Downloads · 30 days
16.6K
0% of all-time downloads
distilbart-mnli-12-3
valhalla/distilbart-mnli-12-3
distilbart-mnli-12-3 is a zero-shot classification model from valhalla. Use it when you need labels you did not train the model on. It is set up for transformers.
distilbart-mnli is the distilled version of bart-large-mnli created using the No Teacher Distillation technique proposed for BART summarisation by Huggingface, here.
All-time downloads
5.1M
Public
Repo size
3.1 GB
Likes
20
Public
Repo makeup
Click a slice to open those files.
.bin1 GB · 50%
At a glance
- Task
- Zero-Shot Classification
- Library
- transformers
- Model type
- bart
- Access
- Public
- Created
- Mar 2, 2022
- Updated
- Jun 14, 2021
- SHA
- ef9a58ce
Try a prompt
Datasets
Model card
From the Hugging Face model README
DistilBart-MNLI
distilbart-mnli is the distilled version of bart-large-mnli created using the No Teacher Distillation technique proposed for BART summarisation by Huggingface, here.
We just copy alternating layers from bart-large-mnli and finetune more on the same data.
| matched acc | mismatched acc | |
|---|---|---|
| bart-large-mnli (baseline, 12-12) | 89.9 | 90.01 |
| distilbart-mnli-12-1 | 87.08 | 87.5 |
| distilbart-mnli-12-3 | 88.1 | 88.19 |
| distilbart-mnli-12-6 | 89.19 | 89.01 |
| distilbart-mnli-12-9 | 89.56 | 89.52 |
This is a very simple and effective technique, as we can see the performance drop is very little.
Detailed performace trade-offs will be posted in this sheet.
Fine-tuning
If you want to train these models yourself, clone the distillbart-mnli repo and follow the steps below
Clone and install transformers from source
git clone https://github.com/huggingface/transformers.git
pip install -qqq -U ./transformers
Download MNLI data
python transformers/utils/download_glue_data.py --data_dir glue_data --tasks MNLI
Create student model
python create_student.py \
--teacher_model_name_or_path facebook/bart-large-mnli \
--student_encoder_layers 12 \
--student_decoder_layers 6 \
--save_path student-bart-mnli-12-6 \
Start fine-tuning
python run_glue.py args.json
You can find the logs of these trained models in this wandb project.
- Task
- Zero-Shot Classification
- Library
- transformers
- Type
- bart
- Hf Inference
- live
- Created
- Mar 2, 2022
- Updated
- Jun 14, 2021