Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
"""
|
2 |
-
DOLPHIN PDF Document AI -
|
3 |
-
Optimized for
|
4 |
-
Features: AI-generated alt text for accessibility using Gemma 3n
|
5 |
"""
|
6 |
|
7 |
import gradio as gr
|
@@ -10,18 +10,16 @@ import markdown
|
|
10 |
import cv2
|
11 |
import numpy as np
|
12 |
from PIL import Image
|
13 |
-
from transformers import AutoProcessor, VisionEncoderDecoderModel,
|
14 |
import torch
|
15 |
try:
|
16 |
from sentence_transformers import SentenceTransformer
|
17 |
import numpy as np
|
18 |
from sklearn.metrics.pairwise import cosine_similarity
|
19 |
-
import google.generativeai as genai
|
20 |
-
from google.generativeai import types
|
21 |
RAG_DEPENDENCIES_AVAILABLE = True
|
22 |
except ImportError as e:
|
23 |
print(f"RAG dependencies not available: {e}")
|
24 |
-
print("Please install: pip install sentence-transformers scikit-learn
|
25 |
RAG_DEPENDENCIES_AVAILABLE = False
|
26 |
SentenceTransformer = None
|
27 |
import os
|
@@ -43,7 +41,7 @@ except ImportError:
|
|
43 |
|
44 |
class DOLPHIN:
|
45 |
def __init__(self, model_id_or_path):
|
46 |
-
"""Initialize the Hugging Face model optimized for
|
47 |
self.processor = AutoProcessor.from_pretrained(model_id_or_path)
|
48 |
self.model = VisionEncoderDecoderModel.from_pretrained(
|
49 |
model_id_or_path,
|
@@ -93,7 +91,7 @@ class DOLPHIN:
|
|
93 |
decoder_input_ids=batch_prompt_ids,
|
94 |
decoder_attention_mask=batch_attention_mask,
|
95 |
min_length=1,
|
96 |
-
max_length=
|
97 |
pad_token_id=self.tokenizer.pad_token_id,
|
98 |
eos_token_id=self.tokenizer.eos_token_id,
|
99 |
use_cache=True,
|
@@ -117,6 +115,139 @@ class DOLPHIN:
|
|
117 |
return results
|
118 |
|
119 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
120 |
def convert_pdf_to_images_gradio(pdf_file):
|
121 |
"""Convert uploaded PDF file to list of PIL Images"""
|
122 |
try:
|
@@ -170,7 +301,7 @@ def process_pdf_document(pdf_file, model, progress=gr.Progress()):
|
|
170 |
padded_image,
|
171 |
dims,
|
172 |
model,
|
173 |
-
max_batch_size=
|
174 |
)
|
175 |
|
176 |
try:
|
@@ -199,8 +330,8 @@ def process_pdf_document(pdf_file, model, progress=gr.Progress()):
|
|
199 |
return error_msg, "error"
|
200 |
|
201 |
|
202 |
-
def process_elements_optimized(layout_results, padded_image, dims, model, max_batch_size=
|
203 |
-
"""Optimized element processing for
|
204 |
layout_results = parse_layout_string(layout_results)
|
205 |
|
206 |
text_elements = []
|
@@ -221,8 +352,8 @@ def process_elements_optimized(layout_results, padded_image, dims, model, max_ba
|
|
221 |
pil_crop = Image.fromarray(cv2.cvtColor(cropped, cv2.COLOR_BGR2RGB))
|
222 |
pil_crop = crop_margin(pil_crop)
|
223 |
|
224 |
-
# Generate alt text for accessibility
|
225 |
-
alt_text =
|
226 |
|
227 |
buffered = io.BytesIO()
|
228 |
pil_crop.save(buffered, format="PNG")
|
@@ -274,8 +405,8 @@ def process_elements_optimized(layout_results, padded_image, dims, model, max_ba
|
|
274 |
return recognition_results
|
275 |
|
276 |
|
277 |
-
def process_element_batch_optimized(elements, model, prompt, max_batch_size=
|
278 |
-
"""Process elements in
|
279 |
results = []
|
280 |
batch_size = min(len(elements), max_batch_size)
|
281 |
|
@@ -316,7 +447,7 @@ def generate_fallback_markdown(recognition_results):
|
|
316 |
return markdown_content
|
317 |
|
318 |
|
319 |
-
# Initialize
|
320 |
model_path = "./hf_model"
|
321 |
if not os.path.exists(model_path):
|
322 |
model_path = "ByteDance/DOLPHIN"
|
@@ -324,164 +455,30 @@ if not os.path.exists(model_path):
|
|
324 |
# Model paths and configuration
|
325 |
model_path = "./hf_model" if os.path.exists("./hf_model") else "ByteDance/DOLPHIN"
|
326 |
hf_token = os.getenv('HF_TOKEN')
|
|
|
327 |
|
328 |
-
#
|
329 |
-
|
|
|
|
|
330 |
|
331 |
-
|
|
|
|
|
|
|
|
|
|
|
332 |
if RAG_DEPENDENCIES_AVAILABLE:
|
333 |
try:
|
334 |
print("Loading embedding model for RAG...")
|
335 |
embedding_model = SentenceTransformer('all-MiniLM-L6-v2', device='cpu')
|
336 |
print("β
Embedding model loaded successfully (CPU)")
|
337 |
-
|
338 |
-
# Initialize Gemini API
|
339 |
-
gemini_api_key = os.getenv('GEMINI_API_KEY')
|
340 |
-
if gemini_api_key:
|
341 |
-
genai.configure(api_key=gemini_api_key)
|
342 |
-
gemini_client = True # Just mark as configured
|
343 |
-
print("β
Gemini API configured successfully")
|
344 |
-
else:
|
345 |
-
print("β GEMINI_API_KEY not found in environment")
|
346 |
-
gemini_client = None
|
347 |
except Exception as e:
|
348 |
-
print(f"β Error loading
|
349 |
-
import traceback
|
350 |
-
traceback.print_exc()
|
351 |
embedding_model = None
|
352 |
-
gemini_client = None
|
353 |
else:
|
354 |
print("β RAG dependencies not available")
|
355 |
embedding_model = None
|
356 |
-
gemini_client = None
|
357 |
-
|
358 |
-
# Model management functions
|
359 |
-
def load_dolphin_model():
|
360 |
-
"""Load DOLPHIN model for PDF processing"""
|
361 |
-
global dolphin_model, current_model
|
362 |
-
|
363 |
-
if current_model == "dolphin":
|
364 |
-
return dolphin_model
|
365 |
-
|
366 |
-
# No need to unload chatbot model (using API now)
|
367 |
-
|
368 |
-
try:
|
369 |
-
print("Loading DOLPHIN model...")
|
370 |
-
dolphin_model = DOLPHIN(model_path)
|
371 |
-
current_model = "dolphin"
|
372 |
-
print(f"β
DOLPHIN model loaded (Device: {dolphin_model.device})")
|
373 |
-
return dolphin_model
|
374 |
-
except Exception as e:
|
375 |
-
print(f"β Error loading DOLPHIN model: {e}")
|
376 |
-
return None
|
377 |
-
|
378 |
-
def unload_dolphin_model():
|
379 |
-
"""Unload DOLPHIN model to free memory"""
|
380 |
-
global dolphin_model, current_model
|
381 |
-
|
382 |
-
if dolphin_model is not None:
|
383 |
-
print("Unloading DOLPHIN model...")
|
384 |
-
del dolphin_model
|
385 |
-
dolphin_model = None
|
386 |
-
if current_model == "dolphin":
|
387 |
-
current_model = None
|
388 |
-
if torch.cuda.is_available():
|
389 |
-
torch.cuda.empty_cache()
|
390 |
-
print("β
DOLPHIN model unloaded")
|
391 |
-
|
392 |
-
def initialize_gemini_client():
|
393 |
-
"""Initialize Gemini API client"""
|
394 |
-
global gemini_client
|
395 |
-
|
396 |
-
if gemini_client is not None:
|
397 |
-
return gemini_client
|
398 |
-
|
399 |
-
try:
|
400 |
-
gemini_api_key = os.getenv('GEMINI_API_KEY')
|
401 |
-
if not gemini_api_key:
|
402 |
-
print("β GEMINI_API_KEY not found in environment")
|
403 |
-
return None
|
404 |
-
|
405 |
-
print("Initializing Gemini API client...")
|
406 |
-
gemini_client = genai.configure(api_key=gemini_api_key)
|
407 |
-
print("β
Gemini API client ready for gemma-3n-e4b-it")
|
408 |
-
return gemini_client
|
409 |
-
except Exception as e:
|
410 |
-
print(f"β Error initializing Gemini client: {e}")
|
411 |
-
import traceback
|
412 |
-
traceback.print_exc()
|
413 |
-
return None
|
414 |
-
|
415 |
-
|
416 |
-
def generate_alt_text_for_image(pil_image):
|
417 |
-
"""Generate alt text for an image using Gemma 3n model via Google AI API"""
|
418 |
-
try:
|
419 |
-
# Initialize Gemini client
|
420 |
-
client = initialize_gemini_client()
|
421 |
-
if client is None:
|
422 |
-
print("β Gemini client not initialized for alt text generation")
|
423 |
-
return "Image description unavailable"
|
424 |
-
|
425 |
-
# Debug: Check image format and properties
|
426 |
-
print(f"π Image format: {pil_image.format}, mode: {pil_image.mode}, size: {pil_image.size}")
|
427 |
-
|
428 |
-
# Ensure image is in RGB mode
|
429 |
-
if pil_image.mode != 'RGB':
|
430 |
-
print(f"Converting image from {pil_image.mode} to RGB")
|
431 |
-
pil_image = pil_image.convert('RGB')
|
432 |
-
|
433 |
-
# Convert PIL image to bytes
|
434 |
-
buffered = io.BytesIO()
|
435 |
-
pil_image.save(buffered, format="JPEG")
|
436 |
-
image_bytes = buffered.getvalue()
|
437 |
-
|
438 |
-
print(f"π Generating alt text for image with Gemma 3n...")
|
439 |
-
|
440 |
-
# Create a detailed prompt for alt text generation
|
441 |
-
prompt = """You are an accessibility expert creating alt text for images to help visually impaired users understand visual content. Analyze this image and provide a clear, concise description that captures the essential visual information.
|
442 |
-
|
443 |
-
Focus on:
|
444 |
-
- Main subject or content of the image
|
445 |
-
- Important details, text, or data shown
|
446 |
-
- Layout and structure if relevant (charts, diagrams, tables)
|
447 |
-
- Context that would help someone understand the image's purpose
|
448 |
-
|
449 |
-
Provide a descriptive alt text in 1-2 sentences that is informative but not overly verbose. Start directly with the description without saying "This image shows" or similar phrases."""
|
450 |
-
|
451 |
-
# Use the Google AI API client with proper format
|
452 |
-
response = genai.GenerativeModel('gemma-3n-e4b-it').generate_content([
|
453 |
-
types.Part.from_bytes(
|
454 |
-
data=image_bytes,
|
455 |
-
mime_type='image/jpeg',
|
456 |
-
),
|
457 |
-
prompt
|
458 |
-
])
|
459 |
-
|
460 |
-
print(f"π‘ API response received: {type(response)}")
|
461 |
-
|
462 |
-
if hasattr(response, 'text') and response.text:
|
463 |
-
alt_text = response.text.strip()
|
464 |
-
print(f"β
Alt text generated: {alt_text[:100]}...")
|
465 |
-
else:
|
466 |
-
print(f"β No text in response. Response: {response}")
|
467 |
-
return "Image description unavailable"
|
468 |
-
|
469 |
-
# Clean up the alt text
|
470 |
-
alt_text = alt_text.replace('\n', ' ').replace('\r', ' ')
|
471 |
-
# Remove common prefixes if they appear
|
472 |
-
prefixes_to_remove = ["This image shows", "The image shows", "This shows", "The figure shows"]
|
473 |
-
for prefix in prefixes_to_remove:
|
474 |
-
if alt_text.startswith(prefix):
|
475 |
-
alt_text = alt_text[len(prefix):].strip()
|
476 |
-
break
|
477 |
-
|
478 |
-
return alt_text if alt_text else "Image description unavailable"
|
479 |
-
|
480 |
-
except Exception as e:
|
481 |
-
print(f"β Error generating alt text: {e}")
|
482 |
-
import traceback
|
483 |
-
traceback.print_exc()
|
484 |
-
return "Image description unavailable"
|
485 |
|
486 |
|
487 |
# Global state for managing tabs
|
@@ -490,14 +487,9 @@ show_results_tab = False
|
|
490 |
document_chunks = []
|
491 |
document_embeddings = None
|
492 |
|
493 |
-
# Global model state
|
494 |
-
dolphin_model = None
|
495 |
-
gemini_client = None
|
496 |
-
current_model = None # Track which model is currently loaded
|
497 |
-
|
498 |
|
499 |
def chunk_document(text, chunk_size=1024, overlap=100):
|
500 |
-
"""Split document into overlapping chunks for RAG
|
501 |
words = text.split()
|
502 |
chunks = []
|
503 |
|
@@ -554,16 +546,9 @@ def process_uploaded_pdf(pdf_file, progress=gr.Progress()):
|
|
554 |
return "β No PDF uploaded", gr.Tabs(visible=False)
|
555 |
|
556 |
try:
|
557 |
-
# Load DOLPHIN model for PDF processing
|
558 |
-
progress(0.1, desc="Loading DOLPHIN model...")
|
559 |
-
dolphin = load_dolphin_model()
|
560 |
-
|
561 |
-
if dolphin is None:
|
562 |
-
return "β Failed to load DOLPHIN model", gr.Tabs(visible=False)
|
563 |
-
|
564 |
# Process PDF
|
565 |
-
progress(0.
|
566 |
-
combined_markdown, status = process_pdf_document(pdf_file,
|
567 |
|
568 |
if status == "processing_complete":
|
569 |
processed_markdown = combined_markdown
|
@@ -574,9 +559,6 @@ def process_uploaded_pdf(pdf_file, progress=gr.Progress()):
|
|
574 |
document_embeddings = create_embeddings(document_chunks)
|
575 |
print(f"Created {len(document_chunks)} chunks")
|
576 |
|
577 |
-
# Keep DOLPHIN model loaded for GPU usage
|
578 |
-
progress(0.95, desc="Preparing chatbot...")
|
579 |
-
|
580 |
show_results_tab = True
|
581 |
progress(1.0, desc="PDF processed successfully!")
|
582 |
return "β
PDF processed successfully! Chatbot is ready in the Chat tab.", gr.Tabs(visible=True)
|
@@ -604,15 +586,16 @@ def clear_all():
|
|
604 |
document_chunks = []
|
605 |
document_embeddings = None
|
606 |
|
607 |
-
#
|
608 |
-
|
|
|
609 |
|
610 |
return None, "", gr.Tabs(visible=False)
|
611 |
|
612 |
|
613 |
# Create Gradio interface
|
614 |
with gr.Blocks(
|
615 |
-
title="DOLPHIN PDF AI",
|
616 |
theme=gr.themes.Soft(),
|
617 |
css="""
|
618 |
@import url('https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700&display=swap');
|
@@ -662,16 +645,15 @@ with gr.Blocks(
|
|
662 |
# Home Tab
|
663 |
with gr.TabItem("π Home", id="home"):
|
664 |
embedding_status = "β
RAG ready" if embedding_model else "β RAG not loaded"
|
665 |
-
gemini_status = "β
Gemini API ready" if gemini_client else "β Gemini API not configured"
|
666 |
-
current_status = f"Currently loaded: {current_model or 'None'}"
|
667 |
gr.Markdown(
|
668 |
-
"# Scholar Express -
|
669 |
-
"### Upload a research paper to get a web-friendly version with AI-generated alt text for accessibility. Includes an AI chatbot powered by
|
670 |
f"**System:** {model_status}\n"
|
671 |
f"**RAG System:** {embedding_status}\n"
|
672 |
-
f"**
|
|
|
673 |
f"**Alt Text:** Gemma 3n generates descriptive alt text for images\n"
|
674 |
-
f"**
|
675 |
)
|
676 |
|
677 |
with gr.Column(elem_classes="upload-container"):
|
@@ -742,7 +724,7 @@ with gr.Blocks(
|
|
742 |
send_btn = gr.Button("Send", variant="primary", scale=1)
|
743 |
|
744 |
gr.Markdown(
|
745 |
-
"*Ask questions about your processed document. The AI uses RAG (Retrieval-Augmented Generation) with
|
746 |
elem_id="chat-notice"
|
747 |
)
|
748 |
|
@@ -771,7 +753,7 @@ with gr.Blocks(
|
|
771 |
outputs=[chat_tab]
|
772 |
)
|
773 |
|
774 |
-
# Chatbot functionality with
|
775 |
def chatbot_response(message, history):
|
776 |
if not message.strip():
|
777 |
return history
|
@@ -780,26 +762,20 @@ with gr.Blocks(
|
|
780 |
return history + [[message, "β Please process a PDF document first before asking questions."]]
|
781 |
|
782 |
try:
|
783 |
-
#
|
784 |
-
client = initialize_gemini_client()
|
785 |
-
|
786 |
-
if client is None:
|
787 |
-
return history + [[message, "β Failed to initialize Gemini client. Please check your GEMINI_API_KEY."]]
|
788 |
-
|
789 |
-
# Use RAG to get relevant chunks from markdown (balanced for performance vs quota)
|
790 |
if document_chunks and len(document_chunks) > 0:
|
791 |
relevant_chunks = retrieve_relevant_chunks(message, document_chunks, document_embeddings, top_k=3)
|
792 |
context = "\n\n".join(relevant_chunks)
|
793 |
-
# Smart truncation: aim for ~
|
794 |
-
if len(context) >
|
795 |
# Try to cut at sentence boundaries
|
796 |
-
sentences = context[:
|
797 |
-
context = '.'.join(sentences[:-1]) + '...' if len(sentences) > 1 else context[:
|
798 |
else:
|
799 |
# Fallback to truncated document if RAG fails
|
800 |
-
context = processed_markdown[:
|
801 |
|
802 |
-
# Create prompt for
|
803 |
prompt = f"""You are a helpful assistant that answers questions about documents. Use the provided context to answer questions accurately and concisely.
|
804 |
|
805 |
Context from the document:
|
@@ -809,26 +785,9 @@ Question: {message}
|
|
809 |
|
810 |
Please provide a clear and helpful answer based on the context provided."""
|
811 |
|
812 |
-
# Generate response using
|
813 |
-
|
814 |
-
|
815 |
-
|
816 |
-
for attempt in range(max_retries):
|
817 |
-
try:
|
818 |
-
response = genai.GenerativeModel('gemma-3n-e4b-it').generate_content(prompt)
|
819 |
-
response_text = response.text if hasattr(response, 'text') else str(response)
|
820 |
-
return history + [[message, response_text]]
|
821 |
-
except Exception as api_error:
|
822 |
-
if "429" in str(api_error) and attempt < max_retries - 1:
|
823 |
-
# Rate limit hit, wait and retry
|
824 |
-
time.sleep(3)
|
825 |
-
continue
|
826 |
-
else:
|
827 |
-
# Other error or final attempt failed
|
828 |
-
if "429" in str(api_error):
|
829 |
-
return history + [[message, "β API quota exceeded. Please wait a moment and try again, or check your Gemini API billing."]]
|
830 |
-
else:
|
831 |
-
raise api_error
|
832 |
|
833 |
except Exception as e:
|
834 |
error_msg = f"β Error generating response: {str(e)}"
|
@@ -863,7 +822,7 @@ if __name__ == "__main__":
|
|
863 |
server_port=7860,
|
864 |
share=False,
|
865 |
show_error=True,
|
866 |
-
max_threads=
|
867 |
inbrowser=False,
|
868 |
quiet=True
|
869 |
)
|
|
|
1 |
"""
|
2 |
+
DOLPHIN PDF Document AI - Local Gemma 3n Version
|
3 |
+
Optimized for powerful GPU deployment with local models
|
4 |
+
Features: AI-generated alt text for accessibility using local Gemma 3n
|
5 |
"""
|
6 |
|
7 |
import gradio as gr
|
|
|
10 |
import cv2
|
11 |
import numpy as np
|
12 |
from PIL import Image
|
13 |
+
from transformers import AutoProcessor, VisionEncoderDecoderModel, AutoModelForImageTextToText
|
14 |
import torch
|
15 |
try:
|
16 |
from sentence_transformers import SentenceTransformer
|
17 |
import numpy as np
|
18 |
from sklearn.metrics.pairwise import cosine_similarity
|
|
|
|
|
19 |
RAG_DEPENDENCIES_AVAILABLE = True
|
20 |
except ImportError as e:
|
21 |
print(f"RAG dependencies not available: {e}")
|
22 |
+
print("Please install: pip install sentence-transformers scikit-learn")
|
23 |
RAG_DEPENDENCIES_AVAILABLE = False
|
24 |
SentenceTransformer = None
|
25 |
import os
|
|
|
41 |
|
42 |
class DOLPHIN:
|
43 |
def __init__(self, model_id_or_path):
|
44 |
+
"""Initialize the Hugging Face model optimized for powerful GPU"""
|
45 |
self.processor = AutoProcessor.from_pretrained(model_id_or_path)
|
46 |
self.model = VisionEncoderDecoderModel.from_pretrained(
|
47 |
model_id_or_path,
|
|
|
91 |
decoder_input_ids=batch_prompt_ids,
|
92 |
decoder_attention_mask=batch_attention_mask,
|
93 |
min_length=1,
|
94 |
+
max_length=2048,
|
95 |
pad_token_id=self.tokenizer.pad_token_id,
|
96 |
eos_token_id=self.tokenizer.eos_token_id,
|
97 |
use_cache=True,
|
|
|
115 |
return results
|
116 |
|
117 |
|
118 |
+
class Gemma3nModel:
|
119 |
+
def __init__(self, model_id="google/gemma-3n-E4B-it"):
|
120 |
+
"""Initialize the Gemma 3n model for text generation and image description"""
|
121 |
+
self.model_id = model_id
|
122 |
+
self.processor = AutoProcessor.from_pretrained(model_id)
|
123 |
+
self.model = AutoModelForImageTextToText.from_pretrained(
|
124 |
+
model_id,
|
125 |
+
torch_dtype="auto",
|
126 |
+
device_map="auto"
|
127 |
+
)
|
128 |
+
self.model.eval()
|
129 |
+
print(f"β
Gemma 3n loaded (Device: {self.model.device}, DType: {self.model.dtype})")
|
130 |
+
|
131 |
+
def generate_alt_text(self, pil_image):
|
132 |
+
"""Generate alt text for an image using local Gemma 3n"""
|
133 |
+
try:
|
134 |
+
# Ensure image is in RGB mode
|
135 |
+
if pil_image.mode != 'RGB':
|
136 |
+
pil_image = pil_image.convert('RGB')
|
137 |
+
|
138 |
+
# Create a detailed prompt for alt text generation
|
139 |
+
prompt = """You are an accessibility expert creating alt text for images to help visually impaired users understand visual content. Analyze this image and provide a clear, concise description that captures the essential visual information.
|
140 |
+
|
141 |
+
Focus on:
|
142 |
+
- Main subject or content of the image
|
143 |
+
- Important details, text, or data shown
|
144 |
+
- Layout and structure if relevant (charts, diagrams, tables)
|
145 |
+
- Context that would help someone understand the image's purpose
|
146 |
+
|
147 |
+
Provide a descriptive alt text in 1-2 sentences that is informative but not overly verbose. Start directly with the description without saying "This image shows" or similar phrases."""
|
148 |
+
|
149 |
+
# Prepare the message format
|
150 |
+
message = {
|
151 |
+
"role": "user",
|
152 |
+
"content": [
|
153 |
+
{"type": "image", "image": pil_image},
|
154 |
+
{"type": "text", "text": prompt}
|
155 |
+
]
|
156 |
+
}
|
157 |
+
|
158 |
+
# Apply chat template and generate
|
159 |
+
input_ids = self.processor.apply_chat_template(
|
160 |
+
[message],
|
161 |
+
add_generation_prompt=True,
|
162 |
+
tokenize=True,
|
163 |
+
return_dict=True,
|
164 |
+
return_tensors="pt",
|
165 |
+
)
|
166 |
+
input_len = input_ids["input_ids"].shape[-1]
|
167 |
+
|
168 |
+
input_ids = input_ids.to(self.model.device, dtype=self.model.dtype)
|
169 |
+
outputs = self.model.generate(
|
170 |
+
**input_ids,
|
171 |
+
max_new_tokens=256,
|
172 |
+
disable_compile=True,
|
173 |
+
do_sample=False,
|
174 |
+
temperature=0.1
|
175 |
+
)
|
176 |
+
|
177 |
+
text = self.processor.batch_decode(
|
178 |
+
outputs[:, input_len:],
|
179 |
+
skip_special_tokens=True,
|
180 |
+
clean_up_tokenization_spaces=True
|
181 |
+
)
|
182 |
+
|
183 |
+
alt_text = text[0].strip()
|
184 |
+
|
185 |
+
# Clean up the alt text
|
186 |
+
alt_text = alt_text.replace('\n', ' ').replace('\r', ' ')
|
187 |
+
# Remove common prefixes if they appear
|
188 |
+
prefixes_to_remove = ["This image shows", "The image shows", "This shows", "The figure shows"]
|
189 |
+
for prefix in prefixes_to_remove:
|
190 |
+
if alt_text.startswith(prefix):
|
191 |
+
alt_text = alt_text[len(prefix):].strip()
|
192 |
+
break
|
193 |
+
|
194 |
+
return alt_text if alt_text else "Image description unavailable"
|
195 |
+
|
196 |
+
except Exception as e:
|
197 |
+
print(f"β Error generating alt text: {e}")
|
198 |
+
import traceback
|
199 |
+
traceback.print_exc()
|
200 |
+
return "Image description unavailable"
|
201 |
+
|
202 |
+
def chat(self, prompt, history=None):
|
203 |
+
"""Chat functionality using Gemma 3n for text-only conversations"""
|
204 |
+
try:
|
205 |
+
# Create message format
|
206 |
+
message = {
|
207 |
+
"role": "user",
|
208 |
+
"content": [
|
209 |
+
{"type": "text", "text": prompt}
|
210 |
+
]
|
211 |
+
}
|
212 |
+
|
213 |
+
# If history exists, include it
|
214 |
+
conversation = history if history else []
|
215 |
+
conversation.append(message)
|
216 |
+
|
217 |
+
# Apply chat template and generate
|
218 |
+
input_ids = self.processor.apply_chat_template(
|
219 |
+
conversation,
|
220 |
+
add_generation_prompt=True,
|
221 |
+
tokenize=True,
|
222 |
+
return_dict=True,
|
223 |
+
return_tensors="pt",
|
224 |
+
)
|
225 |
+
input_len = input_ids["input_ids"].shape[-1]
|
226 |
+
|
227 |
+
input_ids = input_ids.to(self.model.device, dtype=self.model.dtype)
|
228 |
+
outputs = self.model.generate(
|
229 |
+
**input_ids,
|
230 |
+
max_new_tokens=1024,
|
231 |
+
disable_compile=True,
|
232 |
+
do_sample=True,
|
233 |
+
temperature=0.7
|
234 |
+
)
|
235 |
+
|
236 |
+
text = self.processor.batch_decode(
|
237 |
+
outputs[:, input_len:],
|
238 |
+
skip_special_tokens=True,
|
239 |
+
clean_up_tokenization_spaces=True
|
240 |
+
)
|
241 |
+
|
242 |
+
return text[0].strip()
|
243 |
+
|
244 |
+
except Exception as e:
|
245 |
+
print(f"β Error in chat: {e}")
|
246 |
+
import traceback
|
247 |
+
traceback.print_exc()
|
248 |
+
return f"Error generating response: {str(e)}"
|
249 |
+
|
250 |
+
|
251 |
def convert_pdf_to_images_gradio(pdf_file):
|
252 |
"""Convert uploaded PDF file to list of PIL Images"""
|
253 |
try:
|
|
|
301 |
padded_image,
|
302 |
dims,
|
303 |
model,
|
304 |
+
max_batch_size=4
|
305 |
)
|
306 |
|
307 |
try:
|
|
|
330 |
return error_msg, "error"
|
331 |
|
332 |
|
333 |
+
def process_elements_optimized(layout_results, padded_image, dims, model, max_batch_size=4):
|
334 |
+
"""Optimized element processing for powerful GPU"""
|
335 |
layout_results = parse_layout_string(layout_results)
|
336 |
|
337 |
text_elements = []
|
|
|
352 |
pil_crop = Image.fromarray(cv2.cvtColor(cropped, cv2.COLOR_BGR2RGB))
|
353 |
pil_crop = crop_margin(pil_crop)
|
354 |
|
355 |
+
# Generate alt text for accessibility using local Gemma 3n
|
356 |
+
alt_text = gemma_model.generate_alt_text(pil_crop)
|
357 |
|
358 |
buffered = io.BytesIO()
|
359 |
pil_crop.save(buffered, format="PNG")
|
|
|
405 |
return recognition_results
|
406 |
|
407 |
|
408 |
+
def process_element_batch_optimized(elements, model, prompt, max_batch_size=4):
|
409 |
+
"""Process elements in batches for powerful GPU"""
|
410 |
results = []
|
411 |
batch_size = min(len(elements), max_batch_size)
|
412 |
|
|
|
447 |
return markdown_content
|
448 |
|
449 |
|
450 |
+
# Initialize models
|
451 |
model_path = "./hf_model"
|
452 |
if not os.path.exists(model_path):
|
453 |
model_path = "ByteDance/DOLPHIN"
|
|
|
455 |
# Model paths and configuration
|
456 |
model_path = "./hf_model" if os.path.exists("./hf_model") else "ByteDance/DOLPHIN"
|
457 |
hf_token = os.getenv('HF_TOKEN')
|
458 |
+
gemma_model_id = "google/gemma-3n-E4B-it"
|
459 |
|
460 |
+
# Initialize models
|
461 |
+
print("Loading DOLPHIN model...")
|
462 |
+
dolphin_model = DOLPHIN(model_path)
|
463 |
+
print(f"β
DOLPHIN model loaded (Device: {dolphin_model.device})")
|
464 |
|
465 |
+
print("Loading Gemma 3n model...")
|
466 |
+
gemma_model = Gemma3nModel(gemma_model_id)
|
467 |
+
|
468 |
+
model_status = "β
Both models loaded successfully"
|
469 |
+
|
470 |
+
# Initialize embedding model
|
471 |
if RAG_DEPENDENCIES_AVAILABLE:
|
472 |
try:
|
473 |
print("Loading embedding model for RAG...")
|
474 |
embedding_model = SentenceTransformer('all-MiniLM-L6-v2', device='cpu')
|
475 |
print("β
Embedding model loaded successfully (CPU)")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
476 |
except Exception as e:
|
477 |
+
print(f"β Error loading embedding model: {e}")
|
|
|
|
|
478 |
embedding_model = None
|
|
|
479 |
else:
|
480 |
print("β RAG dependencies not available")
|
481 |
embedding_model = None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
482 |
|
483 |
|
484 |
# Global state for managing tabs
|
|
|
487 |
document_chunks = []
|
488 |
document_embeddings = None
|
489 |
|
|
|
|
|
|
|
|
|
|
|
490 |
|
491 |
def chunk_document(text, chunk_size=1024, overlap=100):
|
492 |
+
"""Split document into overlapping chunks for RAG"""
|
493 |
words = text.split()
|
494 |
chunks = []
|
495 |
|
|
|
546 |
return "β No PDF uploaded", gr.Tabs(visible=False)
|
547 |
|
548 |
try:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
549 |
# Process PDF
|
550 |
+
progress(0.1, desc="Processing PDF...")
|
551 |
+
combined_markdown, status = process_pdf_document(pdf_file, dolphin_model, progress)
|
552 |
|
553 |
if status == "processing_complete":
|
554 |
processed_markdown = combined_markdown
|
|
|
559 |
document_embeddings = create_embeddings(document_chunks)
|
560 |
print(f"Created {len(document_chunks)} chunks")
|
561 |
|
|
|
|
|
|
|
562 |
show_results_tab = True
|
563 |
progress(1.0, desc="PDF processed successfully!")
|
564 |
return "β
PDF processed successfully! Chatbot is ready in the Chat tab.", gr.Tabs(visible=True)
|
|
|
586 |
document_chunks = []
|
587 |
document_embeddings = None
|
588 |
|
589 |
+
# Clear GPU cache
|
590 |
+
if torch.cuda.is_available():
|
591 |
+
torch.cuda.empty_cache()
|
592 |
|
593 |
return None, "", gr.Tabs(visible=False)
|
594 |
|
595 |
|
596 |
# Create Gradio interface
|
597 |
with gr.Blocks(
|
598 |
+
title="DOLPHIN PDF AI - Local Gemma 3n",
|
599 |
theme=gr.themes.Soft(),
|
600 |
css="""
|
601 |
@import url('https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700&display=swap');
|
|
|
645 |
# Home Tab
|
646 |
with gr.TabItem("π Home", id="home"):
|
647 |
embedding_status = "β
RAG ready" if embedding_model else "β RAG not loaded"
|
|
|
|
|
648 |
gr.Markdown(
|
649 |
+
"# Scholar Express - Local Gemma 3n Version\n"
|
650 |
+
"### Upload a research paper to get a web-friendly version with AI-generated alt text for accessibility. Includes an AI chatbot powered by local Gemma 3n.\n"
|
651 |
f"**System:** {model_status}\n"
|
652 |
f"**RAG System:** {embedding_status}\n"
|
653 |
+
f"**DOLPHIN:** Local model for PDF processing\n"
|
654 |
+
f"**Gemma 3n:** Local model for alt text generation and chat\n"
|
655 |
f"**Alt Text:** Gemma 3n generates descriptive alt text for images\n"
|
656 |
+
f"**GPU:** {'CUDA available' if torch.cuda.is_available() else 'CPU only'}"
|
657 |
)
|
658 |
|
659 |
with gr.Column(elem_classes="upload-container"):
|
|
|
724 |
send_btn = gr.Button("Send", variant="primary", scale=1)
|
725 |
|
726 |
gr.Markdown(
|
727 |
+
"*Ask questions about your processed document. The AI uses RAG (Retrieval-Augmented Generation) with local Gemma 3n to find relevant sections and provide accurate answers.*",
|
728 |
elem_id="chat-notice"
|
729 |
)
|
730 |
|
|
|
753 |
outputs=[chat_tab]
|
754 |
)
|
755 |
|
756 |
+
# Chatbot functionality with local Gemma 3n
|
757 |
def chatbot_response(message, history):
|
758 |
if not message.strip():
|
759 |
return history
|
|
|
762 |
return history + [[message, "β Please process a PDF document first before asking questions."]]
|
763 |
|
764 |
try:
|
765 |
+
# Use RAG to get relevant chunks from markdown
|
|
|
|
|
|
|
|
|
|
|
|
|
766 |
if document_chunks and len(document_chunks) > 0:
|
767 |
relevant_chunks = retrieve_relevant_chunks(message, document_chunks, document_embeddings, top_k=3)
|
768 |
context = "\n\n".join(relevant_chunks)
|
769 |
+
# Smart truncation: aim for ~6000 chars for local model
|
770 |
+
if len(context) > 6000:
|
771 |
# Try to cut at sentence boundaries
|
772 |
+
sentences = context[:6000].split('.')
|
773 |
+
context = '.'.join(sentences[:-1]) + '...' if len(sentences) > 1 else context[:6000] + '...'
|
774 |
else:
|
775 |
# Fallback to truncated document if RAG fails
|
776 |
+
context = processed_markdown[:6000] + "..." if len(processed_markdown) > 6000 else processed_markdown
|
777 |
|
778 |
+
# Create prompt for Gemma 3n
|
779 |
prompt = f"""You are a helpful assistant that answers questions about documents. Use the provided context to answer questions accurately and concisely.
|
780 |
|
781 |
Context from the document:
|
|
|
785 |
|
786 |
Please provide a clear and helpful answer based on the context provided."""
|
787 |
|
788 |
+
# Generate response using local Gemma 3n
|
789 |
+
response_text = gemma_model.chat(prompt)
|
790 |
+
return history + [[message, response_text]]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
791 |
|
792 |
except Exception as e:
|
793 |
error_msg = f"β Error generating response: {str(e)}"
|
|
|
822 |
server_port=7860,
|
823 |
share=False,
|
824 |
show_error=True,
|
825 |
+
max_threads=4,
|
826 |
inbrowser=False,
|
827 |
quiet=True
|
828 |
)
|