File size: 2,220 Bytes
b4a3748
 
 
 
 
 
63a68e7
b4a3748
 
 
 
 
 
 
 
 
 
e94a6be
b4a3748
e94a6be
 
b4a3748
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
63a68e7
 
b4a3748
 
 
f661e6f
b4a3748
 
 
63a68e7
 
e94a6be
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b4a3748
 
 
 
 
f661e6f
b4a3748
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
---
license: apache-2.0
base_model: google/mt5-small
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: mt5-small-task2-dataset2
  results: []
---

<!-- 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. -->

# mt5-small-task2-dataset2

This model is a fine-tuned version of [google/mt5-small](https://huggingface.co/google/mt5-small) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4320
- Accuracy: 0.37

## 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: 5.6e-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: 15

### Training results

| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 7.018         | 1.0   | 250  | 1.2234          | 0.014    |
| 1.6684        | 2.0   | 500  | 0.8157          | 0.124    |
| 1.0289        | 3.0   | 750  | 0.6527          | 0.222    |
| 0.8021        | 4.0   | 1000 | 0.5877          | 0.282    |
| 0.6964        | 5.0   | 1250 | 0.5360          | 0.3      |
| 0.6252        | 6.0   | 1500 | 0.5118          | 0.32     |
| 0.5828        | 7.0   | 1750 | 0.4899          | 0.318    |
| 0.5436        | 8.0   | 2000 | 0.4718          | 0.35     |
| 0.5232        | 9.0   | 2250 | 0.4625          | 0.34     |
| 0.5005        | 10.0  | 2500 | 0.4556          | 0.342    |
| 0.4789        | 11.0  | 2750 | 0.4436          | 0.356    |
| 0.4733        | 12.0  | 3000 | 0.4379          | 0.356    |
| 0.4651        | 13.0  | 3250 | 0.4347          | 0.366    |
| 0.4591        | 14.0  | 3500 | 0.4320          | 0.37     |
| 0.4508        | 15.0  | 3750 | 0.4320          | 0.37     |


### Framework versions

- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Tokenizers 0.15.0