Spaces:
Build error
Build error
tREeFrOGorigami
commited on
Commit
•
5e1514b
1
Parent(s):
b58280d
structure
Browse files- __pycache__/hello_test.cpython-310.pyc +0 -0
- app.py +161 -115
- diff_color.py +42 -0
- flagged/log.csv +2 -0
- hello_test.py +13 -0
- hf_space_test.py +20 -0
- lm-evaluation-harness +1 -0
__pycache__/hello_test.cpython-310.pyc
ADDED
Binary file (431 Bytes). View file
|
|
app.py
CHANGED
@@ -1,132 +1,178 @@
|
|
1 |
-
|
|
|
2 |
# from transformers import AutoTokenizer
|
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 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
#
|
54 |
-
|
55 |
-
#
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
#
|
61 |
-
|
62 |
-
#
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
#
|
70 |
-
|
71 |
-
|
72 |
-
#
|
73 |
-
#
|
74 |
-
#
|
75 |
-
#
|
76 |
-
#
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
#
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
|
|
|
|
90 |
|
|
|
|
|
91 |
|
92 |
-
|
|
|
|
|
|
|
|
|
|
|
93 |
|
94 |
|
95 |
|
96 |
-
|
97 |
-
|
|
|
|
|
|
|
98 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
99 |
|
100 |
-
|
101 |
-
|
102 |
-
# def predict(input_img):
|
103 |
-
# predictions = pipeline(input_img)
|
104 |
-
# return input_img, {p["label"]: p["score"] for p in predictions}
|
105 |
|
106 |
-
# gradio_app = gr.Interface(
|
107 |
-
# predict,
|
108 |
-
# inputs=gr.Image(label="Select hot dog candidate", sources=['upload', 'webcam'], type="pil"),
|
109 |
-
# outputs=[gr.Image(label="Processed Image"), gr.Label(label="Result", num_top_classes=2)],
|
110 |
-
# title="Hot Dog? Or Not?",
|
111 |
-
# )
|
112 |
|
113 |
-
# if __name__ == "__main__":
|
114 |
-
# gradio_app.launch()
|
115 |
|
116 |
|
117 |
|
118 |
-
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
119 |
|
120 |
-
|
121 |
-
return "Hello, " + name + "!" * int(intensity)
|
122 |
|
123 |
-
demo = gr.Interface(
|
124 |
-
fn=greet,
|
125 |
-
inputs=["text", "slider"],
|
126 |
-
outputs=["text"],
|
127 |
-
)
|
128 |
|
129 |
-
demo.launch(debug=True)
|
130 |
|
131 |
-
# lm-eval
|
132 |
-
# lm-evaluation-harness
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import os
|
3 |
# from transformers import AutoTokenizer
|
4 |
+
os.system('git clone https://github.com/EleutherAI/lm-evaluation-harness')
|
5 |
+
os.system('cd lm-evaluation-harness')
|
6 |
+
os.system('pip install -e .')
|
7 |
+
# 第一个功能:基于输入文本和对应的损失值对文本进行着色展示
|
8 |
+
def color_text(text_list=["hi", "FreshEval"], loss_list=[0.1,0.7]):
|
9 |
+
"""
|
10 |
+
根据损失值为文本着色。
|
11 |
+
"""
|
12 |
+
highlighted_text = []
|
13 |
+
for text, loss in zip(text_list, loss_list):
|
14 |
+
# color = "#FF0000" if float(loss) > 0.5 else "#00FF00"
|
15 |
+
color=loss
|
16 |
+
# highlighted_text.append({"text": text, "bg_color": color})
|
17 |
+
highlighted_text.append((text, color))
|
18 |
+
|
19 |
+
print(highlighted_text)
|
20 |
+
return highlighted_text
|
21 |
+
|
22 |
+
# 第二个功能:根据 ID 列表和 tokenizer 将 ID 转换为文本,并展示
|
23 |
+
def get_text(ids_list=[0.1,0.7], tokenizer=None):
|
24 |
+
"""
|
25 |
+
给定一个 ID 列表和 tokenizer 名称,将这些 ID 转换成文本。
|
26 |
+
"""
|
27 |
+
return ['Hi', 'Adam']
|
28 |
+
# tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)
|
29 |
+
# text = tokenizer.decode(eval(ids_list), skip_special_tokens=True)
|
30 |
+
# 这里只是简单地返回文本,但是可以根据实际需求添加颜色或其他样式
|
31 |
+
# return text
|
32 |
+
|
33 |
+
|
34 |
+
def get_ids_loss(text, tokenizer, model):
|
35 |
+
"""
|
36 |
+
给定一个文本,model and its tokenizer,返回其对应的 IDs 和损失值。
|
37 |
+
"""
|
38 |
+
# tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)
|
39 |
+
# model = AutoModelForCausalLM.from_pretrained(model_name)
|
40 |
+
# 这里只是简单地返回 IDs 和损失值,但是可以根据实际需求添加颜色或其他样式
|
41 |
+
return [1, 2], [0.1, 0.7]
|
42 |
+
|
43 |
+
|
44 |
+
def color_pipeline(text=["hi", "FreshEval"], model=None):
|
45 |
+
"""
|
46 |
+
给定一个文本,返回其对应的着色文本。
|
47 |
+
"""
|
48 |
+
tokenizer=None # get tokenizer
|
49 |
+
ids, loss = get_ids_loss(text, tokenizer, model)
|
50 |
+
text = get_text(ids, tokenizer)
|
51 |
+
return color_text(text, loss)
|
52 |
+
|
53 |
+
|
54 |
+
# TODO can this be global ? maybe need session to store info of the user
|
55 |
+
|
56 |
+
# 创建 Gradio 界面
|
57 |
+
with gr.Blocks() as demo:
|
58 |
+
with gr.Tab("color your text"):
|
59 |
+
with gr.Row():
|
60 |
+
text_input = gr.Textbox(label="input text", placeholder="input your text here...")
|
61 |
+
# TODO craw and drop the file
|
62 |
+
|
63 |
+
# loss_input = gr.Number(label="loss")
|
64 |
+
model_input = gr.Textbox(label="model name", placeholder="input your model name here...")
|
65 |
+
# TODO select models that can be used online
|
66 |
+
# TODO maybe add our own models
|
67 |
+
|
68 |
+
|
69 |
+
color_text_output = gr.HTML(label="colored text")
|
70 |
+
# gr.Markdown("## Text Examples")
|
71 |
+
# gr.Examples(
|
72 |
+
# [["hi", "Adam"], [0.1,0.7]],
|
73 |
+
# [text_input, loss_input],
|
74 |
+
# cache_examples=True,
|
75 |
+
# fn=color_text,
|
76 |
+
# outputs=color_text_output
|
77 |
+
# )
|
78 |
+
color_text_button = gr.Button("color the text").click(color_pipeline, inputs=[text_input, model_input], outputs=gr.HighlightedText(label="colored text"))
|
79 |
+
|
80 |
+
|
81 |
+
date_time_input = gr.Textbox(label="the date when the text is generated")#TODO add date time input
|
82 |
+
description_input = gr.Textbox(label="description of the text")
|
83 |
+
submit_button = gr.Button("submit a post or record").click()
|
84 |
+
#TODO add model and its score
|
85 |
+
|
86 |
+
with gr.Tab('test your qeustion'):
|
87 |
+
'''
|
88 |
+
use extract, or use ppl
|
89 |
+
'''
|
90 |
+
question=gr.Textbox(placeholder='input your question here...')
|
91 |
+
answer=gr.Textbox(placeholder='input your answer here...')
|
92 |
+
other_choices=gr.Textbox(placeholder='input your other choices here...')
|
93 |
|
94 |
+
test_button=gr.Button('test').click()
|
95 |
+
#TODO add the model and its score
|
96 |
|
97 |
+
def test_question(question, answer, other_choices):
|
98 |
+
'''
|
99 |
+
use extract, or use ppl
|
100 |
+
'''
|
101 |
+
answer_ppl, other_choices_ppl = get_ppl(question, answer, other_choices)
|
102 |
+
return answer_ppl, other_choices_ppl
|
103 |
|
104 |
|
105 |
|
106 |
+
with gr.Tab("model text ppl with time"):
|
107 |
+
'''
|
108 |
+
see the matplotlib example, to see ppl with time, select the models
|
109 |
+
'''
|
110 |
+
# load the json file with time,
|
111 |
|
112 |
+
|
113 |
+
with gr.Tab("model quesion acc with time"):
|
114 |
+
'''
|
115 |
+
see the matplotlib example, to see ppl with time, select the models
|
116 |
+
'''
|
117 |
+
#
|
118 |
+
|
119 |
+
|
120 |
+
with gr.Tab("hot questions"):
|
121 |
+
'''
|
122 |
+
see the questions and answers
|
123 |
+
'''
|
124 |
+
with gr.Tab("ppl"):
|
125 |
+
'''
|
126 |
+
see the questions
|
127 |
+
'''
|
128 |
+
|
129 |
|
130 |
+
demo.launch(debug=True)
|
|
|
|
|
|
|
|
|
131 |
|
|
|
|
|
|
|
|
|
|
|
|
|
132 |
|
|
|
|
|
133 |
|
134 |
|
135 |
|
136 |
+
# import gradio as gr
|
137 |
+
# import os
|
138 |
+
# os.system('python -m spacy download en_core_web_sm')
|
139 |
+
# import spacy
|
140 |
+
# from spacy import displacy
|
141 |
+
|
142 |
+
# nlp = spacy.load("en_core_web_sm")
|
143 |
+
|
144 |
+
# def text_analysis(text):
|
145 |
+
# doc = nlp(text)
|
146 |
+
# html = displacy.render(doc, style="dep", page=True)
|
147 |
+
# html = (
|
148 |
+
# "<div style='max-width:100%; max-height:360px; overflow:auto'>"
|
149 |
+
# + html
|
150 |
+
# + "</div>"
|
151 |
+
# )
|
152 |
+
# pos_count = {
|
153 |
+
# "char_count": len(text),
|
154 |
+
# "token_count": 0,
|
155 |
+
# }
|
156 |
+
# pos_tokens = []
|
157 |
+
|
158 |
+
# for token in doc:
|
159 |
+
# pos_tokens.extend([(token.text, token.pos_), (" ", None)])
|
160 |
+
|
161 |
+
# return pos_tokens, pos_count, html
|
162 |
+
|
163 |
+
# demo = gr.Interface(
|
164 |
+
# text_analysis,
|
165 |
+
# gr.Textbox(placeholder="Enter sentence here..."),
|
166 |
+
# ["highlight", "json", "html"],
|
167 |
+
# examples=[
|
168 |
+
# ["What a beautiful morning for a walk!"],
|
169 |
+
# ["It was the best of times, it was the worst of times."],
|
170 |
+
# ],
|
171 |
+
# )
|
172 |
|
173 |
+
# demo.launch()
|
|
|
174 |
|
|
|
|
|
|
|
|
|
|
|
175 |
|
|
|
176 |
|
177 |
+
# # lm-eval
|
178 |
+
# # lm-evaluation-harness
|
diff_color.py
ADDED
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
from difflib import Differ
|
3 |
+
|
4 |
+
import gradio as gr
|
5 |
+
|
6 |
+
|
7 |
+
def diff_texts(text1, text2):
|
8 |
+
d = Differ()
|
9 |
+
rtn =[
|
10 |
+
(token[2:], token[0] if token[0] != " " else None)
|
11 |
+
for token in d.compare(text1, text2)
|
12 |
+
]
|
13 |
+
print(rtn)
|
14 |
+
return rtn
|
15 |
+
|
16 |
+
|
17 |
+
demo = gr.Interface(
|
18 |
+
diff_texts,
|
19 |
+
[
|
20 |
+
gr.Textbox(
|
21 |
+
label="Text 1",
|
22 |
+
info="Initial text",
|
23 |
+
lines=3,
|
24 |
+
value="The quick brown fox jumped over the lazy dogs.",
|
25 |
+
),
|
26 |
+
gr.Textbox(
|
27 |
+
label="Text 2",
|
28 |
+
info="Text to compare",
|
29 |
+
lines=3,
|
30 |
+
value="The fast brown fox jumps over lazy dogs.",
|
31 |
+
),
|
32 |
+
],
|
33 |
+
gr.HighlightedText(
|
34 |
+
label="Diff",
|
35 |
+
combine_adjacent=True,
|
36 |
+
show_legend=True,
|
37 |
+
color_map={"+": "red", "-": "green"}),
|
38 |
+
theme=gr.themes.Base()# the return is here
|
39 |
+
)
|
40 |
+
if __name__ == "__main__":
|
41 |
+
demo.launch()
|
42 |
+
|
flagged/log.csv
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
Text 1,Text 2,Diff,flag,username,timestamp
|
2 |
+
The quick brown fox jumped over the lazy dogs.,The fast brown fox jumps over lazy dogs.,,,,2024-03-13 16:50:01.853095
|
hello_test.py
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
|
3 |
+
def greet(name, intensity):
|
4 |
+
return "Hello, " + name + ",,!" * int(intensity)
|
5 |
+
# return "Hello, " + name + ",,!" * int(0/int(intensity))# you can see the bug in command line
|
6 |
+
|
7 |
+
demo = gr.Interface(
|
8 |
+
fn=greet,
|
9 |
+
inputs=["text", "slider"],
|
10 |
+
outputs=["text"],
|
11 |
+
)
|
12 |
+
|
13 |
+
demo.launch(debug=True)
|
hf_space_test.py
ADDED
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# this need hugginface connection
|
2 |
+
import gradio as gr
|
3 |
+
from transformers import pipeline
|
4 |
+
|
5 |
+
|
6 |
+
pipeline = pipeline(task="image-classification", model="julien-c/hotdog-not-hotdog")
|
7 |
+
|
8 |
+
def predict(input_img):
|
9 |
+
predictions = pipeline(input_img)
|
10 |
+
return input_img, {p["label"]: p["score"] for p in predictions}
|
11 |
+
|
12 |
+
gradio_app = gr.Interface(
|
13 |
+
predict,
|
14 |
+
inputs=gr.Image(label="Select hot dog candidate", sources=['upload', 'webcam'], type="pil"),
|
15 |
+
outputs=[gr.Image(label="Processed Image"), gr.Label(label="Result", num_top_classes=2)],
|
16 |
+
title="Hot Dog? Or Not?",
|
17 |
+
)
|
18 |
+
|
19 |
+
if __name__ == "__main__":
|
20 |
+
gradio_app.launch(debug=True)
|
lm-evaluation-harness
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
Subproject commit 49695e8d94c3ab011b7ae8814d809de30b1b1182
|