Commit
•
1b770ce
1
Parent(s):
2468b1d
uploaded readme
Browse files
README.md
ADDED
@@ -0,0 +1,103 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Quantization made by Richard Erkhov.
|
2 |
+
|
3 |
+
[Github](https://github.com/RichardErkhov)
|
4 |
+
|
5 |
+
[Discord](https://discord.gg/pvy7H8DZMG)
|
6 |
+
|
7 |
+
[Request more models](https://github.com/RichardErkhov/quant_request)
|
8 |
+
|
9 |
+
|
10 |
+
bart-base-finetuned-xsum - bnb 4bits
|
11 |
+
- Model creator: https://huggingface.co/Vexemous/
|
12 |
+
- Original model: https://huggingface.co/Vexemous/bart-base-finetuned-xsum/
|
13 |
+
|
14 |
+
|
15 |
+
|
16 |
+
|
17 |
+
Original model description:
|
18 |
+
---
|
19 |
+
license: apache-2.0
|
20 |
+
base_model: facebook/bart-base
|
21 |
+
tags:
|
22 |
+
- generated_from_trainer
|
23 |
+
datasets:
|
24 |
+
- xsum
|
25 |
+
metrics:
|
26 |
+
- rouge
|
27 |
+
model-index:
|
28 |
+
- name: bart-base-finetuned-xsum
|
29 |
+
results:
|
30 |
+
- task:
|
31 |
+
name: Sequence-to-sequence Language Modeling
|
32 |
+
type: text2text-generation
|
33 |
+
dataset:
|
34 |
+
name: xsum
|
35 |
+
type: xsum
|
36 |
+
config: default
|
37 |
+
split: train[:10%]
|
38 |
+
args: default
|
39 |
+
metrics:
|
40 |
+
- name: Rouge1
|
41 |
+
type: rouge
|
42 |
+
value: 35.8214
|
43 |
+
pipeline_tag: summarization
|
44 |
+
---
|
45 |
+
|
46 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
47 |
+
should probably proofread and complete it, then remove this comment. -->
|
48 |
+
|
49 |
+
# bart-base-finetuned-xsum
|
50 |
+
|
51 |
+
This model is a fine-tuned version of [facebook/bart-base](https://huggingface.co/facebook/bart-base) on the xsum dataset.
|
52 |
+
It achieves the following results on the evaluation set:
|
53 |
+
- Loss: 1.9356
|
54 |
+
- Rouge1: 35.8214
|
55 |
+
- Rouge2: 14.7565
|
56 |
+
- Rougel: 29.4566
|
57 |
+
- Rougelsum: 29.4496
|
58 |
+
- Gen Len: 19.562
|
59 |
+
|
60 |
+
## Model description
|
61 |
+
|
62 |
+
More information needed
|
63 |
+
|
64 |
+
## Intended uses & limitations
|
65 |
+
|
66 |
+
More information needed
|
67 |
+
|
68 |
+
## Training and evaluation data
|
69 |
+
|
70 |
+
More information needed
|
71 |
+
|
72 |
+
## Training procedure
|
73 |
+
|
74 |
+
### Training hyperparameters
|
75 |
+
|
76 |
+
The following hyperparameters were used during training:
|
77 |
+
- learning_rate: 2e-05
|
78 |
+
- train_batch_size: 16
|
79 |
+
- eval_batch_size: 16
|
80 |
+
- seed: 42
|
81 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
82 |
+
- lr_scheduler_type: linear
|
83 |
+
- num_epochs: 5
|
84 |
+
- mixed_precision_training: Native AMP
|
85 |
+
|
86 |
+
### Training results
|
87 |
+
|
88 |
+
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|
89 |
+
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
|
90 |
+
| 2.301 | 1.0 | 1148 | 1.9684 | 34.4715 | 13.6638 | 28.1147 | 28.1204 | 19.5816 |
|
91 |
+
| 2.1197 | 2.0 | 2296 | 1.9442 | 35.2502 | 14.284 | 28.8462 | 28.8384 | 19.5546 |
|
92 |
+
| 1.9804 | 3.0 | 3444 | 1.9406 | 35.7799 | 14.7422 | 29.3669 | 29.3742 | 19.5326 |
|
93 |
+
| 1.8891 | 4.0 | 4592 | 1.9349 | 35.5151 | 14.4668 | 29.0359 | 29.0484 | 19.5492 |
|
94 |
+
| 1.827 | 5.0 | 5740 | 1.9356 | 35.8214 | 14.7565 | 29.4566 | 29.4496 | 19.562 |
|
95 |
+
|
96 |
+
|
97 |
+
### Framework versions
|
98 |
+
|
99 |
+
- Transformers 4.40.1
|
100 |
+
- Pytorch 1.13.1+cu117
|
101 |
+
- Datasets 2.19.0
|
102 |
+
- Tokenizers 0.19.1
|
103 |
+
|