Update app.py
Browse files
app.py
CHANGED
@@ -1,25 +1,73 @@
|
|
1 |
-
import
|
2 |
-
from transformers import
|
3 |
|
4 |
MODEL_NAME_OR_PATH = "flax-community/t5-recipe-generation"
|
5 |
-
tokenizer =
|
6 |
-
model =
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
23 |
|
24 |
if __name__ == "__main__":
|
25 |
-
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from transformers import FlaxAutoModelForSeq2SeqLM, AutoTokenizer
|
3 |
|
4 |
MODEL_NAME_OR_PATH = "flax-community/t5-recipe-generation"
|
5 |
+
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME_OR_PATH, use_fast=True)
|
6 |
+
model = FlaxAutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME_OR_PATH)
|
7 |
+
|
8 |
+
prefix = "items: "
|
9 |
+
generation_kwargs = {
|
10 |
+
"max_length": 512,
|
11 |
+
"min_length": 64,
|
12 |
+
"no_repeat_ngram_size": 3,
|
13 |
+
"do_sample": True,
|
14 |
+
"top_k": 60,
|
15 |
+
"top_p": 0.95,
|
16 |
+
}
|
17 |
+
|
18 |
+
special_tokens = tokenizer.all_special_tokens
|
19 |
+
tokens_map = {
|
20 |
+
"<sep>": "--",
|
21 |
+
"<section>": "\n",
|
22 |
+
}
|
23 |
+
|
24 |
+
def skip_special_tokens(text, special_tokens):
|
25 |
+
for token in special_tokens:
|
26 |
+
text = text.replace(token, "")
|
27 |
+
return text
|
28 |
+
|
29 |
+
def target_postprocessing(texts, special_tokens):
|
30 |
+
if not isinstance(texts, list):
|
31 |
+
texts = [texts]
|
32 |
+
new_texts = []
|
33 |
+
for text in texts:
|
34 |
+
text = skip_special_tokens(text, special_tokens)
|
35 |
+
for k, v in tokens_map.items():
|
36 |
+
text = text.replace(k, v)
|
37 |
+
new_texts.append(text)
|
38 |
+
return new_texts
|
39 |
+
|
40 |
+
def generation_function(texts):
|
41 |
+
_inputs = texts if isinstance(texts, list) else [texts]
|
42 |
+
inputs = [prefix + inp for inp in _inputs]
|
43 |
+
inputs = tokenizer(
|
44 |
+
inputs,
|
45 |
+
max_length=256,
|
46 |
+
padding="max_length",
|
47 |
+
truncation=True,
|
48 |
+
return_tensors="jax",
|
49 |
+
)
|
50 |
+
input_ids = inputs.input_ids
|
51 |
+
attention_mask = inputs.attention_mask
|
52 |
+
output_ids = model.generate(
|
53 |
+
input_ids=input_ids,
|
54 |
+
attention_mask=attention_mask,
|
55 |
+
**generation_kwargs,
|
56 |
+
)
|
57 |
+
generated = output_ids.sequences
|
58 |
+
generated_recipe = target_postprocessing(
|
59 |
+
tokenizer.batch_decode(generated, skip_special_tokens=False),
|
60 |
+
special_tokens,
|
61 |
+
)
|
62 |
+
return generated_recipe[0] # Return the first generated recipe as a string
|
63 |
+
|
64 |
+
iface = gr.Interface(
|
65 |
+
fn=generation_function,
|
66 |
+
inputs="text",
|
67 |
+
outputs="text",
|
68 |
+
title="Recipe Generation",
|
69 |
+
description="Generate a recipe based on an input text.",
|
70 |
+
)
|
71 |
|
72 |
if __name__ == "__main__":
|
73 |
+
iface.launch()
|