Spaces:
Sleeping
Sleeping
Prgckwb
commited on
Commit
•
d9d3f4b
1
Parent(s):
e603ef9
:tada: init
Browse files
app.py
CHANGED
@@ -6,13 +6,15 @@ from transformers import AutoTokenizer, CLIPTokenizerFast, T5TokenizerFast
|
|
6 |
|
7 |
def load_tokenizers(model_id: str) -> list[CLIPTokenizerFast | T5TokenizerFast | None]:
|
8 |
config = DiffusionPipeline.load_config(model_id)
|
9 |
-
num_tokenizers = sum(
|
10 |
|
11 |
if not 1 <= num_tokenizers <= 3:
|
12 |
-
raise gr.Error(f
|
13 |
|
14 |
tokenizers = [
|
15 |
-
AutoTokenizer.from_pretrained(
|
|
|
|
|
16 |
for i in range(num_tokenizers)
|
17 |
]
|
18 |
|
@@ -22,7 +24,7 @@ def load_tokenizers(model_id: str) -> list[CLIPTokenizerFast | T5TokenizerFast |
|
|
22 |
return tokenizers
|
23 |
|
24 |
|
25 |
-
@torch.
|
26 |
def inference(model_id: str, input_text: str):
|
27 |
tokenizers = load_tokenizers(model_id)
|
28 |
|
@@ -30,17 +32,19 @@ def inference(model_id: str, input_text: str):
|
|
30 |
special_tokens_components = []
|
31 |
for i, tokenizer in enumerate(tokenizers):
|
32 |
if tokenizer:
|
33 |
-
label_text = f
|
34 |
|
35 |
# テキストとトークンIDのペアを作成
|
36 |
input_ids = tokenizer(
|
37 |
text=input_text,
|
38 |
truncation=True,
|
39 |
return_length=False,
|
40 |
-
return_overflowing_tokens=False
|
41 |
).input_ids
|
42 |
decoded_tokens = [tokenizer.decode(id_) for id_ in input_ids]
|
43 |
-
token_pairs = [
|
|
|
|
|
44 |
output_text_pair_component = gr.HighlightedText(
|
45 |
label=label_text,
|
46 |
value=token_pairs,
|
@@ -51,7 +55,7 @@ def inference(model_id: str, input_text: str):
|
|
51 |
# スペシャルトークンを追加
|
52 |
special_tokens = []
|
53 |
for k, v in tokenizer.special_tokens_map.items():
|
54 |
-
if k ==
|
55 |
continue
|
56 |
special_token_map = (str(k), str(v))
|
57 |
special_tokens.append(special_token_map)
|
@@ -71,7 +75,7 @@ def inference(model_id: str, input_text: str):
|
|
71 |
return text_pairs_components + special_tokens_components
|
72 |
|
73 |
|
74 |
-
if __name__ ==
|
75 |
theme = gr.themes.Soft(
|
76 |
primary_hue=gr.themes.colors.emerald,
|
77 |
secondary_hue=gr.themes.colors.emerald,
|
@@ -79,29 +83,29 @@ if __name__ == '__main__':
|
|
79 |
with gr.Blocks(theme=theme) as demo:
|
80 |
with gr.Column():
|
81 |
input_model_id = gr.Dropdown(
|
82 |
-
label=
|
83 |
choices=[
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
],
|
92 |
-
value=
|
93 |
)
|
94 |
input_text = gr.Textbox(
|
95 |
-
label=
|
96 |
-
placeholder=
|
97 |
)
|
98 |
|
99 |
-
with gr.Tab(label=
|
100 |
with gr.Column():
|
101 |
output_highlighted_text_1 = gr.HighlightedText()
|
102 |
output_highlighted_text_2 = gr.HighlightedText()
|
103 |
output_highlighted_text_3 = gr.HighlightedText()
|
104 |
-
with gr.Tab(label=
|
105 |
with gr.Column():
|
106 |
output_special_tokens_1 = gr.HighlightedText()
|
107 |
output_special_tokens_2 = gr.HighlightedText()
|
@@ -109,7 +113,7 @@ if __name__ == '__main__':
|
|
109 |
|
110 |
with gr.Row():
|
111 |
clear_button = gr.ClearButton(components=[input_text])
|
112 |
-
submit_button = gr.Button(
|
113 |
|
114 |
all_inputs = [input_model_id, input_text]
|
115 |
all_output = [
|
@@ -120,22 +124,21 @@ if __name__ == '__main__':
|
|
120 |
output_special_tokens_2,
|
121 |
output_special_tokens_3,
|
122 |
]
|
123 |
-
submit_button.click(
|
124 |
-
fn=inference,
|
125 |
-
inputs=all_inputs,
|
126 |
-
outputs=all_output
|
127 |
-
)
|
128 |
|
129 |
examples = gr.Examples(
|
130 |
fn=inference,
|
131 |
inputs=all_inputs,
|
132 |
outputs=all_output,
|
133 |
examples=[
|
134 |
-
[
|
135 |
-
[
|
136 |
-
|
|
|
|
|
|
|
137 |
],
|
138 |
-
cache_examples=True
|
139 |
)
|
140 |
|
141 |
demo.queue().launch()
|
|
|
6 |
|
7 |
def load_tokenizers(model_id: str) -> list[CLIPTokenizerFast | T5TokenizerFast | None]:
|
8 |
config = DiffusionPipeline.load_config(model_id)
|
9 |
+
num_tokenizers = sum("tokenizer" in key for key in config.keys())
|
10 |
|
11 |
if not 1 <= num_tokenizers <= 3:
|
12 |
+
raise gr.Error(f"Invalid number of tokenizers: {num_tokenizers}")
|
13 |
|
14 |
tokenizers = [
|
15 |
+
AutoTokenizer.from_pretrained(
|
16 |
+
model_id, subfolder=f'tokenizer{"" if i == 0 else f"_{i + 1}"}'
|
17 |
+
)
|
18 |
for i in range(num_tokenizers)
|
19 |
]
|
20 |
|
|
|
24 |
return tokenizers
|
25 |
|
26 |
|
27 |
+
@torch.no_grad()
|
28 |
def inference(model_id: str, input_text: str):
|
29 |
tokenizers = load_tokenizers(model_id)
|
30 |
|
|
|
32 |
special_tokens_components = []
|
33 |
for i, tokenizer in enumerate(tokenizers):
|
34 |
if tokenizer:
|
35 |
+
label_text = f"Tokenizer {i + 1}: {tokenizer.__class__.__name__}"
|
36 |
|
37 |
# テキストとトークンIDのペアを作成
|
38 |
input_ids = tokenizer(
|
39 |
text=input_text,
|
40 |
truncation=True,
|
41 |
return_length=False,
|
42 |
+
return_overflowing_tokens=False,
|
43 |
).input_ids
|
44 |
decoded_tokens = [tokenizer.decode(id_) for id_ in input_ids]
|
45 |
+
token_pairs = [
|
46 |
+
(str(token), str(id_)) for token, id_ in zip(decoded_tokens, input_ids)
|
47 |
+
]
|
48 |
output_text_pair_component = gr.HighlightedText(
|
49 |
label=label_text,
|
50 |
value=token_pairs,
|
|
|
55 |
# スペシャルトークンを追加
|
56 |
special_tokens = []
|
57 |
for k, v in tokenizer.special_tokens_map.items():
|
58 |
+
if k == "additional_special_tokens":
|
59 |
continue
|
60 |
special_token_map = (str(k), str(v))
|
61 |
special_tokens.append(special_token_map)
|
|
|
75 |
return text_pairs_components + special_tokens_components
|
76 |
|
77 |
|
78 |
+
if __name__ == "__main__":
|
79 |
theme = gr.themes.Soft(
|
80 |
primary_hue=gr.themes.colors.emerald,
|
81 |
secondary_hue=gr.themes.colors.emerald,
|
|
|
83 |
with gr.Blocks(theme=theme) as demo:
|
84 |
with gr.Column():
|
85 |
input_model_id = gr.Dropdown(
|
86 |
+
label="Model ID",
|
87 |
choices=[
|
88 |
+
"black-forest-labs/FLUX.1-dev",
|
89 |
+
"black-forest-labs/FLUX.1-schnell",
|
90 |
+
"stabilityai/stable-diffusion-3-medium-diffusers",
|
91 |
+
"stabilityai/stable-diffusion-xl-base-1.0",
|
92 |
+
"stable-diffusion-v1-5/stable-diffusion-v1-5",
|
93 |
+
"stabilityai/japanese-stable-diffusion-xl",
|
94 |
+
"rinna/japanese-stable-diffusion",
|
95 |
],
|
96 |
+
value="black-forest-labs/FLUX.1-dev",
|
97 |
)
|
98 |
input_text = gr.Textbox(
|
99 |
+
label="Input Text",
|
100 |
+
placeholder="Enter text here",
|
101 |
)
|
102 |
|
103 |
+
with gr.Tab(label="Tokenization Outputs"):
|
104 |
with gr.Column():
|
105 |
output_highlighted_text_1 = gr.HighlightedText()
|
106 |
output_highlighted_text_2 = gr.HighlightedText()
|
107 |
output_highlighted_text_3 = gr.HighlightedText()
|
108 |
+
with gr.Tab(label="Special Tokens"):
|
109 |
with gr.Column():
|
110 |
output_special_tokens_1 = gr.HighlightedText()
|
111 |
output_special_tokens_2 = gr.HighlightedText()
|
|
|
113 |
|
114 |
with gr.Row():
|
115 |
clear_button = gr.ClearButton(components=[input_text])
|
116 |
+
submit_button = gr.Button("Run", variant="primary")
|
117 |
|
118 |
all_inputs = [input_model_id, input_text]
|
119 |
all_output = [
|
|
|
124 |
output_special_tokens_2,
|
125 |
output_special_tokens_3,
|
126 |
]
|
127 |
+
submit_button.click(fn=inference, inputs=all_inputs, outputs=all_output)
|
|
|
|
|
|
|
|
|
128 |
|
129 |
examples = gr.Examples(
|
130 |
fn=inference,
|
131 |
inputs=all_inputs,
|
132 |
outputs=all_output,
|
133 |
examples=[
|
134 |
+
["black-forest-labs/FLUX.1-dev", "a photo of cat"],
|
135 |
+
[
|
136 |
+
"stabilityai/stable-diffusion-3-medium-diffusers",
|
137 |
+
'cat holding sign saying "I am a cat"',
|
138 |
+
],
|
139 |
+
["rinna/japanese-stable-diffusion", "空を飛んでいるネコの写真 油絵"],
|
140 |
],
|
141 |
+
cache_examples=True,
|
142 |
)
|
143 |
|
144 |
demo.queue().launch()
|