-
Notifications
You must be signed in to change notification settings - Fork 316
Open
Labels
bugSomething isn't workingSomething isn't working
Description
Expected Behavior
When using the given model initialization code:
from open_flamingo import create_model_and_transforms
model, image_processor, tokenizer = create_model_and_transforms(
clip_vision_encoder_path="ViT-L-14",
clip_vision_encoder_pretrained="openai",
lang_encoder_path="anas-awadalla/mpt-7b",
tokenizer_path="anas-awadalla/mpt-7b",
cross_attn_every_n_layers=4
)
the model should be initialized
Current Behavior
Failure with error:
ValueError: We require the attribute name for the nn.ModuleList in the decoder storing the transformer block layers. Please supply this string manually.
Full error output here: https://gist.github.com/JiahuiKChen/d3695f31bcf7702bdf0c4ea86a57a77b
Steps to Reproduce
Run any of the create_model_and_transforms code snippets.
I'm getting errors for all the models. Different error for 4B models, see report here.
Environment
Python 3.12.3
transformers 4.41.2
torch 2.3.1
Detailed Description
No changes made, no code other than the import and initialization method
Metadata
Metadata
Assignees
Labels
bugSomething isn't workingSomething isn't working