File size: 5,381 Bytes
f255c7a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
109
110
111
112
113
114
115
116
117
---
license: mit
base_model: haryoaw/scenario-MDBT-TCR_data-cl-massive_all_1_1
tags:
- generated_from_trainer
datasets:
- massive
metrics:
- accuracy
- f1
model-index:
- name: scenario-KD-SCR-MSV-CL-D2_data-cl-massive_all_1_155
  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. -->

# scenario-KD-SCR-MSV-CL-D2_data-cl-massive_all_1_155

This model is a fine-tuned version of [haryoaw/scenario-MDBT-TCR_data-cl-massive_all_1_1](https://huggingface.co/haryoaw/scenario-MDBT-TCR_data-cl-massive_all_1_1) on the massive dataset.
It achieves the following results on the evaluation set:
- Loss: nan
- Accuracy: 0.0315
- F1: 0.0010

## 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: 32
- eval_batch_size: 32
- seed: 55
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 30

### Training results

| Training Loss | Epoch | Step   | Validation Loss | Accuracy | F1     |
|:-------------:|:-----:|:------:|:---------------:|:--------:|:------:|
| 0.0           | 0.56  | 5000   | nan             | 0.0315   | 0.0010 |
| 0.0           | 1.11  | 10000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 1.67  | 15000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 2.22  | 20000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 2.78  | 25000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 3.33  | 30000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 3.89  | 35000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 4.45  | 40000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 5.0   | 45000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 5.56  | 50000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 6.11  | 55000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 6.67  | 60000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 7.23  | 65000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 7.78  | 70000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 8.34  | 75000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 8.89  | 80000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 9.45  | 85000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 10.0  | 90000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 10.56 | 95000  | nan             | 0.0315   | 0.0010 |
| 0.0           | 11.12 | 100000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 11.67 | 105000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 12.23 | 110000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 12.78 | 115000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 13.34 | 120000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 13.9  | 125000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 14.45 | 130000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 15.01 | 135000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 15.56 | 140000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 16.12 | 145000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 16.67 | 150000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 17.23 | 155000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 17.79 | 160000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 18.34 | 165000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 18.9  | 170000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 19.45 | 175000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 20.01 | 180000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 20.56 | 185000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 21.12 | 190000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 21.68 | 195000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 22.23 | 200000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 22.79 | 205000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 23.34 | 210000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 23.9  | 215000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 24.46 | 220000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 25.01 | 225000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 25.57 | 230000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 26.12 | 235000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 26.68 | 240000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 27.23 | 245000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 27.79 | 250000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 28.35 | 255000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 28.9  | 260000 | nan             | 0.0315   | 0.0010 |
| 0.0           | 29.46 | 265000 | nan             | 0.0315   | 0.0010 |


### Framework versions

- Transformers 4.33.3
- Pytorch 2.1.1+cu121
- Datasets 2.14.5
- Tokenizers 0.13.3