Commit
•
8147896
1
Parent(s):
ead58ea
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- generated_from_trainer
|
4 |
+
datasets:
|
5 |
+
- cnn_dailymail
|
6 |
+
metrics:
|
7 |
+
- rouge
|
8 |
+
model-index:
|
9 |
+
- name: pegasus-newsroom-cnn-adam8bit-bs4x64acc
|
10 |
+
results:
|
11 |
+
- task:
|
12 |
+
name: Sequence-to-sequence Language Modeling
|
13 |
+
type: text2text-generation
|
14 |
+
dataset:
|
15 |
+
name: cnn_dailymail
|
16 |
+
type: cnn_dailymail
|
17 |
+
args: 3.0.0
|
18 |
+
metrics:
|
19 |
+
- name: Rouge1
|
20 |
+
type: rouge
|
21 |
+
value: 44.2881
|
22 |
+
---
|
23 |
+
|
24 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
25 |
+
should probably proofread and complete it, then remove this comment. -->
|
26 |
+
|
27 |
+
# pegasus-newsroom-cnn-adam8bit-bs4x64acc
|
28 |
+
|
29 |
+
This model is a fine-tuned version of [oMateos2020/pegasus-newsroom-cnn-adam8bit-bs16x64acc](https://huggingface.co/oMateos2020/pegasus-newsroom-cnn-adam8bit-bs16x64acc) on the cnn_dailymail dataset.
|
30 |
+
It achieves the following results on the evaluation set:
|
31 |
+
- Loss: 2.8608
|
32 |
+
- Rouge1: 44.2881
|
33 |
+
- Rouge2: 21.5487
|
34 |
+
- Rougel: 31.3798
|
35 |
+
- Rougelsum: 41.2326
|
36 |
+
- Gen Len: 71.7744
|
37 |
+
|
38 |
+
## Model description
|
39 |
+
|
40 |
+
More information needed
|
41 |
+
|
42 |
+
## Intended uses & limitations
|
43 |
+
|
44 |
+
More information needed
|
45 |
+
|
46 |
+
## Training and evaluation data
|
47 |
+
|
48 |
+
More information needed
|
49 |
+
|
50 |
+
## Training procedure
|
51 |
+
|
52 |
+
### Training hyperparameters
|
53 |
+
|
54 |
+
The following hyperparameters were used during training:
|
55 |
+
- learning_rate: 6.4e-05
|
56 |
+
- train_batch_size: 4
|
57 |
+
- eval_batch_size: 32
|
58 |
+
- seed: 42
|
59 |
+
- gradient_accumulation_steps: 64
|
60 |
+
- total_train_batch_size: 256
|
61 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
62 |
+
- lr_scheduler_type: linear
|
63 |
+
- lr_scheduler_warmup_ratio: 0.4
|
64 |
+
- num_epochs: 1
|
65 |
+
- mixed_precision_training: Native AMP
|
66 |
+
- label_smoothing_factor: 0.1
|
67 |
+
|
68 |
+
### Training results
|
69 |
+
|
70 |
+
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|
71 |
+
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
|
72 |
+
| 2.9307 | 1.0 | 1121 | 2.8608 | 44.2881 | 21.5487 | 31.3798 | 41.2326 | 71.7744 |
|
73 |
+
|
74 |
+
|
75 |
+
### Framework versions
|
76 |
+
|
77 |
+
- Transformers 4.20.1
|
78 |
+
- Pytorch 1.12.1+cu113
|
79 |
+
- Datasets 2.4.0
|
80 |
+
- Tokenizers 0.12.1
|