SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Support for easier multimodal use of modular (#35056)

Browse Source 
* update modular and add examples

* style

* improve example comments

* style

* fix small logic issue for imports

* fix relative order issue when files do not make sense

* Improve comments

* trigger CIs
This commit is contained in:
Cyril Vallez
2024-12-04 15:13:11 +01:00
committed by GitHub
parent 46df859975
commit 1da1e0d7f2
13 changed files with 2424 additions and 103 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
examples/modular-transformers/modeling_new_task_model.py
View File

@@ -265,7 +265,7 @@ class NewTaskModelForNewTask(NewTaskModelPreTrainedModel, GenerationMixin):
min_dtype = torch.finfo(dtype).min
sequence_length = inputs_embeds.shape[1]
if using_static_cache:
target_length = past_key_values.get_max_length()
target_length = past_key_values.get_max_cache_shape()
else:
target_length = (
attention_mask.shape[-1]
@@ -358,9 +358,9 @@ class NewTaskModelForNewTask(NewTaskModelPreTrainedModel, GenerationMixin):
```python
>>> from PIL import Image
>>> import requests
>>> from transformers import AutoProcessor, NewTaskModelForConditionalGeneration
>>> from transformers import AutoProcessor, NewTaskModelForNewTask
>>> model = NewTaskModelForConditionalGeneration.from_pretrained("google/NewTaskModel-test-224px-hf")
>>> model = NewTaskModelForNewTask.from_pretrained("google/NewTaskModel-test-224px-hf")
>>> processor = AutoProcessor.from_pretrained("google/NewTaskModel-test-224px-hf")
>>> prompt = "answer en Where is the cow standing?"