Spaces:
Build error
Build error
modified model path
Browse files
app.py
CHANGED
@@ -4,19 +4,26 @@ import os
|
|
4 |
import torch
|
5 |
|
6 |
theme = "darkgrass"
|
7 |
-
title = "
|
8 |
-
model_name = "EleutherAI/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
10 |
-
description = "
|
11 |
-
|
12 |
examples = [
|
13 |
-
["
|
14 |
-
["์ง๋ฌธ:
|
15 |
["2040๋
๋ฏธ๊ตญ์, "]
|
16 |
]
|
17 |
-
tokenizer = AutoTokenizer.from_pretrained(
|
18 |
model = AutoModelForCausalLM.from_pretrained(
|
19 |
-
|
20 |
)
|
21 |
model.eval()
|
22 |
|
@@ -30,10 +37,7 @@ def predict(text):
|
|
30 |
tokens, do_sample=True, temperature=0.8, max_new_tokens=64, top_k=50, top_p=0.8,
|
31 |
no_repeat_ngram_size=3, repetition_penalty=1.2,
|
32 |
bad_words_ids=[
|
33 |
-
tokenizer.encode(
|
34 |
-
tokenizer.encode('....'),
|
35 |
-
tokenizer.encode('(์ค๋ต)'),
|
36 |
-
tokenizer.encode('http')
|
37 |
],
|
38 |
eos_token_id=tokenizer.eos_token_id,
|
39 |
pad_token_id=tokenizer.pad_token_id
|
|
|
4 |
import torch
|
5 |
|
6 |
theme = "darkgrass"
|
7 |
+
title = "Polyglot(Korean) Demo"
|
8 |
+
model_name = "EleutherAI/polyglot-ko-1.3b"
|
9 |
+
|
10 |
+
bad_words = [
|
11 |
+
'...',
|
12 |
+
'....',
|
13 |
+
'(์ค๋ต)',
|
14 |
+
'http'
|
15 |
+
]
|
16 |
|
17 |
+
description = "polyglot (1.3B ํ๋ผ๋ฏธํฐ ์ฌ์ด์ฆ) ํ๊ตญ์ด ๋ชจ๋ธ์ ์์ฐํ๋ ๋ฐ๋ชจํ์ด์ง ์
๋๋ค."
|
18 |
+
article = "<p style='text-align: center'><a href='https://github.com/EleutherAI/polyglot' target='_blank'>Polyglot: Large Language Models of Well-balanced Competence in Multi-languages</a></p>"
|
19 |
examples = [
|
20 |
+
["CPU์ GPU์ ์ฐจ์ด๋,"],
|
21 |
+
["์ง๋ฌธ: ์ฐํฌ๋ผ์ด๋ ์ ์์ด ์ธ๊ณ3์ฐจ๋์ ์ผ๋ก ํ์ ์ด ๋ ๊น์? \n๋ต๋ณ:"],
|
22 |
["2040๋
๋ฏธ๊ตญ์, "]
|
23 |
]
|
24 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
25 |
model = AutoModelForCausalLM.from_pretrained(
|
26 |
+
model_name
|
27 |
)
|
28 |
model.eval()
|
29 |
|
|
|
37 |
tokens, do_sample=True, temperature=0.8, max_new_tokens=64, top_k=50, top_p=0.8,
|
38 |
no_repeat_ngram_size=3, repetition_penalty=1.2,
|
39 |
bad_words_ids=[
|
40 |
+
tokenizer.encode(bad_word) for bad_word in bad_words
|
|
|
|
|
|
|
41 |
],
|
42 |
eos_token_id=tokenizer.eos_token_id,
|
43 |
pad_token_id=tokenizer.pad_token_id
|