/
gpt2_run.py
48 lines (41 loc) · 1.63 KB
/
gpt2_run.py
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
import tensorflow as tf
from transformers import TFGPT2LMHeadModel, GPT2Tokenizer
import streamlit as st
def load_gpt2():
tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
model = TFGPT2LMHeadModel.from_pretrained("gpt2",
pad_token_id=tokenizer.eos_token_id)
return tokenizer, model
tokenizer, model = load_gpt2()
random_seed = st.sidebar.slider('seed', min_value=0, max_value=1234,
value=0, step=1)
max_length = st.sidebar.slider('max_length', min_value=1, max_value=100,
value=50, step=1)
top_p = st.sidebar.slider('top_p', min_value=0.0, max_value=1.0,
value=0.92, step=0.01)
top_k = st.sidebar.slider('top_k', min_value=0, max_value=100,
value=0, step=1)
def gpt2_generate(user_input,
random_seed=0,
max_length=50,
top_p=0.92,
top_k=0):
input_ids = tokenizer.encode(user_input,
return_tensors='tf')
tf.random.set_seed(0)
sample_output = model.generate(
input_ids,
do_sample=True,
max_length=max_length,
top_p=top_p,
top_k=top_k
)
result = tokenizer.decode(sample_output[0], skip_special_tokens=True)
return(result)
user_input = st.text_area('', 'I love System Containers')
if st.button('Talk to me!'):
output = gpt2_generate(user_input,
random_seed=random_seed,
max_length=max_length,
top_p=top_p, top_k=top_k)
st.write(output)