Spaces:
Running
Running
jbdel
commited on
Commit
•
8c5c31d
1
Parent(s):
2886238
chat_with_paper_update
Browse files- README.md +1 -1
- app.py +13 -10
- paper_chat_tab.py +240 -193
README.md
CHANGED
@@ -5,7 +5,7 @@ emoji: ⚡
|
|
5 |
colorFrom: red
|
6 |
colorTo: purple
|
7 |
sdk: gradio
|
8 |
-
sdk_version: 5.
|
9 |
app_file: app.py
|
10 |
pinned: false
|
11 |
header: mini
|
|
|
5 |
colorFrom: red
|
6 |
colorTo: purple
|
7 |
sdk: gradio
|
8 |
+
sdk_version: 5.8.0
|
9 |
app_file: app.py
|
10 |
pinned: false
|
11 |
header: mini
|
app.py
CHANGED
@@ -82,6 +82,12 @@ with gr.Blocks(css_paths="style.css") as demo:
|
|
82 |
link="https://huggingface.co/datasets/huggingface/paper-central-data")
|
83 |
|
84 |
with gr.Tabs() as tabs:
|
|
|
|
|
|
|
|
|
|
|
|
|
85 |
with gr.Tab("Paper-central", id="tab-paper-central"):
|
86 |
# Create a row for navigation buttons and calendar
|
87 |
with gr.Row():
|
@@ -178,6 +184,8 @@ with gr.Blocks(css_paths="style.css") as demo:
|
|
178 |
wrap=True,
|
179 |
)
|
180 |
|
|
|
|
|
181 |
with gr.Tab("Edit papers", id="tab-pr"):
|
182 |
pr_paper_central_tab(paper_central_df.df_raw)
|
183 |
|
@@ -187,19 +195,13 @@ with gr.Blocks(css_paths="style.css") as demo:
|
|
187 |
with gr.Tab("Contributors"):
|
188 |
author_resource_leaderboard_tab()
|
189 |
|
190 |
-
with gr.Tab("Chat With Paper", id="tab-chat-with-paper", visible=False) as tab_chat_paper:
|
191 |
-
gr.Markdown("## Chat with Paper")
|
192 |
-
arxiv_id = gr.State(value=None)
|
193 |
-
paper_from = gr.State(value=None)
|
194 |
-
paper_chat_tab(arxiv_id, paper_from)
|
195 |
-
|
196 |
|
197 |
# chat with paper
|
198 |
def get_selected(evt: gr.SelectData, dataframe_origin):
|
199 |
|
200 |
paper_id = gr.update(value=None)
|
201 |
paper_from = gr.update(value=None)
|
202 |
-
tab_chat_paper = gr.update(visible=
|
203 |
selected_tab = gr.Tabs()
|
204 |
|
205 |
try:
|
@@ -516,7 +518,7 @@ with gr.Blocks(css_paths="style.css") as demo:
|
|
516 |
selected_tab = gr.Tabs()
|
517 |
paper_id = gr.update(value=None)
|
518 |
paper_from = gr.update(value=None)
|
519 |
-
tab_chat_paper = gr.update(visible=
|
520 |
|
521 |
if request:
|
522 |
# print("Request headers dictionary:", dict(request.headers))
|
@@ -568,7 +570,8 @@ with gr.Blocks(css_paths="style.css") as demo:
|
|
568 |
api_name="update_data",
|
569 |
).then(
|
570 |
fn=echo,
|
571 |
-
outputs=[calendar, date_range_radio, conference_options, hf_options, tabs, arxiv_id, paper_from,
|
|
|
572 |
api_name=False,
|
573 |
).then(
|
574 |
# New then to handle LoginButton and HTML components
|
@@ -583,7 +586,7 @@ def main():
|
|
583 |
"""
|
584 |
Launches the Gradio app.
|
585 |
"""
|
586 |
-
demo.launch(
|
587 |
|
588 |
|
589 |
# Run the main function when the script is executed
|
|
|
82 |
link="https://huggingface.co/datasets/huggingface/paper-central-data")
|
83 |
|
84 |
with gr.Tabs() as tabs:
|
85 |
+
with gr.Tab("Chat With Paper", id="tab-chat-with-paper", visible=True) as tab_chat_paper:
|
86 |
+
gr.Markdown("## Chat with Paper")
|
87 |
+
arxiv_id = gr.State(value=None)
|
88 |
+
paper_from = gr.State(value=None)
|
89 |
+
paper_chat_tab(arxiv_id, paper_from, paper_central_df)
|
90 |
+
|
91 |
with gr.Tab("Paper-central", id="tab-paper-central"):
|
92 |
# Create a row for navigation buttons and calendar
|
93 |
with gr.Row():
|
|
|
184 |
wrap=True,
|
185 |
)
|
186 |
|
187 |
+
|
188 |
+
|
189 |
with gr.Tab("Edit papers", id="tab-pr"):
|
190 |
pr_paper_central_tab(paper_central_df.df_raw)
|
191 |
|
|
|
195 |
with gr.Tab("Contributors"):
|
196 |
author_resource_leaderboard_tab()
|
197 |
|
|
|
|
|
|
|
|
|
|
|
|
|
198 |
|
199 |
# chat with paper
|
200 |
def get_selected(evt: gr.SelectData, dataframe_origin):
|
201 |
|
202 |
paper_id = gr.update(value=None)
|
203 |
paper_from = gr.update(value=None)
|
204 |
+
tab_chat_paper = gr.update(visible=True)
|
205 |
selected_tab = gr.Tabs()
|
206 |
|
207 |
try:
|
|
|
518 |
selected_tab = gr.Tabs()
|
519 |
paper_id = gr.update(value=None)
|
520 |
paper_from = gr.update(value=None)
|
521 |
+
tab_chat_paper = gr.update(visible=True)
|
522 |
|
523 |
if request:
|
524 |
# print("Request headers dictionary:", dict(request.headers))
|
|
|
570 |
api_name="update_data",
|
571 |
).then(
|
572 |
fn=echo,
|
573 |
+
outputs=[calendar, date_range_radio, conference_options, hf_options, tabs, arxiv_id, paper_from,
|
574 |
+
tab_chat_paper],
|
575 |
api_name=False,
|
576 |
).then(
|
577 |
# New then to handle LoginButton and HTML components
|
|
|
586 |
"""
|
587 |
Launches the Gradio app.
|
588 |
"""
|
589 |
+
demo.launch(share=True)
|
590 |
|
591 |
|
592 |
# Run the main function when the script is executed
|
paper_chat_tab.py
CHANGED
@@ -7,9 +7,10 @@ import requests
|
|
7 |
from io import BytesIO
|
8 |
from transformers import AutoTokenizer
|
9 |
import json
|
10 |
-
|
11 |
import os
|
12 |
from openai import OpenAI
|
|
|
13 |
|
14 |
# Cache for tokenizers to avoid reloading
|
15 |
tokenizer_cache = {}
|
@@ -23,7 +24,6 @@ PROVIDERS = {
|
|
23 |
"api_key_env_var": "SAMBANOVA_API_KEY",
|
24 |
"models": [
|
25 |
"Meta-Llama-3.1-70B-Instruct",
|
26 |
-
# Add more models if needed
|
27 |
],
|
28 |
"type": "tuples",
|
29 |
"max_total_tokens": "50000",
|
@@ -43,12 +43,12 @@ PROVIDERS = {
|
|
43 |
}
|
44 |
|
45 |
|
46 |
-
#
|
47 |
def fetch_paper_info_neurips(paper_id):
|
48 |
url = f"https://openreview.net/forum?id={paper_id}"
|
49 |
response = requests.get(url)
|
50 |
if response.status_code != 200:
|
51 |
-
return None
|
52 |
|
53 |
html_content = response.content
|
54 |
soup = BeautifulSoup(html_content, 'html.parser')
|
@@ -73,66 +73,104 @@ def fetch_paper_info_neurips(paper_id):
|
|
73 |
else:
|
74 |
abstract = 'Abstract not found'
|
75 |
|
76 |
-
# Construct preamble
|
77 |
-
|
|
|
78 |
|
79 |
-
return preamble
|
80 |
|
81 |
-
|
82 |
-
def fetch_paper_content_arxiv(paper_id):
|
83 |
try:
|
84 |
-
|
85 |
-
url = f"https://arxiv.org/pdf/{paper_id}.pdf"
|
86 |
-
|
87 |
-
# Fetch the PDF
|
88 |
response = requests.get(url)
|
89 |
-
response.raise_for_status()
|
90 |
-
|
91 |
-
# Read the PDF content
|
92 |
pdf_content = BytesIO(response.content)
|
93 |
reader = PdfReader(pdf_content)
|
94 |
-
|
95 |
-
# Extract text from the PDF
|
96 |
text = ""
|
97 |
for page in reader.pages:
|
98 |
text += page.extract_text()
|
99 |
-
|
100 |
-
|
101 |
-
except Exception as e:
|
102 |
-
print(f"Error fetching paper content: {e}")
|
103 |
return None
|
104 |
|
105 |
|
106 |
-
def
|
107 |
try:
|
108 |
-
|
109 |
-
url = f"https://openreview.net/pdf?id={paper_id}"
|
110 |
-
|
111 |
-
# Fetch the PDF
|
112 |
response = requests.get(url)
|
113 |
-
response.raise_for_status()
|
114 |
-
|
115 |
-
# Read the PDF content
|
116 |
pdf_content = BytesIO(response.content)
|
117 |
reader = PdfReader(pdf_content)
|
118 |
-
|
119 |
-
# Extract text from the PDF
|
120 |
text = ""
|
121 |
for page in reader.pages:
|
122 |
text += page.extract_text()
|
123 |
-
|
124 |
-
return text # Return full text; truncation will be handled later
|
125 |
-
|
126 |
except Exception as e:
|
127 |
-
print(f"
|
128 |
return None
|
129 |
|
130 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
131 |
def create_chat_interface(provider_dropdown, model_dropdown, paper_content, hf_token_input, default_type,
|
132 |
provider_max_total_tokens):
|
133 |
# Define the function to handle the chat
|
134 |
-
print("the type is", default_type.value)
|
135 |
-
|
136 |
def get_fn(message, history, paper_content_value, hf_token_value, provider_name_value, model_name_value,
|
137 |
max_total_tokens):
|
138 |
provider_info = PROVIDERS[provider_name_value]
|
@@ -141,11 +179,9 @@ def create_chat_interface(provider_dropdown, model_dropdown, paper_content, hf_t
|
|
141 |
models = provider_info['models']
|
142 |
max_total_tokens = int(max_total_tokens)
|
143 |
|
144 |
-
# Load tokenizer
|
145 |
tokenizer_key = f"{provider_name_value}_{model_name_value}"
|
146 |
if tokenizer_key not in tokenizer_cache:
|
147 |
-
# Load the tokenizer; adjust the model path based on the provider and model
|
148 |
-
# This is a placeholder; you need to provide the correct tokenizer path
|
149 |
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.2-1B-Instruct",
|
150 |
token=os.environ.get("HF_TOKEN"))
|
151 |
tokenizer_cache[tokenizer_key] = tokenizer
|
@@ -189,32 +225,28 @@ def create_chat_interface(provider_dropdown, model_dropdown, paper_content, hf_t
|
|
189 |
|
190 |
# Check if total tokens exceed the maximum allowed tokens
|
191 |
if total_tokens > max_total_tokens:
|
192 |
-
# Attempt to truncate
|
193 |
available_tokens = max_total_tokens - (total_tokens - context_token_length)
|
194 |
if available_tokens > 0:
|
195 |
-
# Truncate the context to fit the available tokens
|
196 |
truncated_context_tokens = context_tokens[:available_tokens]
|
197 |
context = tokenizer.decode(truncated_context_tokens)
|
198 |
context_token_length = available_tokens
|
199 |
total_tokens = total_tokens - len(context_tokens) + context_token_length
|
200 |
else:
|
201 |
-
# Not enough space for context; remove it
|
202 |
context = ""
|
203 |
total_tokens -= context_token_length
|
204 |
context_token_length = 0
|
205 |
|
206 |
-
#
|
207 |
while total_tokens > max_total_tokens and len(messages) > 1:
|
208 |
-
# Remove the oldest message
|
209 |
removed_message = messages.pop(0)
|
210 |
removed_tokens = message_tokens_list.pop(0)
|
211 |
total_tokens -= removed_tokens
|
212 |
|
213 |
-
# Rebuild the final messages
|
214 |
final_messages = []
|
215 |
if context:
|
216 |
-
final_messages.append(
|
217 |
-
{"role": "system", "content": f"{context}"})
|
218 |
final_messages.extend(messages)
|
219 |
|
220 |
# Use the provider's API key
|
@@ -222,14 +254,13 @@ def create_chat_interface(provider_dropdown, model_dropdown, paper_content, hf_t
|
|
222 |
if not api_key:
|
223 |
raise ValueError("API token is not provided.")
|
224 |
|
225 |
-
# Initialize the OpenAI client
|
226 |
client = OpenAI(
|
227 |
base_url=endpoint,
|
228 |
api_key=api_key,
|
229 |
)
|
230 |
|
231 |
try:
|
232 |
-
# Create the chat completion
|
233 |
completion = client.chat.completions.create(
|
234 |
model=model_name_value,
|
235 |
messages=final_messages,
|
@@ -241,29 +272,13 @@ def create_chat_interface(provider_dropdown, model_dropdown, paper_content, hf_t
|
|
241 |
response_text += delta
|
242 |
yield response_text
|
243 |
except json.JSONDecodeError as e:
|
244 |
-
|
245 |
-
print(f"Error Message: {e.msg}")
|
246 |
-
print(f"Error Position: Line {e.lineno}, Column {e.colno} (Character {e.pos})")
|
247 |
-
print(f"Problematic JSON Data: {e.doc}")
|
248 |
-
yield f"{e.doc}"
|
249 |
except openai.OpenAIError as openai_err:
|
250 |
-
|
251 |
-
print(f"An OpenAI error occurred: {openai_err}")
|
252 |
-
yield f"{openai_err}"
|
253 |
except Exception as ex:
|
254 |
-
|
255 |
-
print(f"An unexpected error occurred: {ex}")
|
256 |
-
yield f"{ex}"
|
257 |
-
|
258 |
-
# Create the Chatbot separately to access it later
|
259 |
-
chatbot = gr.Chatbot(
|
260 |
-
label="Chatbot",
|
261 |
-
scale=1,
|
262 |
-
height=400,
|
263 |
-
autoscroll=True,
|
264 |
-
)
|
265 |
|
266 |
-
|
267 |
chat_interface = gr.ChatInterface(
|
268 |
fn=get_fn,
|
269 |
chatbot=chatbot,
|
@@ -273,142 +288,164 @@ def create_chat_interface(provider_dropdown, model_dropdown, paper_content, hf_t
|
|
273 |
return chat_interface, chatbot
|
274 |
|
275 |
|
276 |
-
def paper_chat_tab(paper_id, paper_from):
|
277 |
-
with gr.
|
278 |
-
|
279 |
-
|
280 |
-
|
|
|
281 |
|
282 |
-
|
283 |
-
|
284 |
-
|
|
|
|
|
|
|
|
|
285 |
|
286 |
-
|
287 |
-
|
288 |
|
289 |
-
|
290 |
-
|
291 |
-
choices=provider_names,
|
292 |
-
value=default_provider
|
293 |
-
)
|
294 |
|
295 |
-
|
296 |
-
|
297 |
-
|
298 |
-
|
299 |
-
|
300 |
-
|
301 |
-
# Dropdown for selecting the model
|
302 |
-
model_dropdown = gr.Dropdown(
|
303 |
-
label="Select Model",
|
304 |
-
choices=PROVIDERS[default_provider]['models'],
|
305 |
-
value=PROVIDERS[default_provider]['models'][0]
|
306 |
-
)
|
307 |
|
308 |
-
|
309 |
-
|
310 |
-
value=f'<img src="{PROVIDERS[default_provider]["logo"]}" width="100px" />'
|
311 |
-
)
|
312 |
|
313 |
-
#
|
314 |
-
|
|
|
|
|
|
|
315 |
|
316 |
-
|
317 |
-
|
318 |
|
319 |
-
|
320 |
-
|
|
|
|
|
|
|
321 |
|
322 |
-
|
323 |
-
|
324 |
-
|
325 |
-
|
|
|
326 |
|
327 |
-
|
328 |
-
|
329 |
-
|
330 |
-
|
331 |
-
|
332 |
-
chatbot_message_type = provider_info['type']
|
333 |
-
max_total_tokens = provider_info['max_total_tokens']
|
334 |
|
335 |
-
|
336 |
-
|
|
|
337 |
|
338 |
-
|
339 |
-
logo_html_content = f'<img src="{logo_url}" width="100px" />'
|
340 |
-
logo_html_update = gr.update(value=logo_html_content)
|
341 |
|
342 |
-
|
343 |
-
note_markdown_update = gr.update(value=f"**Note:** This model is supported by {selected_provider}.")
|
344 |
|
345 |
-
#
|
346 |
-
|
347 |
-
|
348 |
-
placeholder=f"Enter your {selected_provider} API token to avoid rate limits"
|
349 |
-
)
|
350 |
|
351 |
-
|
352 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
353 |
|
354 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
355 |
|
356 |
-
|
357 |
-
|
358 |
-
|
359 |
-
|
360 |
-
|
361 |
-
|
362 |
-
)
|
363 |
|
364 |
-
|
365 |
-
|
366 |
-
|
367 |
-
|
368 |
-
|
369 |
-
|
370 |
-
preamble = "Paper not found or could not retrieve paper information."
|
371 |
-
if text is None:
|
372 |
-
return preamble, None, []
|
373 |
-
return preamble, text, []
|
374 |
-
elif paper_from_value == "paper_page":
|
375 |
-
# Fetch the paper information from Hugging Face API
|
376 |
-
url = f"https://huggingface.co/api/papers/{paper_id_value}?field=comments"
|
377 |
-
response = requests.get(url)
|
378 |
-
if response.status_code != 200:
|
379 |
-
return "Paper not found or could not retrieve paper information.", None, []
|
380 |
-
paper_info = response.json()
|
381 |
-
|
382 |
-
# Extract required information
|
383 |
-
title = paper_info.get('title', 'No Title')
|
384 |
-
link = f"https://huggingface.co/papers/{paper_id_value}"
|
385 |
-
authors_list = [author.get('name', 'Unknown') for author in paper_info.get('authors', [])]
|
386 |
-
authors = ', '.join(authors_list)
|
387 |
-
summary = paper_info.get('summary', 'No Summary')
|
388 |
-
num_comments = len(paper_info.get('comments', []))
|
389 |
-
num_upvotes = paper_info.get('upvotes', 0)
|
390 |
-
|
391 |
-
# Format the preamble
|
392 |
-
preamble = f"🤗 [paper-page]({link})<br/>"
|
393 |
-
preamble += f"**Title:** {title}<br/>"
|
394 |
-
preamble += f"**Authors:** {authors}<br/>"
|
395 |
-
preamble += f"**Summary:**<br/>>\n{summary}<br/>"
|
396 |
-
preamble += f"👍{num_comments} 💬{num_upvotes} <br/>"
|
397 |
-
|
398 |
-
# Fetch the paper content
|
399 |
-
text = fetch_paper_content_arxiv(paper_id_value)
|
400 |
-
if text is None:
|
401 |
-
text = "Paper content could not be retrieved."
|
402 |
-
return preamble, text, []
|
403 |
-
else:
|
404 |
-
return "", "", []
|
405 |
|
406 |
-
|
407 |
-
|
408 |
-
|
409 |
-
|
410 |
-
|
411 |
-
|
412 |
|
413 |
|
414 |
def main():
|
@@ -416,10 +453,7 @@ def main():
|
|
416 |
Launches the Gradio app.
|
417 |
"""
|
418 |
with gr.Blocks(css_paths="style.css") as demo:
|
419 |
-
# Create an input for paper_id
|
420 |
paper_id = gr.Textbox(label="Paper ID", value="")
|
421 |
-
|
422 |
-
# Create an input for paper_from (e.g., 'neurips' or 'paper_page')
|
423 |
paper_from = gr.Radio(
|
424 |
label="Paper Source",
|
425 |
choices=["neurips", "paper_page"],
|
@@ -427,11 +461,24 @@ def main():
|
|
427 |
)
|
428 |
|
429 |
# Build the paper chat tab
|
430 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
431 |
|
432 |
demo.launch(ssr_mode=False)
|
433 |
|
434 |
|
435 |
-
# Run the main function when the script is executed
|
436 |
if __name__ == "__main__":
|
437 |
main()
|
|
|
7 |
from io import BytesIO
|
8 |
from transformers import AutoTokenizer
|
9 |
import json
|
10 |
+
from datetime import datetime
|
11 |
import os
|
12 |
from openai import OpenAI
|
13 |
+
import re
|
14 |
|
15 |
# Cache for tokenizers to avoid reloading
|
16 |
tokenizer_cache = {}
|
|
|
24 |
"api_key_env_var": "SAMBANOVA_API_KEY",
|
25 |
"models": [
|
26 |
"Meta-Llama-3.1-70B-Instruct",
|
|
|
27 |
],
|
28 |
"type": "tuples",
|
29 |
"max_total_tokens": "50000",
|
|
|
43 |
}
|
44 |
|
45 |
|
46 |
+
# Functions for paper fetching
|
47 |
def fetch_paper_info_neurips(paper_id):
|
48 |
url = f"https://openreview.net/forum?id={paper_id}"
|
49 |
response = requests.get(url)
|
50 |
if response.status_code != 200:
|
51 |
+
return None, None, None
|
52 |
|
53 |
html_content = response.content
|
54 |
soup = BeautifulSoup(html_content, 'html.parser')
|
|
|
73 |
else:
|
74 |
abstract = 'Abstract not found'
|
75 |
|
76 |
+
# Construct preamble
|
77 |
+
link = f"https://openreview.net/forum?id={paper_id}"
|
78 |
+
return title, author_list, f"**Abstract:** {abstract}\n\n[View on OpenReview]({link})"
|
79 |
|
|
|
80 |
|
81 |
+
def fetch_paper_content_neurips(paper_id):
|
|
|
82 |
try:
|
83 |
+
url = f"https://openreview.net/pdf?id={paper_id}"
|
|
|
|
|
|
|
84 |
response = requests.get(url)
|
85 |
+
response.raise_for_status()
|
|
|
|
|
86 |
pdf_content = BytesIO(response.content)
|
87 |
reader = PdfReader(pdf_content)
|
|
|
|
|
88 |
text = ""
|
89 |
for page in reader.pages:
|
90 |
text += page.extract_text()
|
91 |
+
return text
|
92 |
+
except:
|
|
|
|
|
93 |
return None
|
94 |
|
95 |
|
96 |
+
def fetch_paper_content_arxiv(paper_id):
|
97 |
try:
|
98 |
+
url = f"https://arxiv.org/pdf/{paper_id}.pdf"
|
|
|
|
|
|
|
99 |
response = requests.get(url)
|
100 |
+
response.raise_for_status()
|
|
|
|
|
101 |
pdf_content = BytesIO(response.content)
|
102 |
reader = PdfReader(pdf_content)
|
|
|
|
|
103 |
text = ""
|
104 |
for page in reader.pages:
|
105 |
text += page.extract_text()
|
106 |
+
return text
|
|
|
|
|
107 |
except Exception as e:
|
108 |
+
print(f"Error fetching paper content: {e}")
|
109 |
return None
|
110 |
|
111 |
|
112 |
+
def fetch_paper_info_paperpage(paper_id_value):
|
113 |
+
# Extract paper_id from paper_page link or input
|
114 |
+
def extract_paper_id(input_string):
|
115 |
+
# Already in correct form?
|
116 |
+
if re.fullmatch(r'\d+\.\d+', input_string.strip()):
|
117 |
+
return input_string.strip()
|
118 |
+
# If URL
|
119 |
+
match = re.search(r'https://huggingface\.co/papers/(\d+\.\d+)', input_string)
|
120 |
+
if match:
|
121 |
+
return match.group(1)
|
122 |
+
return input_string.strip()
|
123 |
+
|
124 |
+
paper_id_value = extract_paper_id(paper_id_value)
|
125 |
+
url = f"https://huggingface.co/api/papers/{paper_id_value}?field=comments"
|
126 |
+
response = requests.get(url)
|
127 |
+
if response.status_code != 200:
|
128 |
+
return None, None, None
|
129 |
+
paper_info = response.json()
|
130 |
+
title = paper_info.get('title', 'No Title')
|
131 |
+
authors_list = [author.get('name', 'Unknown') for author in paper_info.get('authors', [])]
|
132 |
+
authors = ', '.join(authors_list)
|
133 |
+
summary = paper_info.get('summary', 'No Summary')
|
134 |
+
num_comments = len(paper_info.get('comments', []))
|
135 |
+
num_upvotes = paper_info.get('upvotes', 0)
|
136 |
+
link = f"https://huggingface.co/papers/{paper_id_value}"
|
137 |
+
|
138 |
+
details = f"{summary}<br/>👍{num_comments} 💬{num_upvotes}<br/> <a href='{link}' " \
|
139 |
+
f"target='_blank'>View on 🤗 hugging face</a>"
|
140 |
+
return title, authors, details
|
141 |
+
|
142 |
+
|
143 |
+
def fetch_paper_content_paperpage(paper_id_value):
|
144 |
+
# Extract paper_id
|
145 |
+
def extract_paper_id(input_string):
|
146 |
+
if re.fullmatch(r'\d+\.\d+', input_string.strip()):
|
147 |
+
return input_string.strip()
|
148 |
+
match = re.search(r'https://huggingface\.co/papers/(\d+\.\d+)', input_string)
|
149 |
+
if match:
|
150 |
+
return match.group(1)
|
151 |
+
return input_string.strip()
|
152 |
+
|
153 |
+
paper_id_value = extract_paper_id(paper_id_value)
|
154 |
+
text = fetch_paper_content_arxiv(paper_id_value)
|
155 |
+
return text
|
156 |
+
|
157 |
+
|
158 |
+
# Dictionary for paper sources
|
159 |
+
PAPER_SOURCES = {
|
160 |
+
"neurips": {
|
161 |
+
"fetch_info": fetch_paper_info_neurips,
|
162 |
+
"fetch_pdf": fetch_paper_content_neurips
|
163 |
+
},
|
164 |
+
"paper_page": {
|
165 |
+
"fetch_info": fetch_paper_info_paperpage,
|
166 |
+
"fetch_pdf": fetch_paper_content_paperpage
|
167 |
+
}
|
168 |
+
}
|
169 |
+
|
170 |
+
|
171 |
def create_chat_interface(provider_dropdown, model_dropdown, paper_content, hf_token_input, default_type,
|
172 |
provider_max_total_tokens):
|
173 |
# Define the function to handle the chat
|
|
|
|
|
174 |
def get_fn(message, history, paper_content_value, hf_token_value, provider_name_value, model_name_value,
|
175 |
max_total_tokens):
|
176 |
provider_info = PROVIDERS[provider_name_value]
|
|
|
179 |
models = provider_info['models']
|
180 |
max_total_tokens = int(max_total_tokens)
|
181 |
|
182 |
+
# Load tokenizer
|
183 |
tokenizer_key = f"{provider_name_value}_{model_name_value}"
|
184 |
if tokenizer_key not in tokenizer_cache:
|
|
|
|
|
185 |
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.2-1B-Instruct",
|
186 |
token=os.environ.get("HF_TOKEN"))
|
187 |
tokenizer_cache[tokenizer_key] = tokenizer
|
|
|
225 |
|
226 |
# Check if total tokens exceed the maximum allowed tokens
|
227 |
if total_tokens > max_total_tokens:
|
228 |
+
# Attempt to truncate context
|
229 |
available_tokens = max_total_tokens - (total_tokens - context_token_length)
|
230 |
if available_tokens > 0:
|
|
|
231 |
truncated_context_tokens = context_tokens[:available_tokens]
|
232 |
context = tokenizer.decode(truncated_context_tokens)
|
233 |
context_token_length = available_tokens
|
234 |
total_tokens = total_tokens - len(context_tokens) + context_token_length
|
235 |
else:
|
|
|
236 |
context = ""
|
237 |
total_tokens -= context_token_length
|
238 |
context_token_length = 0
|
239 |
|
240 |
+
# Truncate message history if needed
|
241 |
while total_tokens > max_total_tokens and len(messages) > 1:
|
|
|
242 |
removed_message = messages.pop(0)
|
243 |
removed_tokens = message_tokens_list.pop(0)
|
244 |
total_tokens -= removed_tokens
|
245 |
|
246 |
+
# Rebuild the final messages
|
247 |
final_messages = []
|
248 |
if context:
|
249 |
+
final_messages.append({"role": "system", "content": f"{context}"})
|
|
|
250 |
final_messages.extend(messages)
|
251 |
|
252 |
# Use the provider's API key
|
|
|
254 |
if not api_key:
|
255 |
raise ValueError("API token is not provided.")
|
256 |
|
257 |
+
# Initialize the OpenAI client
|
258 |
client = OpenAI(
|
259 |
base_url=endpoint,
|
260 |
api_key=api_key,
|
261 |
)
|
262 |
|
263 |
try:
|
|
|
264 |
completion = client.chat.completions.create(
|
265 |
model=model_name_value,
|
266 |
messages=final_messages,
|
|
|
272 |
response_text += delta
|
273 |
yield response_text
|
274 |
except json.JSONDecodeError as e:
|
275 |
+
yield f"JSON decoding error: {e.msg}"
|
|
|
|
|
|
|
|
|
276 |
except openai.OpenAIError as openai_err:
|
277 |
+
yield f"OpenAI error: {openai_err}"
|
|
|
|
|
278 |
except Exception as ex:
|
279 |
+
yield f"Unexpected error: {ex}"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
280 |
|
281 |
+
chatbot = gr.Chatbot(label="Chatbot", scale=1, height=400, autoscroll=True)
|
282 |
chat_interface = gr.ChatInterface(
|
283 |
fn=get_fn,
|
284 |
chatbot=chatbot,
|
|
|
288 |
return chat_interface, chatbot
|
289 |
|
290 |
|
291 |
+
def paper_chat_tab(paper_id, paper_from, paper_central_df):
|
292 |
+
with gr.Row():
|
293 |
+
# Left column: Paper selection and display
|
294 |
+
with gr.Column(scale=1):
|
295 |
+
gr.Markdown("### Select a Paper")
|
296 |
+
todays_date = datetime.today().strftime('%Y-%m-%d')
|
297 |
|
298 |
+
# Filter papers for today's date and having a paper_page
|
299 |
+
selectable_papers = paper_central_df.df_prettified
|
300 |
+
selectable_papers = selectable_papers[
|
301 |
+
selectable_papers['paper_page'].notna() &
|
302 |
+
(selectable_papers['paper_page'] != "") &
|
303 |
+
(selectable_papers['date'] == todays_date)
|
304 |
+
]
|
305 |
|
306 |
+
paper_choices = [(row['title'], row['paper_page']) for _, row in selectable_papers.iterrows()]
|
307 |
+
paper_choices = sorted(paper_choices, key=lambda x: x[0])
|
308 |
|
309 |
+
if not paper_choices:
|
310 |
+
paper_choices = [("No available papers for today", "")]
|
|
|
|
|
|
|
311 |
|
312 |
+
paper_select = gr.Dropdown(
|
313 |
+
label="Select a paper to chat with:",
|
314 |
+
choices=[p[0] for p in paper_choices],
|
315 |
+
value=paper_choices[0][0] if paper_choices else None
|
316 |
+
)
|
317 |
+
select_paper_button = gr.Button("Load this paper")
|
|
|
|
|
|
|
|
|
|
|
|
|
318 |
|
319 |
+
# Paper info display - styled card
|
320 |
+
content = gr.HTML(value="", elem_id="paper_info_card")
|
|
|
|
|
321 |
|
322 |
+
# Right column: Provider and model selection + chat
|
323 |
+
with gr.Column(scale=1, visible=False) as provider_section:
|
324 |
+
gr.Markdown("### LLM Provider and Model")
|
325 |
+
provider_names = list(PROVIDERS.keys())
|
326 |
+
default_provider = provider_names[0]
|
327 |
|
328 |
+
default_type = gr.State(value=PROVIDERS[default_provider]["type"])
|
329 |
+
default_max_total_tokens = gr.State(value=PROVIDERS[default_provider]["max_total_tokens"])
|
330 |
|
331 |
+
provider_dropdown = gr.Dropdown(
|
332 |
+
label="Select Provider",
|
333 |
+
choices=provider_names,
|
334 |
+
value=default_provider
|
335 |
+
)
|
336 |
|
337 |
+
hf_token_input = gr.Textbox(
|
338 |
+
label=f"Enter your {default_provider} API token (optional)",
|
339 |
+
type="password",
|
340 |
+
placeholder=f"Enter your {default_provider} API token to avoid rate limits"
|
341 |
+
)
|
342 |
|
343 |
+
model_dropdown = gr.Dropdown(
|
344 |
+
label="Select Model",
|
345 |
+
choices=PROVIDERS[default_provider]['models'],
|
346 |
+
value=PROVIDERS[default_provider]['models'][0]
|
347 |
+
)
|
|
|
|
|
348 |
|
349 |
+
logo_html = gr.HTML(
|
350 |
+
value=f'<img src="{PROVIDERS[default_provider]["logo"]}" width="100px" />'
|
351 |
+
)
|
352 |
|
353 |
+
note_markdown = gr.Markdown(f"**Note:** This model is supported by {default_provider}.")
|
|
|
|
|
354 |
|
355 |
+
paper_content = gr.State()
|
|
|
356 |
|
357 |
+
# Create chat interface
|
358 |
+
chat_interface, chatbot = create_chat_interface(provider_dropdown, model_dropdown, paper_content,
|
359 |
+
hf_token_input, default_type, default_max_total_tokens)
|
|
|
|
|
360 |
|
361 |
+
def update_provider(selected_provider):
|
362 |
+
provider_info = PROVIDERS[selected_provider]
|
363 |
+
models = provider_info['models']
|
364 |
+
logo_url = provider_info['logo']
|
365 |
+
chatbot_message_type = provider_info['type']
|
366 |
+
max_total_tokens = provider_info['max_total_tokens']
|
367 |
+
|
368 |
+
model_dropdown_choices = gr.update(choices=models, value=models[0])
|
369 |
+
logo_html_content = f'<img src="{logo_url}" width="100px" />'
|
370 |
+
logo_html_update = gr.update(value=logo_html_content)
|
371 |
+
note_markdown_update = gr.update(value=f"**Note:** This model is supported by {selected_provider}.")
|
372 |
+
hf_token_input_update = gr.update(
|
373 |
+
label=f"Enter your {selected_provider} API token (optional)",
|
374 |
+
placeholder=f"Enter your {selected_provider} API token to avoid rate limits"
|
375 |
+
)
|
376 |
+
chatbot_reset = []
|
377 |
+
return model_dropdown_choices, logo_html_update, note_markdown_update, hf_token_input_update, chatbot_message_type, max_total_tokens, chatbot_reset
|
378 |
+
|
379 |
+
provider_dropdown.change(
|
380 |
+
fn=update_provider,
|
381 |
+
inputs=provider_dropdown,
|
382 |
+
outputs=[model_dropdown, logo_html, note_markdown, hf_token_input, default_type, default_max_total_tokens,
|
383 |
+
chatbot],
|
384 |
+
queue=False
|
385 |
+
)
|
386 |
|
387 |
+
def update_paper_info(paper_id_value, paper_from_value, selected_model, old_content):
|
388 |
+
# Use PAPER_SOURCES to fetch info
|
389 |
+
source_info = PAPER_SOURCES.get(paper_from_value, {})
|
390 |
+
fetch_info_fn = source_info.get("fetch_info")
|
391 |
+
fetch_pdf_fn = source_info.get("fetch_pdf")
|
392 |
+
|
393 |
+
if not fetch_info_fn or not fetch_pdf_fn:
|
394 |
+
return gr.update(value="<div>No information available.</div>"), None, []
|
395 |
+
|
396 |
+
title, authors, details = fetch_info_fn(paper_id_value)
|
397 |
+
if title is None and authors is None and details is None:
|
398 |
+
return gr.update(value="<div>No information could be retrieved.</div>"), None, []
|
399 |
+
|
400 |
+
text = fetch_pdf_fn(paper_id_value)
|
401 |
+
if text is None:
|
402 |
+
text = "Paper content could not be retrieved."
|
403 |
+
|
404 |
+
# Create a styled card for the paper info
|
405 |
+
card_html = f"""
|
406 |
+
<div style="border:1px solid #ccc; border-radius:6px; background:#f9f9f9; padding:15px; margin-bottom:10px;">
|
407 |
+
<center><h3 style="margin-top:0; text-decoration:underline;">You are talking with:</h3></center>
|
408 |
+
<h3>{title}</h3>
|
409 |
+
<p><strong>Authors:</strong> {authors}</p>
|
410 |
+
<p>{details}</p>
|
411 |
+
</div>
|
412 |
+
"""
|
413 |
+
|
414 |
+
return gr.update(value=card_html), text, []
|
415 |
+
|
416 |
+
def select_paper(paper_title):
|
417 |
+
# Find the corresponding paper_page from the title
|
418 |
+
for t, ppage in paper_choices:
|
419 |
+
if t == paper_title:
|
420 |
+
return ppage, "paper_page"
|
421 |
+
return "", ""
|
422 |
+
|
423 |
+
select_paper_button.click(
|
424 |
+
fn=select_paper,
|
425 |
+
inputs=[paper_select],
|
426 |
+
outputs=[paper_id, paper_from]
|
427 |
+
)
|
428 |
|
429 |
+
# After updating paper_id, we update paper info
|
430 |
+
paper_id.change(
|
431 |
+
fn=update_paper_info,
|
432 |
+
inputs=[paper_id, paper_from, model_dropdown, content],
|
433 |
+
outputs=[content, paper_content, chatbot]
|
434 |
+
)
|
|
|
435 |
|
436 |
+
# Function to toggle visibility of the right column based on paper_id
|
437 |
+
def toggle_provider_visibility(paper_id_value):
|
438 |
+
if paper_id_value and paper_id_value.strip():
|
439 |
+
return gr.update(visible=True)
|
440 |
+
else:
|
441 |
+
return gr.update(visible=False)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
442 |
|
443 |
+
# Chain a then call to toggle visibility of the provider_section after paper info update
|
444 |
+
paper_id.change(
|
445 |
+
fn=toggle_provider_visibility,
|
446 |
+
inputs=[paper_id],
|
447 |
+
outputs=[provider_section]
|
448 |
+
)
|
449 |
|
450 |
|
451 |
def main():
|
|
|
453 |
Launches the Gradio app.
|
454 |
"""
|
455 |
with gr.Blocks(css_paths="style.css") as demo:
|
|
|
456 |
paper_id = gr.Textbox(label="Paper ID", value="")
|
|
|
|
|
457 |
paper_from = gr.Radio(
|
458 |
label="Paper Source",
|
459 |
choices=["neurips", "paper_page"],
|
|
|
461 |
)
|
462 |
|
463 |
# Build the paper chat tab
|
464 |
+
dummy_calendar = gr.State(datetime.now().strftime("%Y-%m-%d"))
|
465 |
+
|
466 |
+
class MockPaperCentral:
|
467 |
+
def __init__(self):
|
468 |
+
import pandas as pd
|
469 |
+
data = {
|
470 |
+
'date': [datetime.today().strftime('%Y-%m-%d')],
|
471 |
+
'paper_page': ['1234.56789'],
|
472 |
+
'title': ['An Example Paper']
|
473 |
+
}
|
474 |
+
self.df_prettified = pd.DataFrame(data)
|
475 |
+
|
476 |
+
paper_central_df = MockPaperCentral()
|
477 |
+
|
478 |
+
paper_chat_tab(paper_id, paper_from, paper_central_df)
|
479 |
|
480 |
demo.launch(ssr_mode=False)
|
481 |
|
482 |
|
|
|
483 |
if __name__ == "__main__":
|
484 |
main()
|