update
Browse files- generate_poem.py +65 -3
generate_poem.py
CHANGED
|
@@ -1,14 +1,76 @@
|
|
|
|
|
| 1 |
import gradio as gr
|
|
|
|
| 2 |
|
|
|
|
|
|
|
| 3 |
|
| 4 |
-
|
| 5 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
|
| 7 |
|
| 8 |
generate_poem_interface = gr.Interface(
|
|
|
|
| 9 |
fn=generate_poem,
|
| 10 |
inputs=[
|
| 11 |
-
gr.components.Textbox(
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
],
|
| 13 |
outputs="text",
|
| 14 |
)
|
|
|
|
| 1 |
+
import torch
|
| 2 |
import gradio as gr
|
| 3 |
+
from transformers import T5Tokenizer, T5ForConditionalGeneration
|
| 4 |
|
| 5 |
+
tokenizer = T5Tokenizer.from_pretrained("VietAI/vit5-base")
|
| 6 |
+
model = T5ForConditionalGeneration.from_pretrained("Libosa2707/vietnamese-poem-t5")
|
| 7 |
|
| 8 |
+
|
| 9 |
+
def generate_poem(input_text):
|
| 10 |
+
# Define the parameters for the generate function
|
| 11 |
+
min_length = 50
|
| 12 |
+
max_length = 100
|
| 13 |
+
rep_penalty = 1.2
|
| 14 |
+
temp = 0.7
|
| 15 |
+
top_k = 50
|
| 16 |
+
top_p = 0.92
|
| 17 |
+
no_repeat_ngram_size = 2
|
| 18 |
+
|
| 19 |
+
# Tokenize the input
|
| 20 |
+
input_ids = tokenizer(
|
| 21 |
+
input_text,
|
| 22 |
+
return_tensors="pt",
|
| 23 |
+
padding="max_length",
|
| 24 |
+
truncation=True,
|
| 25 |
+
max_length=42,
|
| 26 |
+
).input_ids.to(model.device)
|
| 27 |
+
|
| 28 |
+
# Generate text
|
| 29 |
+
model.eval()
|
| 30 |
+
with torch.no_grad():
|
| 31 |
+
output = model.generate(
|
| 32 |
+
do_sample=True,
|
| 33 |
+
input_ids=input_ids,
|
| 34 |
+
min_length=min_length,
|
| 35 |
+
max_length=max_length,
|
| 36 |
+
top_p=top_p,
|
| 37 |
+
top_k=top_k,
|
| 38 |
+
temperature=temp,
|
| 39 |
+
repetition_penalty=rep_penalty,
|
| 40 |
+
no_repeat_ngram_size=no_repeat_ngram_size,
|
| 41 |
+
num_return_sequences=1,
|
| 42 |
+
)
|
| 43 |
+
|
| 44 |
+
# Process the generated text
|
| 45 |
+
gen = tokenizer.decode(
|
| 46 |
+
output[0], skip_special_tokens=False, clean_up_tokenization_spaces=False
|
| 47 |
+
)
|
| 48 |
+
sentences = gen.split("<unk>")
|
| 49 |
+
gen_poem = "\n".join(sentences).replace("<pad>", "").replace("</s>", "")
|
| 50 |
+
gen_poem = gen_poem.strip()
|
| 51 |
+
|
| 52 |
+
# Post-process the poem text
|
| 53 |
+
pretty_text = ""
|
| 54 |
+
for line in gen_poem.split("\n"):
|
| 55 |
+
line = line.strip()
|
| 56 |
+
if not line:
|
| 57 |
+
continue
|
| 58 |
+
line = line[0].upper() + line[1:]
|
| 59 |
+
pretty_text += line + "\n"
|
| 60 |
+
|
| 61 |
+
# Return the generated poem
|
| 62 |
+
return pretty_text
|
| 63 |
|
| 64 |
|
| 65 |
generate_poem_interface = gr.Interface(
|
| 66 |
+
title="Làm thơ theo yêu cầu",
|
| 67 |
fn=generate_poem,
|
| 68 |
inputs=[
|
| 69 |
+
gr.components.Textbox(
|
| 70 |
+
lines=1,
|
| 71 |
+
placeholder="Làm thơ với thể thơ tám chữ và tiêu đề mùa xuân nho nhỏ",
|
| 72 |
+
label="Yêu cầu về thể thơ và tiêu đề",
|
| 73 |
+
),
|
| 74 |
],
|
| 75 |
outputs="text",
|
| 76 |
)
|