

from transformers import T5ForConditionalGeneration

# you may also try to change "declare-lab/flan-alpaca-large" to "declare-lab/flan-alpaca-base" to see if it goes well. model = T5ForConditionalGeneration.from_pretrained("declare-lab/flan-alpaca-large")

[Sosycs]

after a unit test by loading the model from huggingface.

(…)an-alpaca-large/resolve/main/config.json: 100%
787/787 [00:00<00:00, 56.3kB/s]
model.safetensors: 100%
3.13G/3.13G [00:16<00:00, 261MB/s]
(…)arge/resolve/main/generation_config.json: 100%
142/142 [00:00<00:00, 13.6kB/s]

I have changed the model from large to base but I have encountered the same:

2023-10-17 16:48:58.529434: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
args Namespace(data_root='/content/mm-cot/data', output_dir='/content/mm-cot/experiments', model='declare-lab/flan-alpaca-base', options=['A', 'B', 'C', 'D', 'E'], epoch=50, lr=5e-05, bs=2, input_len=512, output_len=512, eval_bs=4, eval_acc=None, train_split='train', val_split='val', test_split='test', use_generate=True, final_eval=False, user_msg='rationale', img_type='vit', eval_le=None, test_le=None, evaluate_dir=None, caption_file='data/instruct_captions.json', use_caption=True, prompt_format='QCM-E', seed=42)
====Input Arguments====
{
  "data_root": "/content/mm-cot/data",
  "output_dir": "/content/mm-cot/experiments",
  "model": "declare-lab/flan-alpaca-base",
  "options": [
    "A",
    "B",
    "C",
    "D",
    "E"
  ],
  "epoch": 50,
  "lr": 5e-05,
  "bs": 2,
  "input_len": 512,
  "output_len": 512,
  "eval_bs": 4,
  "eval_acc": null,
  "train_split": "train",
  "val_split": "val",
  "test_split": "test",
  "use_generate": true,
  "final_eval": false,
  "user_msg": "rationale",
  "img_type": "vit",
  "eval_le": null,
  "test_le": null,
  "evaluate_dir": null,
  "caption_file": "data/instruct_captions.json",
  "use_caption": true,
  "prompt_format": "QCM-E",
  "seed": 42
}
img_features size:  torch.Size([11208, 145, 1024])
number of train problems: 12726

number of val problems: 4241

number of test problems: 4241

[16:49:05] [Model]: Loading declare-lab/flan-alpaca-base...    

I am using google colab T4 with high RAM

[cooelf]

The hanging may also be reasonable as the main process could be handling the data after loading the model (there is no signal for indicating the completion of model loading).