IProject-10
commited on
Commit
•
fe1e26a
1
Parent(s):
e5f9d40
End of training
Browse files- README.md +20 -90
- config.json +2 -2
- training_args.bin +2 -2
README.md
CHANGED
@@ -1,99 +1,41 @@
|
|
1 |
---
|
2 |
-
|
3 |
license: mit
|
4 |
-
base_model:
|
5 |
tags:
|
6 |
- generated_from_trainer
|
7 |
datasets:
|
8 |
- squad_v2
|
9 |
model-index:
|
10 |
- name: deberta-v3-base-finetuned-squad2
|
11 |
-
results:
|
12 |
-
- task:
|
13 |
-
name: Question Answering
|
14 |
-
type: question-answering
|
15 |
-
dataset:
|
16 |
-
type: squad_v2
|
17 |
-
name: SQuAD 2
|
18 |
-
config: squad_v2
|
19 |
-
split: validation
|
20 |
-
metrics:
|
21 |
-
- type: exact_match
|
22 |
-
value: 84.56161037648447
|
23 |
-
name: Exact-Match
|
24 |
-
verified: true
|
25 |
-
- type: f1
|
26 |
-
value: 87.81110592215731
|
27 |
-
name: F1-score
|
28 |
-
verified: true
|
29 |
-
|
30 |
-
language:
|
31 |
-
- en
|
32 |
-
pipeline_tag: question-answering
|
33 |
-
metrics:
|
34 |
-
- exact_match
|
35 |
-
- f1
|
36 |
---
|
37 |
|
38 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
39 |
should probably proofread and complete it, then remove this comment. -->
|
40 |
|
41 |
-
|
42 |
-
|
43 |
-
DeBERTa-v3-base fine-tuned on SQuAD 2.0 : Encoder-based Transformer Language model.
|
44 |
-
The DeBERTa V3 base model comes with 12 layers and a hidden size of 768.
|
45 |
-
It has only 86M backbone parameters with a vocabulary containing 128K tokens which introduces 98M parameters in the Embedding layer.
|
46 |
-
This model was trained using the 160GB data as DeBERTa V2.
|
47 |
-
Suitable for Question-Answering tasks, predicts answer spans within the context provided.<br>
|
48 |
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
## Intended uses & limitations
|
57 |
|
58 |
-
|
59 |
|
60 |
-
|
61 |
-
!pip install transformers
|
62 |
-
from transformers import pipeline
|
63 |
-
model_checkpoint = "IProject-10/deberta-v3-base-finetuned-squad2"
|
64 |
-
question_answerer = pipeline("question-answering", model=model_checkpoint)
|
65 |
|
66 |
-
|
67 |
-
🤗 Transformers is backed by the three most popular deep learning libraries — Jax, PyTorch and TensorFlow — with a seamless integration
|
68 |
-
between them. It's straightforward to train your models with one before loading them for inference with the other.
|
69 |
-
"""
|
70 |
|
71 |
-
|
72 |
-
question_answerer(question=question, context=context)
|
73 |
-
```
|
74 |
|
75 |
-
##
|
76 |
|
77 |
-
|
78 |
|
79 |
-
|
80 |
-
exact: 84.56161037648447,
|
81 |
-
f1: 87.81110592215731,
|
82 |
-
total: 11873,
|
83 |
-
HasAns_exact: 81.62955465587045,
|
84 |
-
HasAns_f1: 88.13786447600818,
|
85 |
-
HasAns_total: 5928,
|
86 |
-
NoAns_exact: 87.48528174936922,
|
87 |
-
NoAns_f1: 87.48528174936922,
|
88 |
-
NoAns_total: 5945,
|
89 |
-
best_exact: 84.56161037648447,
|
90 |
-
best_exact_thresh: 0.9994288682937622,
|
91 |
-
best_f1: 87.81110592215778,
|
92 |
-
best_f1_thresh: 0.9994288682937622,
|
93 |
-
total_time_in_seconds: 336.43560706100106,
|
94 |
-
samples_per_second: 35.29055709566211,
|
95 |
-
latency_in_seconds: 0.028336191953255374
|
96 |
-
```
|
97 |
|
98 |
### Training hyperparameters
|
99 |
|
@@ -106,21 +48,9 @@ The following hyperparameters were used during training:
|
|
106 |
- lr_scheduler_type: linear
|
107 |
- num_epochs: 3
|
108 |
|
109 |
-
### Training results
|
110 |
-
|
111 |
-
| Training Loss | Epoch | Step | Validation Loss |
|
112 |
-
|:-------------:|:-----:|:-----:|:---------------:|
|
113 |
-
| 0.7299 | 1.0 | 8217 | 0.7246 |
|
114 |
-
| 0.5104 | 2.0 | 16434 | 0.7321 |
|
115 |
-
| 0.3547 | 3.0 | 24651 | 0.8493 |
|
116 |
-
|
117 |
-
This model is a fine-tuned version of [microsoft/deberta-v3-base](https://huggingface.co/microsoft/deberta-v3-base) on the squad_v2 dataset.
|
118 |
-
It achieves the following results on the evaluation set:
|
119 |
-
- Loss: 0.8493
|
120 |
-
|
121 |
### Framework versions
|
122 |
|
123 |
-
- Transformers 4.
|
124 |
- Pytorch 2.0.1+cu118
|
125 |
-
- Datasets 2.14.
|
126 |
-
- Tokenizers 0.13.3
|
|
|
1 |
---
|
|
|
2 |
license: mit
|
3 |
+
base_model: IProject-10/deberta-v3-base-finetuned-squad2
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
datasets:
|
7 |
- squad_v2
|
8 |
model-index:
|
9 |
- name: deberta-v3-base-finetuned-squad2
|
10 |
+
results: []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
---
|
12 |
|
13 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
14 |
should probably proofread and complete it, then remove this comment. -->
|
15 |
|
16 |
+
# deberta-v3-base-finetuned-squad2
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
|
18 |
+
This model is a fine-tuned version of [IProject-10/deberta-v3-base-finetuned-squad2](https://huggingface.co/IProject-10/deberta-v3-base-finetuned-squad2) on the squad_v2 dataset.
|
19 |
+
It achieves the following results on the evaluation set:
|
20 |
+
- eval_loss: 0.8493
|
21 |
+
- eval_runtime: 168.8242
|
22 |
+
- eval_samples_per_second: 71.678
|
23 |
+
- eval_steps_per_second: 4.484
|
24 |
+
- step: 0
|
|
|
25 |
|
26 |
+
## Model description
|
27 |
|
28 |
+
More information needed
|
|
|
|
|
|
|
|
|
29 |
|
30 |
+
## Intended uses & limitations
|
|
|
|
|
|
|
31 |
|
32 |
+
More information needed
|
|
|
|
|
33 |
|
34 |
+
## Training and evaluation data
|
35 |
|
36 |
+
More information needed
|
37 |
|
38 |
+
## Training procedure
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
|
40 |
### Training hyperparameters
|
41 |
|
|
|
48 |
- lr_scheduler_type: linear
|
49 |
- num_epochs: 3
|
50 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
### Framework versions
|
52 |
|
53 |
+
- Transformers 4.32.1
|
54 |
- Pytorch 2.0.1+cu118
|
55 |
+
- Datasets 2.14.4
|
56 |
+
- Tokenizers 0.13.3
|
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "
|
3 |
"architectures": [
|
4 |
"DebertaV2ForQuestionAnswering"
|
5 |
],
|
@@ -29,7 +29,7 @@
|
|
29 |
"relative_attention": true,
|
30 |
"share_att_key": true,
|
31 |
"torch_dtype": "float32",
|
32 |
-
"transformers_version": "4.
|
33 |
"type_vocab_size": 0,
|
34 |
"vocab_size": 128100
|
35 |
}
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "IProject-10/deberta-v3-base-finetuned-squad2",
|
3 |
"architectures": [
|
4 |
"DebertaV2ForQuestionAnswering"
|
5 |
],
|
|
|
29 |
"relative_attention": true,
|
30 |
"share_att_key": true,
|
31 |
"torch_dtype": "float32",
|
32 |
+
"transformers_version": "4.32.1",
|
33 |
"type_vocab_size": 0,
|
34 |
"vocab_size": 128100
|
35 |
}
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b5f933eba458dad2a124e1528e500389fab2da5fe18a9680ca05957e087976d1
|
3 |
+
size 4091
|