transformers

transformers copied to clipboard

issue with loading pretrained model using DeepSpeed Zero Stage 3

12

### System Info ```shell - `transformers` version: 4.19.0.dev0 - Platform: Linux-5.4.0-90-generic-x86_64-with-glibc2.29 - Python version: 3.8.10 - Huggingface_hub version: 0.5.1 - PyTorch version (GPU?): 1.12.0.dev20220505+cu113 (True) - Tensorflow version (GPU?): not...

pacman100

BigBird inference: same input data gives different outputs

7

### System Info - pytorch==1.10.2 - transformers==4.20.1 - python=3.9.7 - ### Who can help? _No response_ ### Information - [ ] The official example scripts - [X] My own modified...

NautiyalAmit

Fix for issue #12182 to ensure that the tutorial for zero shot distillation works

2

# What does this PR do? The code for training models via zero shot distillation was breaking because the .map() function was removing the _labels_ field from the dataset object....

pramodith

DeBERTa can't load some parameters

2

### System Info - `transformers` version: 4.21.1 - Platform: Linux-5.4.0-81-generic-x86_64-with-glibc2.31 - Python version: 3.9.12 - Huggingface_hub version: 0.8.1 - PyTorch version (GPU?): 1.11.0+cu113 (True) - Tensorflow version (GPU?): not installed...

sooftware

Refactor Pytorch `model.generate` method to work on TPU

3

### Feature request Refactor PT version of the method `model.generate` for text generating models to make it compatible with XLA and speed up inference on TPU. ### Motivation Right now,...

mikcnt

Cannot import pipelines from transformers

2

Hi, I use a Windows 10 Proffessional laptop for development and am using python 3.7.6 conda virtual env. I get the following run time error when running my code (below,...

balachander1964

LEDForSequenceClassification fine-tuning model gives: IndexError: index out of range in self

5

### System Info transformers - 4.21.1 Python - 3.8.13 torch - 1.12.0+cu113 ### Who can help? @patrickvonplaten ### Information - [X] The official example scripts - [ ] My own...

darshan2203

fix the description of token used for Bart classification

2

# What does this PR do? Fixes the description of token used for Bart classification. It uses last EOS token and not a special pooled output. https://github.com/huggingface/transformers/blob/7a8118947f3c6a802a9f63dc22c394961d38860f/src/transformers/models/bart/modeling_bart.py#L1520-L1527 ## Before submitting...

ekagra-ranjan

Remove unused `cur_len` in generation_utils.py

2

# What does this PR do? Removes unused `cur_len` in `generation_utils.py` ## Before submitting - [ ] This PR fixes a typo or improves the docs (you can dismiss the...

ekagra-ranjan

BertLMHeadModel (w/ relative position embedding) does not work correctly when use_cache = True

3

### System Info - `transformers` version: 4.20.1 - Platform: Linux-5.4.0-92-generic-x86_64-with-glibc2.17 - Python version: 3.8.13 - Huggingface_hub version: 0.8.1 - PyTorch version (GPU?): 1.12.0 (True) - Tensorflow version (GPU?): not installed...

jsh710101