Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -172,7 +172,7 @@ def build_leaderboard_tab(leaderboard_table_file, text_recog_file, Inaccessible_
|
|
172 |
# arena table
|
173 |
with gr.Tab("OCRBench", id=0):
|
174 |
arena_table_vals = get_arena_table(model_table_df)
|
175 |
-
md = "OCRBench is a comprehensive evaluation benchmark designed to assess the OCR capabilities of Large Multimodal Models. It
|
176 |
gr.Markdown(md, elem_id="leaderboard_markdown")
|
177 |
gr.Dataframe(
|
178 |
headers=[
|
@@ -207,7 +207,7 @@ def build_leaderboard_tab(leaderboard_table_file, text_recog_file, Inaccessible_
|
|
207 |
)
|
208 |
with gr.Tab("Text Recognition", id=1):
|
209 |
arena_table_vals = get_recog_table(recog_table_df)
|
210 |
-
md = "OCRBench is a comprehensive evaluation benchmark designed to assess the OCR capabilities of Large Multimodal Models. It
|
211 |
gr.Markdown(md, elem_id="leaderboard_markdown")
|
212 |
gr.Dataframe(
|
213 |
headers=[
|
@@ -281,7 +281,7 @@ def build_leaderboard_tab(leaderboard_table_file, text_recog_file, Inaccessible_
|
|
281 |
pass
|
282 |
md_tail = f"""
|
283 |
# Notice
|
284 |
-
|
285 |
If you would like to include your model in the OCRBench leaderboard, please follow the evaluation instructions provided on [GitHub](https://github.com/Yuliang-Liu/MultimodalOCR), [VLMEvalKit](https://github.com/open-compass/VLMEvalKit) or [lmms-eval](https://github.com/EvolvingLMMs-Lab/lmms-eval) and feel free to contact us via email at [email protected]. We will update the leaderboard in time."""
|
286 |
gr.Markdown(md_tail, elem_id="leaderboard_markdown")
|
287 |
|
|
|
172 |
# arena table
|
173 |
with gr.Tab("OCRBench", id=0):
|
174 |
arena_table_vals = get_arena_table(model_table_df)
|
175 |
+
md = "OCRBench is a comprehensive evaluation benchmark designed to assess the OCR capabilities of Large Multimodal Models. It consists of five components: Text Recognition, SceneText-Centric VQA, Document-Oriented VQA, Key Information Extraction, and Handwritten Mathematical Expression Recognition. The benchmark includes 1000 question-answer pairs, with all answers containing no fewer than 4 symbols to reduce false positives. All answers undergo manual verification and correction to ensure a more precise evaluation."
|
176 |
gr.Markdown(md, elem_id="leaderboard_markdown")
|
177 |
gr.Dataframe(
|
178 |
headers=[
|
|
|
207 |
)
|
208 |
with gr.Tab("Text Recognition", id=1):
|
209 |
arena_table_vals = get_recog_table(recog_table_df)
|
210 |
+
md = "OCRBench is a comprehensive evaluation benchmark designed to assess the OCR capabilities of Large Multimodal Models. It consists of five components: Text Recognition, SceneText-Centric VQA, Document-Oriented VQA, Key Information Extraction, and Handwritten Mathematical Expression Recognition. The benchmark includes 1000 question-answer pairs, with all answers containing no fewer than 4 symbols to reduce false positives. All answers undergo manual verification and correction to ensure a more precise evaluation."
|
211 |
gr.Markdown(md, elem_id="leaderboard_markdown")
|
212 |
gr.Dataframe(
|
213 |
headers=[
|
|
|
281 |
pass
|
282 |
md_tail = f"""
|
283 |
# Notice
|
284 |
+
Sometimes, API calls to closed-source models may not succeed. In such cases, we will repeat the calls for unsuccessful samples until it becomes impossible to obtain a successful response. It is important to note that due to rigorous security reviews by OpenAI, GPT4V refuses to provide results for the 84 samples in OCRBench.
|
285 |
If you would like to include your model in the OCRBench leaderboard, please follow the evaluation instructions provided on [GitHub](https://github.com/Yuliang-Liu/MultimodalOCR), [VLMEvalKit](https://github.com/open-compass/VLMEvalKit) or [lmms-eval](https://github.com/EvolvingLMMs-Lab/lmms-eval) and feel free to contact us via email at [email protected]. We will update the leaderboard in time."""
|
286 |
gr.Markdown(md_tail, elem_id="leaderboard_markdown")
|
287 |
|