Update README.md
Browse files
README.md
CHANGED
@@ -1,20 +1,43 @@
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
tags:
|
4 |
-
-
|
|
|
5 |
metrics:
|
6 |
- bleu
|
|
|
|
|
|
|
|
|
|
|
7 |
model-index:
|
8 |
- name: h1
|
9 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
---
|
11 |
|
12 |
-
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
13 |
-
should probably proofread and complete it, then remove this comment. -->
|
14 |
-
|
15 |
# h1
|
16 |
|
17 |
-
This model is a fine-tuned version of [distilgpt2](https://huggingface.co/distilgpt2) on
|
18 |
It achieves the following results on the evaluation set:
|
19 |
- Loss: 0.0890
|
20 |
- Exact Match: 0.1970
|
@@ -28,15 +51,24 @@ It achieves the following results on the evaluation set:
|
|
28 |
|
29 |
## Model description
|
30 |
|
31 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
32 |
|
33 |
## Intended uses & limitations
|
34 |
|
35 |
-
|
36 |
|
37 |
## Training and evaluation data
|
38 |
|
39 |
-
|
40 |
|
41 |
## Training procedure
|
42 |
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
tags:
|
4 |
+
- distigpt2
|
5 |
+
- hearthstone
|
6 |
metrics:
|
7 |
- bleu
|
8 |
+
- dvitel/codebleu
|
9 |
+
- exact_match
|
10 |
+
- chrf
|
11 |
+
datasets:
|
12 |
+
- dvitel/hearthstone
|
13 |
model-index:
|
14 |
- name: h1
|
15 |
+
results:
|
16 |
+
- task:
|
17 |
+
type: text-generation
|
18 |
+
name: Python Code Synthesis
|
19 |
+
dataset:
|
20 |
+
type: dvitel/hearthstone
|
21 |
+
name: HearthStone
|
22 |
+
split: test
|
23 |
+
metrics:
|
24 |
+
- type: exact_match
|
25 |
+
value: 0.21212121212121213
|
26 |
+
name: Exact Match
|
27 |
+
- type: bleu
|
28 |
+
value: 0.9637468196180485
|
29 |
+
name: BLEU
|
30 |
+
- type: dvitel/codebleu
|
31 |
+
value: 0.8884667222252154
|
32 |
+
name: CodeBLEU
|
33 |
+
- type: dvitel/codebleu
|
34 |
+
value: 96.5942286007928
|
35 |
+
name: chrF
|
36 |
---
|
37 |
|
|
|
|
|
|
|
38 |
# h1
|
39 |
|
40 |
+
This model is a fine-tuned version of [distilgpt2](https://huggingface.co/distilgpt2) on [hearthstone](https://huggingface.co/datasets/dvitel/hearthstone) dataset.
|
41 |
It achieves the following results on the evaluation set:
|
42 |
- Loss: 0.0890
|
43 |
- Exact Match: 0.1970
|
|
|
51 |
|
52 |
## Model description
|
53 |
|
54 |
+
DistilGPT2 applied onto HearthStone dataset with preprocessing of python code to dumped AST. Example:
|
55 |
+
```python
|
56 |
+
#gold labels
|
57 |
+
Module([ClassDef('Innervate', [Name('SpellCard', Load())], [], [FunctionDef('__init__', arguments([], [arg('self', None, None)], None, [], [], None, []), [Expr(Call(Attribute(Call(Name('super', Load()), [], []), '__init__', Load()), [Constant('Innervate', None), Constant(0, None), Attribute(Name('CHARACTER_CLASS', Load()), 'DRUID', Load()), Attribute(Name('CARD_RARITY', Load()), 'FREE', Load())], []))], [], None, None), FunctionDef('use', arguments([], [arg('self', None, None), arg('player', None, None), arg('game', None, None)], None, [], [], None, []), [Expr(Call(Attribute(Call(Name('super', Load()), [], []), 'use', Load()), [Name('player', Load()), Name('game', Load())], [])), If(Compare(Attribute(Name('player', Load()),'mana', Load()), [Lt()], [Constant(8, None)]), [AugAssign(Attribute(Name('player', Load()),'mana', Store()), Add(), Constant(2, None))], [Assign([Attribute(Name('player', Load()),'mana', Store())], Constant(10, None), None)])], [], None, None)], [])], [])
|
58 |
+
```
|
59 |
+
```python
|
60 |
+
#wrong prediction (example of error after training)
|
61 |
+
Module([ClassDef('Innervate', [Name('SpellCard', Load())], [], [FunctionDef('__init__', arguments([], [arg('self', None, None)], None, [], [], None, []), [Expr(Call(Attribute(Call(Name('super', Load()), [], []), '__init__', Load()), [Constant('Innervate', None), Constant(0, None), Attribute(Name('CHARACTER_CLASS', Load()), 'DRUID', Load()), Attribute(Name('CARD_RARITY', Load()), 'FREE', Load())], []))], [], None, None), FunctionDef('use', arguments([], [arg('self', None, None), arg('player', None, None), arg('game', None, None)], None, [], [], None, []), [Expr(Call(Attribute(Call(Name('super', Load()), [], []), 'use', Load()), [Name('player', Load()), Name('game', Load())], [])), For(Compare(Attribute(Name('player', Load()),'maxa', Load()), [Lt()], [Constant(10, None)]), [AugAssign(Attribute(Name('player', Load()),'mana', Store()), Add(), Constant(2, None))], Exign([Name(Name('player', Load()),'mana', Store())], Constant(None, None), None)],], [], None, None)], [])], [])
|
62 |
+
```
|
63 |
+
|
64 |
|
65 |
## Intended uses & limitations
|
66 |
|
67 |
+
HearthStone card code synthesis.
|
68 |
|
69 |
## Training and evaluation data
|
70 |
|
71 |
+
See split of [hearthstone](https://huggingface.co/datasets/dvitel/hearthstone) dataset
|
72 |
|
73 |
## Training procedure
|
74 |
|