Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# -*-coding:utf-8-*-
|
2 |
+
import streamlit as st
|
3 |
+
# code from https://huggingface.co/kakaobrain/kogpt
|
4 |
+
import torch
|
5 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
6 |
+
|
7 |
+
tokenizer = AutoTokenizer.from_pretrained(
|
8 |
+
'kakaobrain/kogpt', revision='KoGPT6B-ryan1.5b',
|
9 |
+
bos_token='[BOS]', eos_token='[EOS]', unk_token='[UNK]', pad_token='[PAD]', mask_token='[MASK]'
|
10 |
+
)
|
11 |
+
model = AutoModelForCausalLM.from_pretrained(
|
12 |
+
'kakaobrain/kogpt', revision='KoGPT6B-ryan1.5b',
|
13 |
+
pad_token_id=tokenizer.eos_token_id,
|
14 |
+
torch_dtype=torch.float16, low_cpu_mem_usage=False
|
15 |
+
).to(device='cpu', non_blocking=True)
|
16 |
+
_ = model.eval()
|
17 |
+
|
18 |
+
print("Model loading done!")
|
19 |
+
|
20 |
+
def gpt(prompt):
|
21 |
+
with torch.no_grad():
|
22 |
+
tokens = tokenizer.encode(prompt, return_tensors='pt').to(device='cpu', non_blocking=True)
|
23 |
+
gen_tokens = model.generate(tokens, do_sample=True, temperature=0.8, max_length=256)
|
24 |
+
generated = tokenizer.batch_decode(gen_tokens)[0]
|
25 |
+
|
26 |
+
return generated
|
27 |
+
|
28 |
+
#prompts
|
29 |
+
st.title("μ¬λ¬λΆλ€μ λ¬Έμ₯μ μμ±ν΄μ€λλ€. π€")
|
30 |
+
st.markdown("μΉ΄μΉ΄μ€ gpt μ¬μ©ν©λλ€.")
|
31 |
+
st.subheader("λͺκ°μ§ μμ : ")
|
32 |
+
example_1_str = "μ€λμ λ μ¨λ λ무 λλΆμλ€. λ΄μΌμ "
|
33 |
+
example_2_str = "μ°λ¦¬λ ν볡μ μΈμ λ κ°λ§νμ§λ§ νμ "
|
34 |
+
example_1 = st.button(example_1_str)
|
35 |
+
example_2 = st.button(example_2_str)
|
36 |
+
textbox = st.text_area('μ€λμ μλ¦λ€μμ ν₯ν΄ λ¬λ¦¬κ³ ', '',height=100, max_chars=500 )
|
37 |
+
button = st.button('μμ±:')
|
38 |
+
# output
|
39 |
+
st.subheader("κ²°κ³Όκ°: ")
|
40 |
+
if example_1:
|
41 |
+
with st.spinner('In progress.......'):
|
42 |
+
output_text = gpt(example_1_str)
|
43 |
+
st.markdown("## "+output_text)
|
44 |
+
if example_2:
|
45 |
+
with st.spinner('In progress.......'):
|
46 |
+
output_text = gpt(example_2_str)
|
47 |
+
st.markdown("## "+output_text)
|
48 |
+
if button:
|
49 |
+
with st.spinner('In progress.......'):
|
50 |
+
if textbox:
|
51 |
+
output_text = gpt(textbox)
|
52 |
+
else:
|
53 |
+
output_text = " "
|
54 |
+
st.markdown("## "+output_text)
|