File size: 3,439 Bytes
c4125a8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ea437e5
c4125a8
 
 
 
 
 
 
 
 
ea437e5
 
c4125a8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ea437e5
c4125a8
 
 
 
 
ea437e5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c4125a8
 
 
 
 
ea437e5
c4125a8
 
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
---
license: apache-2.0
base_model: t5-large
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- accuracy
model-index:
- name: t5-large_sst2_sp0_ar0
  results:
  - task:
      name: Text Classification
      type: text-classification
    dataset:
      name: glue
      type: glue
      config: sst2
      split: validation
      args: sst2
    metrics:
    - name: Accuracy
      type: accuracy
      value: 0.9453125
---

<!-- 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. -->

# t5-large_sst2_sp0_ar0

This model is a fine-tuned version of [t5-large](https://huggingface.co/t5-large) on the glue dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1944
- Accuracy: 0.9453

## 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: 5e-05
- train_batch_size: 16
- eval_batch_size: 32
- seed: 1
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 20
- num_epochs: 6

### Training results

| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 0.6815        | 0.01  | 25   | 0.6999          | 0.5092   |
| 0.6592        | 0.01  | 50   | 0.6221          | 0.6445   |
| 0.5832        | 0.02  | 75   | 0.4570          | 0.7993   |
| 0.2882        | 0.02  | 100  | 0.2076          | 0.9358   |
| 0.1894        | 0.03  | 125  | 0.3499          | 0.9404   |
| 0.1864        | 0.04  | 150  | 0.2963          | 0.9461   |
| 0.2553        | 0.04  | 175  | 0.6929          | 0.9289   |
| 0.245         | 0.05  | 200  | 0.4761          | 0.9323   |
| 0.2042        | 0.05  | 225  | 0.5294          | 0.9461   |
| 0.2002        | 0.06  | 250  | 0.8441          | 0.9472   |
| 0.1633        | 0.07  | 275  | 0.8560          | 0.9495   |
| 0.1939        | 0.07  | 300  | 0.3197          | 0.9450   |
| 0.1928        | 0.08  | 325  | 0.4214          | 0.9472   |
| 0.2201        | 0.08  | 350  | 0.5266          | 0.9484   |
| 0.143         | 0.09  | 375  | 0.8642          | 0.9450   |
| 0.2354        | 0.1   | 400  | 1.2116          | 0.9335   |
| 0.1692        | 0.1   | 425  | 0.1807          | 0.9472   |
| 0.1531        | 0.11  | 450  | 0.6431          | 0.9484   |
| 0.152         | 0.11  | 475  | 1.4046          | 0.9553   |
| 0.1948        | 0.12  | 500  | 0.1596          | 0.9553   |
| 0.2007        | 0.13  | 525  | 0.1779          | 0.9438   |
| 0.1338        | 0.13  | 550  | 0.6476          | 0.9495   |
| 0.3812        | 0.14  | 575  | 0.3901          | 0.9484   |
| 0.7052        | 0.14  | 600  | 0.1740          | 0.9507   |
| 0.8601        | 0.15  | 625  | 1.5226          | 0.9484   |
| 1.384         | 0.16  | 650  | 0.6605          | 0.9427   |
| 0.6833        | 0.16  | 675  | 0.7313          | 0.9484   |
| 0.1833        | 0.17  | 700  | 0.4110          | 0.9438   |
| 0.1968        | 0.17  | 725  | 0.2914          | 0.9450   |
| 0.2001        | 0.18  | 750  | 0.1947          | 0.9335   |


### Framework versions

- Transformers 4.33.2
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.11.6