Instructions to use OpenVINO/bert-base-uncased-sst2-unstructured80-int8-ov with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use OpenVINO/bert-base-uncased-sst2-unstructured80-int8-ov with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, NNCFNetwork tokenizer = AutoTokenizer.from_pretrained("OpenVINO/bert-base-uncased-sst2-unstructured80-int8-ov") model = NNCFNetwork.from_pretrained("OpenVINO/bert-base-uncased-sst2-unstructured80-int8-ov", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,108 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
datasets:
|
| 7 |
+
- glue
|
| 8 |
+
metrics:
|
| 9 |
+
- accuracy
|
| 10 |
+
model-index:
|
| 11 |
+
- name: yujiepan/bert-base-uncased-sst2-int8-unstructured80
|
| 12 |
+
results:
|
| 13 |
+
- task:
|
| 14 |
+
name: Text Classification
|
| 15 |
+
type: text-classification
|
| 16 |
+
dataset:
|
| 17 |
+
name: GLUE SST2
|
| 18 |
+
type: glue
|
| 19 |
+
config: sst2
|
| 20 |
+
split: validation
|
| 21 |
+
args: sst2
|
| 22 |
+
metrics:
|
| 23 |
+
- name: Accuracy
|
| 24 |
+
type: accuracy
|
| 25 |
+
value: 0.91284
|
| 26 |
+
pipeline_tag: text-classification
|
| 27 |
+
---
|
| 28 |
+
|
| 29 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 30 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 31 |
+
|
| 32 |
+
# Joint magnitude pruning, quantization and distillation on BERT-base/SST-2
|
| 33 |
+
|
| 34 |
+
This model conducts unstructured magnitude pruning, quantization and distillation at the same time on BERT-base when finetuning on the GLUE SST2 dataset.
|
| 35 |
+
It achieves the following results on the evaluation set:
|
| 36 |
+
- Torch accuracy: 0.9128
|
| 37 |
+
- OpenVINO IR accuracy: 0.9128
|
| 38 |
+
- Sparsity in transformer block linear layers: 0.80
|
| 39 |
+
|
| 40 |
+
## Setup
|
| 41 |
+
|
| 42 |
+
```
|
| 43 |
+
conda install pytorch torchvision torchaudio pytorch-cuda=11.6 -c pytorch -c nvidia
|
| 44 |
+
pip install optimum[openvino,nncf]==1.7.0
|
| 45 |
+
# TODO
|
| 46 |
+
pip install wandb # optional
|
| 47 |
+
```
|
| 48 |
+
|
| 49 |
+
## NNCF config
|
| 50 |
+
|
| 51 |
+
See `nncf_config.json` in this repo.
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
## Run
|
| 55 |
+
|
| 56 |
+
We use one card for training.
|
| 57 |
+
|
| 58 |
+
```
|
| 59 |
+
NNCFCFG=/path/to/nncf/config
|
| 60 |
+
python run_glue.py \
|
| 61 |
+
--lr_scheduler_type cosine_with_restarts \
|
| 62 |
+
--cosine_cycle_ratios 11,6 \
|
| 63 |
+
--cosine_cycle_decays 1,1 \
|
| 64 |
+
--save_best_model_after_epoch -1 \
|
| 65 |
+
--save_best_model_after_sparsity 0.7999 \
|
| 66 |
+
--model_name_or_path textattack/bert-base-uncased-SST-2 \
|
| 67 |
+
--teacher_model_or_path yoshitomo-matsubara/bert-large-uncased-sst2 \
|
| 68 |
+
--distillation_temperature 2 \
|
| 69 |
+
--task_name sst2 \
|
| 70 |
+
--nncf_compression_config $NNCFCFG \
|
| 71 |
+
--distillation_weight 0.95 \
|
| 72 |
+
--output_dir /tmp/bert-base-uncased-sst2-int8-unstructured80-17epoch \
|
| 73 |
+
--run_name bert-base-uncased-sst2-int8-unstructured80-17epoch \
|
| 74 |
+
--overwrite_output_dir \
|
| 75 |
+
--do_train \
|
| 76 |
+
--do_eval \
|
| 77 |
+
--max_seq_length 128 \
|
| 78 |
+
--per_device_train_batch_size 32 \
|
| 79 |
+
--per_device_eval_batch_size 32 \
|
| 80 |
+
--learning_rate 5e-05 \
|
| 81 |
+
--optim adamw_torch \
|
| 82 |
+
--num_train_epochs 17 \
|
| 83 |
+
--logging_steps 1 \
|
| 84 |
+
--evaluation_strategy steps \
|
| 85 |
+
--eval_steps 250 \
|
| 86 |
+
--save_strategy steps \
|
| 87 |
+
--save_steps 250 \
|
| 88 |
+
--save_total_limit 1 \
|
| 89 |
+
--fp16 \
|
| 90 |
+
--seed 1
|
| 91 |
+
```
|
| 92 |
+
|
| 93 |
+
The best model checkpoint is stored in the `best_model` folder. Here we only upload that checkpoint folder together with some config files.
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
## inference
|
| 97 |
+
|
| 98 |
+
https://gist.github.com/yujiepan-work/c38dc4e56c7a9d803c42988f7b7d260a
|
| 99 |
+
|
| 100 |
+
|
| 101 |
+
### Framework versions
|
| 102 |
+
|
| 103 |
+
- Transformers 4.26.0
|
| 104 |
+
- Pytorch 1.13.1+cu116
|
| 105 |
+
- Datasets 2.8.0
|
| 106 |
+
- Tokenizers 0.13.2
|
| 107 |
+
|
| 108 |
+
For a full description of the environment, please refer to `pip-requirements.txt` and `conda-requirements.txt`.
|