Create README
Browse files
README.md
ADDED
|
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: en
|
| 3 |
+
tags:
|
| 4 |
+
- bart
|
| 5 |
+
- seq2seq
|
| 6 |
+
- summarization
|
| 7 |
+
license: apache-2.0
|
| 8 |
+
datasets:
|
| 9 |
+
- samsum
|
| 10 |
+
widget:
|
| 11 |
+
- text: |
|
| 12 |
+
Hannah: Hey, do you have Betty's number?
|
| 13 |
+
Amanda: Lemme check
|
| 14 |
+
Amanda: Sorry, can't find it.
|
| 15 |
+
Amanda: Ask Larry
|
| 16 |
+
Amanda: He called her last time we were at the park together
|
| 17 |
+
Hannah: I don't know him well
|
| 18 |
+
Amanda: Don't be shy, he's very nice
|
| 19 |
+
Hannah: If you say so..
|
| 20 |
+
Hannah: I'd rather you texted him
|
| 21 |
+
Amanda: Just text him 🙂
|
| 22 |
+
Hannah: Urgh.. Alright
|
| 23 |
+
Hannah: Bye
|
| 24 |
+
Amanda: Bye bye
|
| 25 |
+
model-index:
|
| 26 |
+
- name: bart-large-xsum-samsum
|
| 27 |
+
results:
|
| 28 |
+
- task:
|
| 29 |
+
name: Abstractive Text Summarization
|
| 30 |
+
type: abstractive-text-summarization
|
| 31 |
+
dataset:
|
| 32 |
+
name: "SAMSum Corpus: A Human-annotated Dialogue Dataset for Abstractive Summarization"
|
| 33 |
+
type: samsum
|
| 34 |
+
metrics:
|
| 35 |
+
- name: Validation ROGUE-1
|
| 36 |
+
type: rogue-1
|
| 37 |
+
value: 54.3921
|
| 38 |
+
- name: Validation ROGUE-2
|
| 39 |
+
type: rogue-2
|
| 40 |
+
value: 29.8078
|
| 41 |
+
- name: Validation ROGUE-L
|
| 42 |
+
type: rogue-l
|
| 43 |
+
value: 45.1543
|
| 44 |
+
- name: Test ROGUE-1
|
| 45 |
+
type: rogue-1
|
| 46 |
+
value: 53.3059
|
| 47 |
+
- name: Test ROGUE-2
|
| 48 |
+
type: rogue-2
|
| 49 |
+
value: 28.355
|
| 50 |
+
- name: Test ROGUE-L
|
| 51 |
+
type: rogue-l
|
| 52 |
+
value: 44.0953
|
| 53 |
+
---
|
| 54 |
+
## `bart-large-xsum-samsum`
|
| 55 |
+
This model was obtained by fine-tuning `facebook/bart-large-xsum` on [Samsum](https://huggingface.co/datasets/samsum) dataset.
|
| 56 |
+
## Usage
|
| 57 |
+
```python
|
| 58 |
+
from transformers import pipeline
|
| 59 |
+
|
| 60 |
+
summarizer = pipeline("summarization", model="lidiya/bart-large-xsum-samsum")
|
| 61 |
+
conversation = '''Hannah: Hey, do you have Betty's number?
|
| 62 |
+
Amanda: Lemme check
|
| 63 |
+
Amanda: Sorry, can't find it.
|
| 64 |
+
Amanda: Ask Larry
|
| 65 |
+
Amanda: He called her last time we were at the park together
|
| 66 |
+
Hannah: I don't know him well
|
| 67 |
+
Amanda: Don't be shy, he's very nice
|
| 68 |
+
Hannah: If you say so..
|
| 69 |
+
Hannah: I'd rather you texted him
|
| 70 |
+
Amanda: Just text him 🙂
|
| 71 |
+
Hannah: Urgh.. Alright
|
| 72 |
+
Hannah: Bye
|
| 73 |
+
Amanda: Bye bye
|
| 74 |
+
'''
|
| 75 |
+
summarizer(conversation)
|
| 76 |
+
```
|
| 77 |
+
## Training procedure
|
| 78 |
+
- Colab notebook: https://colab.research.google.com/drive/1dul0Sg-TTMy9xZCJzmDRajXbyzDwtYx6?usp=sharing
|
| 79 |
+
## Results
|
| 80 |
+
| key | value |
|
| 81 |
+
| --- | ----- |
|
| 82 |
+
| eval_rouge1 | 54.3921 |
|
| 83 |
+
| eval_rouge2 | 29.8078 |
|
| 84 |
+
| eval_rougeL | 45.1543 |
|
| 85 |
+
| eval_rougeLsum | 49.942 |
|
| 86 |
+
| test_rouge1 | 53.3059 |
|
| 87 |
+
| test_rouge2 | 28.355 |
|
| 88 |
+
| test_rougeL | 44.0953 |
|
| 89 |
+
| test_rougeLsum | 48.9246 |
|