mdipierro
/
SmolDocling-256M-Demo
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Update app.py

This commit is contained in:
Ahmed Nassar 2025-03-17 13:20:29 +00:00 committed by system
parent 4ff2750a5c
commit 61303b8155
No known key found for this signature in database
GPG Key ID: 6A528E38E0733467
1 changed files with 14 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
app.py
View File

@ -99,7 +99,6 @@ def model_inference(
"pixel_values": inputs.pixel_values, "pixel_values": inputs.pixel_values,
"attention_mask": inputs.attention_mask, "attention_mask": inputs.attention_mask,
"num_return_sequences": 1, "num_return_sequences": 1,
"no_repeat_ngram_size": 10,
"max_new_tokens": 8192, "max_new_tokens": 8192,
} }
@ -111,24 +110,30 @@ def model_inference(
yield "..." yield "..."
buffer = "" buffer = ""
doctag_output = "" full_output = ""
for new_text in streamer: for new_text in streamer:
if new_text != "<end_of_utterance>": full_output += new_text
buffer += html.escape(new_text) buffer += html.escape(new_text)
doctag_output += new_text
yield buffer yield buffer
if any(tag in doctag_output for tag in ["<doctag>", "<otsl>", "<code>", "<formula>", "<chart>"]): cleaned_output = full_output.replace("<end_of_utterance>", "").strip()
# final_output = buffer
# cleaned_output = final_output[len(inputs.input_ids):] if len(final_output) > prompt_length else final_output if cleaned_output:
doctag_output = cleaned_output
yield cleaned_output
if any(tag in doctag_output for tag in ["<doctag>", "<otsl>", "<code>", "<chart>", "<formula>"]):
doc = DoclingDocument(name="Document") doc = DoclingDocument(name="Document")
if "<chart>" in doctag_output: if "<chart>" in doctag_output:
doctag_output = doctag_output.replace("<chart>", "<otsl>").replace("</chart>", "</otsl>") doctag_output = doctag_output.replace("<chart>", "<otsl>").replace("</chart>", "</otsl>")
doctag_output = re.sub(r'(<loc_500>)(?!.*<loc_500>)<[^>]+>', r'\1', doctag_output) doctag_output = re.sub(r'(<loc_500>)(?!.*<loc_500>)<[^>]+>', r'\1', doctag_output)
print(doctag_output)
doctags_doc = DocTagsDocument.from_doctags_and_image_pairs([doctag_output], images) doctags_doc = DocTagsDocument.from_doctags_and_image_pairs([doctag_output], images)
doc.load_from_doctags(doctags_doc) doc.load_from_doctags(doctags_doc)
print(doc)
yield f"**MD Output:**\n\n{doc.export_to_markdown()}" yield f"**MD Output:**\n\n{doc.export_to_markdown()}"
examples=[[{"text": "Convert this page to docling.", "files": ["example_images/2d0fbcc50e88065a040a537b717620e964fb4453314b71d83f3ed3425addcef6.png"]}], examples=[[{"text": "Convert this page to docling.", "files": ["example_images/2d0fbcc50e88065a040a537b717620e964fb4453314b71d83f3ed3425addcef6.png"]}],