File size: 4,195 Bytes
f81a490
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3525990
 
 
 
 
 
 
f81a490
 
 
 
 
 
 
781e572
 
 
 
f81a490
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
103
104
105
106
107
108
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- winograd_wsc
metrics:
- rouge
model-index:
- name: flan-t5-small-coref
  results:
  - task:
      name: Sequence-to-sequence Language Modeling
      type: text2text-generation
    dataset:
      name: winograd_wsc
      type: winograd_wsc
      config: wsc285
      split: test
      args: wsc285
    metrics:
    - name: Rouge1
      type: rouge
      value: 0.906
widget:
- text: "Sam has a Parker pen. He loves writing with it."
  example_title: "Example 1"
- text: "Coronavirus quickly spread worldwide in 2020. The virus mostly affects elderly people. They can easily catch it."
  example_title: "Example 2"
- text: "First, the manager evaluates the candidates. Afterwards, he notifies the candidates regarding the evaluation."
  example_title: "Example 3"
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# flan-t5-small-coref

This model is a fine-tuned version of [google/flan-t5-small](https://huggingface.co/google/flan-t5-small) on the winograd_wsc dataset.  

The model was trained on the task of coreference resolution.  

It achieves the following results on the evaluation set:
- Loss: 0.5656
- Rouge1: 0.906
- Rouge2: 0.8192
- Rougel: 0.9016
- Rougelsum: 0.9026
- Gen Len: 23.1724

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 20

### Training results

| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------:|:---------:|:-------:|
| No log        | 1.0   | 16   | 1.0901          | 0.6849 | 0.561  | 0.6734 | 0.6746    | 18.4483 |
| No log        | 2.0   | 32   | 0.9083          | 0.8512 | 0.7509 | 0.8438 | 0.8437    | 21.1379 |
| No log        | 3.0   | 48   | 0.8132          | 0.8638 | 0.7728 | 0.8588 | 0.8595    | 21.8276 |
| No log        | 4.0   | 64   | 0.7590          | 0.8786 | 0.7842 | 0.8744 | 0.876     | 22.2069 |
| No log        | 5.0   | 80   | 0.7225          | 0.8846 | 0.7928 | 0.8805 | 0.8817    | 22.3793 |
| No log        | 6.0   | 96   | 0.6920          | 0.886  | 0.7942 | 0.8821 | 0.8827    | 22.4483 |
| No log        | 7.0   | 112  | 0.6660          | 0.8861 | 0.7922 | 0.8816 | 0.8827    | 22.5172 |
| No log        | 8.0   | 128  | 0.6470          | 0.8879 | 0.7953 | 0.8836 | 0.8849    | 22.6897 |
| No log        | 9.0   | 144  | 0.6318          | 0.8968 | 0.806  | 0.8923 | 0.8933    | 23.069  |
| No log        | 10.0  | 160  | 0.6160          | 0.8968 | 0.806  | 0.8923 | 0.8933    | 23.069  |
| No log        | 11.0  | 176  | 0.6055          | 0.9056 | 0.822  | 0.9014 | 0.9021    | 23.1724 |
| No log        | 12.0  | 192  | 0.5962          | 0.9056 | 0.822  | 0.9014 | 0.9021    | 23.1724 |
| No log        | 13.0  | 208  | 0.5884          | 0.9074 | 0.8246 | 0.9033 | 0.9042    | 23.2069 |
| No log        | 14.0  | 224  | 0.5825          | 0.9049 | 0.8182 | 0.9005 | 0.9016    | 23.2414 |
| No log        | 15.0  | 240  | 0.5769          | 0.9049 | 0.8182 | 0.9005 | 0.9016    | 23.2414 |
| No log        | 16.0  | 256  | 0.5727          | 0.903  | 0.8132 | 0.8991 | 0.8997    | 23.1724 |
| No log        | 17.0  | 272  | 0.5698          | 0.906  | 0.8192 | 0.9016 | 0.9026    | 23.1724 |
| No log        | 18.0  | 288  | 0.5673          | 0.906  | 0.8192 | 0.9016 | 0.9026    | 23.1724 |
| No log        | 19.0  | 304  | 0.5661          | 0.906  | 0.8192 | 0.9016 | 0.9026    | 23.1724 |
| No log        | 20.0  | 320  | 0.5656          | 0.906  | 0.8192 | 0.9016 | 0.9026    | 23.1724 |


### Framework versions

- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.7.1
- Tokenizers 0.13.2