Skip to content

Match document language in LLM image descriptions #1000

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Br1an67 wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Match document language in LLM image descriptions #1000

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 9 additions & 1 deletion marker/processors/llm/llm_image_description.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@ class LLMImageDescriptionProcessor(BaseLLMSimpleBlockProcessor):
1. Carefully examine the provided image.
2. Analyze any text that was extracted from within the image.
3. Output a faithful description of the image. Make sure there is enough specific detail to accurately reconstruct the image. If the image is a figure or contains numeric data, include the numeric data in the output.
4. Output the description in the same language as the input text. If the input text is in French, write the description in French. If the input text is in German, write the description in German, and so on. Only use English if the input text is in English or if there is no input text.
**Example:**
Input:
```text
Expand All @@ -49,8 +50,15 @@ def block_prompts(self, document: Document) -> List[PromptData]:
prompt_data = []
for block_data in self.inference_blocks(document):
block = block_data["block"]
page = block_data["page"]
raw_text = block.raw_text(document)
# Include surrounding page text for language context if image has little text
if len(raw_text.strip()) < 20:
page_text = page.raw_text(document)[:200]
if page_text.strip():
raw_text = raw_text + "\n\n[Surrounding document text for language context:]\n" + page_text
prompt = self.image_description_prompt.replace(
"{raw_text}", block.raw_text(document)
"{raw_text}", raw_text
)
image = self.extract_image(document, block)

Expand Down