File size: 4,880 Bytes
3e09900 3803679 3e09900 c7b5b70 3e09900 53103db 3e09900 c7b5b70 3e09900 c7b5b70 3e09900 53103db 3e09900 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 |
---
language: en
thumbnail:
license: mit
tags:
- question-answering
- bert
- bert-base
datasets:
- squad
metrics:
- squad
widget:
- text: "Where is the Eiffel Tower located?"
context: "The Eiffel Tower is a wrought-iron lattice tower on the Champ de Mars in Paris, France. It is named after the engineer Gustave Eiffel, whose company designed and built the tower."
- text: "Who is Frederic Chopin?"
context: "Frédéric François Chopin, born Fryderyk Franciszek Chopin (1 March 1810 – 17 October 1849), was a Polish composer and virtuoso pianist of the Romantic era who wrote primarily for solo piano."
---
## BERT-base uncased model fine-tuned on SQuAD v1
This model was created using the [nn_pruning](https://github.com/huggingface/nn_pruning) python library: the **linear layers contains 27.0%** of the original weights.
This model CANNOT be used without using nn_pruning `optimize_model` function, as it uses NoNorms instead of LayerNorms and this is not currently supported by the Transformers library.
It uses ReLUs instead of GeLUs as in the initial BERT network, to speedup inference.
This does not need special handling, as it is supported by the Transformers library, and flagged in the model config by the ```"hidden_act": "relu"``` entry.
The model contains **43.0%** of the original weights **overall** (the embeddings account for a significant part of the model, and they are not pruned by this method).
With a simple resizing of the linear matrices it ran **1.96x as fast as BERT-base** on the evaluation.
This is possible because the pruning method lead to structured matrices: to visualize them, hover below on the plot to see the non-zero/zero parts of each matrix.
<div class="graph"><script src="/madlag/bert-base-uncased-squadv1-x1.96-f88.3-d27-hybrid-filled-opt-v1/raw/main/model_card/density_info.js" id="a069faa9-ad3b-4c4d-b3eb-aac9a32aa6dc"></script></div>
In terms of accuracy, its **F1 is 88.33**, compared with 88.5 for BERT-base, a **F1 drop of 0.17**.
## Fine-Pruning details
This model was fine-tuned from the HuggingFace [BERT](https://www.aclweb.org/anthology/N19-1423/) base uncased checkpoint on [SQuAD1.1](https://rajpurkar.github.io/SQuAD-explorer), and distilled from the model [bert-large-uncased-whole-word-masking-finetuned-squad](https://huggingface.co/bert-large-uncased-whole-word-masking-finetuned-squad).
This model is case-insensitive: it does not make a difference between english and English.
A side-effect of the block pruning is that some of the attention heads are completely removed: 55 heads were removed on a total of 144 (38.2%).
Here is a detailed view on how the remaining heads are distributed in the network after pruning.
<div class="graph"><script src="/madlag/bert-base-uncased-squadv1-x1.96-f88.3-d27-hybrid-filled-opt-v1/raw/main/model_card/pruning_info.js" id="2df293b0-bd3d-4a5c-8f0c-a35f8fdaef04"></script></div>
## Details of the SQuAD1.1 dataset
| Dataset | Split | # samples |
| -------- | ----- | --------- |
| SQuAD1.1 | train | 90.6K |
| SQuAD1.1 | eval | 11.1k |
### Fine-tuning
- Python: `3.8.5`
- Machine specs:
```CPU: Intel(R) Core(TM) i7-6700K CPU
Memory: 64 GiB
GPUs: 1 GeForce GTX 3090, with 24GiB memory
GPU driver: 455.23.05, CUDA: 11.1
```
### Results
**Pytorch model file size**: `374M` (original BERT: `438M`)
| Metric | # Value | # Original ([Table 2](https://www.aclweb.org/anthology/N19-1423.pdf))| Variation |
| ------ | --------- | --------- | --------- |
| **EM** | **81.31** | **80.8** | **+0.51**|
| **F1** | **88.33** | **88.5** | **-0.17**|
## Example Usage
Install nn_pruning: it contains the optimization script, which just pack the linear layers into smaller ones by removing empty rows/columns.
`pip install git+https://github.com//huggingface/nn_pruning`
Then you can use the `transformers library` almost as usual: you just have to call `optimize_model` when the pipeline has loaded.
```python
from transformers import pipeline
from nn_pruning.inference_model_patcher import optimize_model
qa_pipeline = pipeline(
"question-answering",
model="madlag/bert-base-uncased-squadv1-x1.96-f88.3-d27-hybrid-filled-opt-v1",
tokenizer="madlag/bert-base-uncased-squadv1-x1.96-f88.3-d27-hybrid-filled-opt-v1"
)
print("BERT-base parameters: 110M")
print(f"Parameters count (includes head pruning)={int(qa_pipeline.model.num_parameters() / 1E6)}M")
qa_pipeline.model = optimize_model(qa_pipeline.model, "dense")
print(f"Parameters count after optimization={int(qa_pipeline.model.num_parameters() / 1E6)}M")
predictions = qa_pipeline({
'context': "Frédéric François Chopin, born Fryderyk Franciszek Chopin (1 March 1810 – 17 October 1849), was a Polish composer and virtuoso pianist of the Romantic era who wrote primarily for solo piano.",
'question': "Who is Frederic Chopin?",
})
print("Predictions", predictions)
``` |