Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
---
|
| 3 |
+
language: en
|
| 4 |
+
tags:
|
| 5 |
+
- sagemaker
|
| 6 |
+
- bart
|
| 7 |
+
- summarization
|
| 8 |
+
license: apache-2.0
|
| 9 |
+
datasets:
|
| 10 |
+
- samsum
|
| 11 |
+
model-index:
|
| 12 |
+
- name: bart-large-cnn-samsum
|
| 13 |
+
results:
|
| 14 |
+
- task:
|
| 15 |
+
name: Abstractive Text Summarization
|
| 16 |
+
type: abstractive-text-summarization
|
| 17 |
+
dataset:
|
| 18 |
+
name: "SAMSum Corpus: A Human-annotated Dialogue Dataset for Abstractive Summarization"
|
| 19 |
+
type: samsum
|
| 20 |
+
metrics:
|
| 21 |
+
- name: Validation ROGUE-1
|
| 22 |
+
type: rogue-1
|
| 23 |
+
value: 43.2111
|
| 24 |
+
- name: Validation ROGUE-2
|
| 25 |
+
type: rogue-2
|
| 26 |
+
value: 22.3519
|
| 27 |
+
- name: Validation ROGUE-L
|
| 28 |
+
type: rogue-l
|
| 29 |
+
value: 33.315
|
| 30 |
+
- name: Test ROGUE-1
|
| 31 |
+
type: rogue-1
|
| 32 |
+
value: 41.8283
|
| 33 |
+
- name: Test ROGUE-2
|
| 34 |
+
type: rogue-2
|
| 35 |
+
value: 20.9857
|
| 36 |
+
- name: Test ROGUE-L
|
| 37 |
+
type: rogue-l
|
| 38 |
+
value: 32.3602
|
| 39 |
+
widget:
|
| 40 |
+
- text: |
|
| 41 |
+
Sugi: I am tired of everything in my life.
|
| 42 |
+
Tommy: What? How happy you life is! I do envy you.
|
| 43 |
+
Sugi: You don't know that I have been over-protected by my mother these years. I am really about to leave the family and spread my wings.
|
| 44 |
+
Tommy: Maybe you are right.
|
| 45 |
+
---
|
| 46 |
+
## `bart-large-cnn-samsum`
|
| 47 |
+
This model was trained using Amazon SageMaker and the new Hugging Face Deep Learning container.
|
| 48 |
+
For more information look at:
|
| 49 |
+
- [🤗 Transformers Documentation: Amazon SageMaker](https://huggingface.co/transformers/sagemaker.html)
|
| 50 |
+
- [Example Notebooks](https://github.com/huggingface/notebooks/tree/master/sagemaker)
|
| 51 |
+
- [Amazon SageMaker documentation for Hugging Face](https://docs.aws.amazon.com/sagemaker/latest/dg/hugging-face.html)
|
| 52 |
+
- [Python SDK SageMaker documentation for Hugging Face](https://sagemaker.readthedocs.io/en/stable/frameworks/huggingface/index.html)
|
| 53 |
+
- [Deep Learning Container](https://github.com/aws/deep-learning-containers/blob/master/available_images.md#huggingface-training-containers)
|
| 54 |
+
## Hyperparameters
|
| 55 |
+
{
|
| 56 |
+
"dataset_name": "samsum",
|
| 57 |
+
"do_eval": true,
|
| 58 |
+
"do_predict": true,
|
| 59 |
+
"do_train": true,
|
| 60 |
+
"fp16": true,
|
| 61 |
+
"learning_rate": 5e-05,
|
| 62 |
+
"model_name_or_path": "facebook/bart-large-cnn",
|
| 63 |
+
"num_train_epochs": 3,
|
| 64 |
+
"output_dir": "/opt/ml/model",
|
| 65 |
+
"per_device_eval_batch_size": 4,
|
| 66 |
+
"per_device_train_batch_size": 4,
|
| 67 |
+
"predict_with_generate": true,
|
| 68 |
+
"seed": 7
|
| 69 |
+
}
|
| 70 |
+
## Usage
|
| 71 |
+
from transformers import pipeline
|
| 72 |
+
summarizer = pipeline("summarization", model="slauw87/bart-large-cnn-samsum")
|
| 73 |
+
conversation = '''Sugi: I am tired of everything in my life.
|
| 74 |
+
Tommy: What? How happy you life is! I do envy you.
|
| 75 |
+
Sugi: You don't know that I have been over-protected by my mother these years. I am really about to leave the family and spread my wings.
|
| 76 |
+
Tommy: Maybe you are right.
|
| 77 |
+
'''
|
| 78 |
+
nlp(conversation)
|
| 79 |
+
## Results
|
| 80 |
+
| key | value |
|
| 81 |
+
| --- | ----- |
|
| 82 |
+
| eval_rouge1 | 43.2111 |
|
| 83 |
+
| eval_rouge2 | 22.3519 |
|
| 84 |
+
| eval_rougeL | 33.3153 |
|
| 85 |
+
| eval_rougeLsum | 40.0527 |
|
| 86 |
+
| predict_rouge1 | 41.8283 |
|
| 87 |
+
| predict_rouge2 | 20.9857 |
|
| 88 |
+
| predict_rougeL | 32.3602 |
|
| 89 |
+
| predict_rougeLsum | 38.7316 |
|