File size: 1,709 Bytes
f17de0d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
base_model: mika5883/inverse_gec
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: inverse_gec_finetuned
  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. -->

# inverse_gec_finetuned

This model is a fine-tuned version of [mika5883/inverse_gec](https://huggingface.co/mika5883/inverse_gec) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2517
- Bleu: 59.3585
- Gen Len: 16.2412

## 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: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Bleu    | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|
| No log        | 1.0   | 40   | 0.3337          | 59.0634 | 16.2428 |
| No log        | 2.0   | 80   | 0.2966          | 59.1606 | 16.2464 |
| No log        | 3.0   | 120  | 0.2708          | 59.229  | 16.2444 |
| No log        | 4.0   | 160  | 0.2561          | 59.3149 | 16.2408 |
| No log        | 5.0   | 200  | 0.2517          | 59.3585 | 16.2412 |


### Framework versions

- Transformers 4.39.3
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.15.2