Skip to content

fix: update LLMHtmlExtractCompareV3 response type #389

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

Merged
e06084 merged 2 commits into from
Apr 10, 2026
Merged

fix: update LLMHtmlExtractCompareV3 response type #389

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ecde4ec
fix: update LLMHtmlExtractCompareV3 response type
e06084 Apr 10, 2026
61236d9
update ut
e06084 Apr 10, 2026
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
2 changes: 1 addition & 1 deletion dingo/model/llm/compare/llm_html_extract_compare_v3.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -215,7 +215,7 @@ def process_response(cls, response: str) -> EvalDetail:
tmp_type = "EXTRACTION_EQUAL"

result.status = response_model.score != 1
result.label = [f"{tmp_type}.{response_model.name}"]
result.label = [f"{tmp_type}"]
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist bot Apr 10, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

Removing the specific error type (response_model.name) from the label reduces the granularity of the evaluation results. This makes it harder to aggregate or filter results by the specific cause of quality differences (e.g., Error_Formula vs Error_Table) without parsing the full JSON reason for every record.

If the goal was to clean up the label (e.g., avoiding the .None suffix when the score is 0), consider a conditional approach that preserves the detail when a specific error is identified. Additionally, the f-string f"{tmp_type}" is redundant as tmp_type is already a string.

Suggested change
result.label = [f"{tmp_type}"]
result.label = [f"{tmp_type}.{response_model.name}" if response_model.name != "None" else tmp_type]

result.reason = [json.dumps(response_json, ensure_ascii=False)]

return result
8 changes: 4 additions & 4 deletions test/scripts/model/llm/test_llm_html_extract_compare_v3.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,7 +75,7 @@ def test_score_1_prompt_better(self):
)
result = LLMHtmlExtractCompareV3.process_response(raw)
assert result.metric == "LLMHtmlExtractCompareV3"
assert result.label == ["PROMPT_BETTER.Error_Content_Coverage"]
assert result.label == ["PROMPT_BETTER"]
assert result.status is False
parsed = json.loads(result.reason[0])
assert parsed["score"] == 1
Expand All @@ -86,7 +86,7 @@ def test_score_2_content_better(self):
ensure_ascii=False,
)
result = LLMHtmlExtractCompareV3.process_response(raw)
assert result.label == ["CONTENT_BETTER.Error_Formula"]
assert result.label == ["CONTENT_BETTER"]
assert result.status is True

def test_score_0_extraction_equal(self):
Expand All @@ -95,7 +95,7 @@ def test_score_0_extraction_equal(self):
ensure_ascii=False,
)
result = LLMHtmlExtractCompareV3.process_response(raw)
assert result.label == ["EXTRACTION_EQUAL.None"]
assert result.label == ["EXTRACTION_EQUAL"]
assert result.status is True

def test_json_fenced_with_markdown(self):
Expand All @@ -111,7 +111,7 @@ def test_redacted_thinking_appended_to_reason(self):
'{"score": 2, "name": "Error_Table", "reason": "Brief."}'
)
result = LLMHtmlExtractCompareV3.process_response(body)
assert "CONTENT_BETTER.Error_Table" == result.label[0]
assert "CONTENT_BETTER" == result.label[0]
parsed = json.loads(result.reason[0])
assert "internal" in parsed["reason"]

Expand Down
Loading