File size: 1,831 Bytes
cb1ad45
 
 
 
 
 
 
e6cbaa0
 
b8eab9c
 
 
 
 
 
cb1ad45
 
 
 
39b096f
cb1ad45
 
 
 
 
39b096f
cb1ad45
 
 
39b096f
cb1ad45
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e6cbaa0
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
---
base_model: gpt2
tags:
- generated_from_trainer
model-index:
- name: midi_model_2
  results: []
datasets:
- TristanBehrens/js-fakes-4bars

widget:
 - text: "PIECE_START"
 - text: "PIECE_START STYLE=JSFAKES GENRE=JSFAKES TRACK_START INST=48 BAR_START NOTE_ON=32"
 - text: "PIECE_START STYLE=JSFAKES GENRE=JSFAKES TRACK_START INST=48 BAR_START NOTE_ON=64"

---

# midi_model_2

This model is a fine-tuned version of [gpt2](https://huggingface.co/gpt2) on the js-fakes-4bars dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8079

## Model description

This model generates encoded midi that follows the format of [Magenta](https://github.com/magenta/note-seq).

## Intended uses & limitations

For generating basic encoded midi.

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 4
- eval_batch_size: 2
- seed: 1
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.01
- num_epochs: 1

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 2.3022        | 0.11  | 100  | 1.7587          |
| 1.5783        | 0.22  | 200  | 1.2644          |
| 1.1475        | 0.33  | 300  | 1.0365          |
| 1.0012        | 0.44  | 400  | 0.9359          |
| 0.936         | 0.55  | 500  | 0.8844          |
| 0.8895        | 0.66  | 600  | 0.8532          |
| 0.8714        | 0.77  | 700  | 0.8273          |
| 0.8521        | 0.88  | 800  | 0.8112          |
| 0.8455        | 1.0   | 900  | 0.8079          |


### Framework versions

- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0