Upload 2 files

app.py

@@ -6,7 +6,8 @@ import pandas as pd
 ###########################################
 
 llm_judge_filename = "llm_judge_results.jsonl"
-response_generation_filename = "report_generation_w_docs.jsonl"
+response_generation_filename = "report_generation.jsonl"
+response_generation_w_docs_filename = "report_generation_w_docs.jsonl"
 
 def load_filename_into_df(filename):
     df = pd.read_json(filename, lines=True)
@@ -106,10 +107,10 @@ with gr.Blocks(theme=theme) as app:
             gr.Markdown(CAPTION_V2)
 
     with gr.Tabs(elem_id="outer-tabs", elem_classes="tabs-big") as tabs_big:
-        with gr.TabItem("Report Generation w Docs"):
+        with gr.TabItem("Report Generation"):
             with gr.Row():
                 with gr.Column(scale=7):
-                    gr.Markdown("Report Generation Leaderboard with Grounding Documents")
+                    gr.Markdown("Report Generation Leaderboard")
                 
             with gr.Tabs(elem_id="inner-tabs", elem_classes="tabs-small") as tabs:
                 with gr.TabItem("Leaderboard"):
@@ -191,11 +192,58 @@ with gr.Blocks(theme=theme) as app:
                             elem_id="llm_judge_dataframe",
                             row_count=(25, "dynamic"),
                         )
+        
+        with gr.TabItem("Report Generation w Docs"):
+            with gr.Row():
+                with gr.Column(scale=7):
+                    gr.Markdown("Report Generation Leaderboard with Grounding Documents")
+                
+            with gr.Tabs(elem_id="inner-tabs", elem_classes="tabs-small") as tabs:
+                with gr.TabItem("Leaderboard"):
+                    with gr.Row():
+                        search_1_v2 = gr.Textbox(
+                            label="Model Search (delimit with , )",
+                            placeholder="Model Search (delimit with , )",
+                            show_label=False,
+                            scale=8,
+                        )
+                        model_types_1_v2 = gr.CheckboxGroup(
+                            ["Open-weight", "Closed-source", "Reasoning", "Instruct"],
+                            value=["Open-weight", "Closed-source", "Reasoning", "Instruct"],
+                            show_label=False,
+                            scale=8,
+                        )
+
+                    with gr.Row():
+                        col_types_response_generation = ["number"] + ["markdown"] + ["str"] + ["number"] * 12
+                        df_response_generation_w_docs = load_filename_into_df(response_generation_w_docs_filename)
+
+                        rewardbench_table_hidden_v2 = gr.Dataframe(
+                            df_response_generation_w_docs.values,
+                            datatype=col_types_response_generation,
+                            headers=df_response_generation_w_docs.columns.tolist(),
+                            visible=False,
+                        )
+
+                        rewardbench_table_v2 = gr.Dataframe(
+                            regex_table(
+                                df_response_generation_w_docs.copy(),
+                                "",
+                                ["Open-weight", "Closed-source", "Reasoning", "Instruct"]
+                            ),
+                            datatype=col_types_response_generation,
+                            headers=df_response_generation_w_docs.columns.tolist(),
+                            elem_id="response_generation_dataframe",
+                            row_count=(25, "dynamic"),
+                        )
             
     search_1.change(regex_table, inputs=[rewardbench_table_hidden, search_1, model_types_1], outputs=rewardbench_table)
     search_1_v1.change(
         regex_table, inputs=[rewardbench_table_hidden_v1, search_1_v1, model_types_1_v1], outputs=rewardbench_table_v1
     )
+    search_1_v2.change(
+        regex_table, inputs=[rewardbench_table_hidden_v2, search_1_v2, model_types_1_v2], outputs=rewardbench_table_v2
+    )
 
     model_types_1.change(
         regex_table, inputs=[rewardbench_table_hidden, search_1, model_types_1], outputs=rewardbench_table
@@ -204,6 +252,10 @@ with gr.Blocks(theme=theme) as app:
         regex_table, inputs=[rewardbench_table_hidden_v1, search_1_v1, model_types_1_v1], outputs=rewardbench_table_v1
     )
 
+    model_types_1_v2.change(
+        regex_table, inputs=[rewardbench_table_hidden_v2, search_1_v2, model_types_1_v2], outputs=rewardbench_table_v2
+    )
+
     with gr.Row():
         with gr.Accordion("📚 Citation and Credits", open=False):
             citation_button = gr.Textbox(

report_generation.jsonl

@@ -0,0 +1,28 @@
+{"Model": "OpenAI/GPT-5 (high)", "Category": "Closed-source Reasoning", "Overall": 49.4, "Physics": 45.7, "Chemistry": 62.6, "Finance": 29.9, "Consulting": 59.2, "Extraction": 37.1, "Reasoning": 51.7, "Style": 65.5, "Response Characters": 5876, "Input Tokens": 467, "Output Tokens": 16123, "Cost": 25.89}
+{"Model": "OpenAI/GPT-5-mini (high)", "Category": "Closed-source Reasoning", "Overall": 41.6, "Physics": 40.0, "Chemistry": 59.5, "Finance": 11.8, "Consulting": 55.1, "Extraction": 32.7, "Reasoning": 42.1, "Style": 65.9, "Response Characters": 7870, "Input Tokens": 956, "Output Tokens": 15280, "Cost": 4.93}
+{"Model": "OpenAI/GPT-5-nano (high)", "Category": "Closed-source Reasoning", "Overall": 36.9, "Physics": 29.1, "Chemistry": 37.9, "Finance": 23.6, "Consulting": 56.9, "Extraction": 26.5, "Reasoning": 35.6, "Style": 58.0, "Response Characters": 8915, "Input Tokens": 467, "Output Tokens": 23008, "Cost": 1.48}
+{"Model": "OpenAI/o3", "Category": "Closed-source Reasoning", "Overall": 52.4, "Physics": 38.6, "Chemistry": 57.2, "Finance": 44.1, "Consulting": 69.8, "Extraction": 43.0, "Reasoning": 54.1, "Style": 59.2, "Response Characters": 4226, "Input Tokens": 467, "Output Tokens": 5569, "Cost": 7.28}
+{"Model": "OpenAI/o4-mini", "Category": "Closed-source Reasoning", "Overall": 47.5, "Physics": 34.6, "Chemistry": 50.1, "Finance": 38.1, "Consulting": 67.2, "Extraction": 37.2, "Reasoning": 47.7, "Style": 60.4, "Response Characters": 3046, "Input Tokens": 467, "Output Tokens": 4335, "Cost": 0.77}
+{"Model": "Google/Gemini-2.5-Pro", "Category": "Closed-source Reasoning", "Overall": 52.1, "Physics": 40.4, "Chemistry": 63.8, "Finance": 36.7, "Consulting": 67.5, "Extraction": 45.9, "Reasoning": 53.1, "Style": 62.6, "Response Characters": 8492, "Input Tokens": 480, "Output Tokens": 9102, "Cost": 14.66}
+{"Model": "Google/Gemini-2.5-Flash (Thinking)", "Category": "Closed-source Reasoning", "Overall": 49.2, "Physics": 35.9, "Chemistry": 63.9, "Finance": 33.2, "Consulting": 63.8, "Extraction": 43.6, "Reasoning": 51.4, "Style": 57.3, "Response Characters": 18559, "Input Tokens": 480, "Output Tokens": 12943, "Cost": 5.2}
+{"Model": "Google/Gemini-2.5-Flash-Lite (Thinking)", "Category": "Closed-source Reasoning", "Overall": 44.3, "Physics": 32.3, "Chemistry": 52.7, "Finance": 31.3, "Consulting": 61.0, "Extraction": 35.8, "Reasoning": 43.3, "Style": 56.7, "Response Characters": 12153, "Input Tokens": 480, "Output Tokens": 17302, "Cost": 1.12}
+{"Model": "xAI/grok-4-0709", "Category": "Closed-source Reasoning", "Overall": 45.1, "Physics": 20.6, "Chemistry": 59.8, "Finance": 29.4, "Consulting": 70.5, "Extraction": 40.1, "Reasoning": 48.4, "Style": 65.2, "Response Characters": 4977, "Input Tokens": 1126, "Output Tokens": 17957, "Cost": 43.64}
+{"Model": "Anthropic/claude-sonnet-4 (Thinking)", "Category": "Closed-source Reasoning", "Overall": 42.5, "Physics": 39.5, "Chemistry": 53.3, "Finance": 21.2, "Consulting": 56.1, "Extraction": 29.5, "Reasoning": 42.5, "Style": 66.1, "Response Characters": 3621, "Input Tokens": 559, "Output Tokens": 7924, "Cost": 19.29}
+{"Model": "OpenAI/gpt-oss-120b", "Category": "Open-weight Reasoning", "Overall": 50.0, "Physics": 43.6, "Chemistry": 53.5, "Finance": 35.3, "Consulting": 67.6, "Extraction": 39.7, "Reasoning": 51.5, "Style": 63.4, "Response Characters": 8657, "Input Tokens": 530, "Output Tokens": 4817, "Cost": 0.31}
+{"Model": "OpenAI/gpt-oss-20b", "Category": "Open-weight Reasoning", "Overall": 42.3, "Physics": 33.6, "Chemistry": 40.5, "Finance": 28.7, "Consulting": 66.4, "Extraction": 29.9, "Reasoning": 44.1, "Style": 59.1, "Response Characters": 5609, "Input Tokens": 508, "Output Tokens": 5375, "Cost": 0.12}
+{"Model": "DeepSeek-AI/DeepSeek-V3.1 (Thinking)", "Category": "Open-weight Reasoning", "Overall": 45.9, "Physics": 32.3, "Chemistry": 53.9, "Finance": 35.6, "Consulting": 61.9, "Extraction": 39.4, "Reasoning": 50.0, "Style": 59.1, "Response Characters": 5760, "Input Tokens": 415, "Output Tokens": 6253, "Cost": 0.81}
+{"Model": "Qwen/Qwen3-235B-A22B-Thinking-2507", "Category": "Open-weight Reasoning", "Overall": 41.1, "Physics": 32.7, "Chemistry": 46.4, "Finance": 23.5, "Consulting": 61.9, "Extraction": 35.0, "Reasoning": 46.1, "Style": 63.4, "Response Characters": 11390, "Input Tokens": 490, "Output Tokens": 5568, "Cost": 0.54}
+{"Model": "Qwen/Qwen3-30B-A3B-Thinking-2507", "Category": "Open-weight Reasoning", "Overall": 37.6, "Physics": 19.0, "Chemistry": 44.7, "Finance": 25.6, "Consulting": 61.3, "Extraction": 29.4, "Reasoning": 42.4, "Style": 73.1, "Response Characters": 5892, "Input Tokens": 469, "Output Tokens": 6376, "Cost": 0.3}
+{"Model": "OpenAI/GPT-4.1", "Category": "Closed-source Instruct", "Overall": 46.3, "Physics": 34.3, "Chemistry": 48.5, "Finance": 36.4, "Consulting": 65.9, "Extraction": 36.5, "Reasoning": 47.4, "Style": 55.7, "Response Characters": 7386, "Input Tokens": 468, "Output Tokens": 2394, "Cost": 3.21}
+{"Model": "OpenAI/GPT-4.1-mini", "Category": "Closed-source Instruct", "Overall": 42.8, "Physics": 40.5, "Chemistry": 50.6, "Finance": 23.1, "Consulting": 57.0, "Extraction": 33.5, "Reasoning": 43.1, "Style": 58.7, "Response Characters": 7550, "Input Tokens": 468, "Output Tokens": 2322, "Cost": 0.62}
+{"Model": "OpenAI/GPT-4.1-nano", "Category": "Closed-source Instruct", "Overall": 33.2, "Physics": 21.4, "Chemistry": 35.1, "Finance": 23.3, "Consulting": 53.2, "Extraction": 24.9, "Reasoning": 32.9, "Style": 48.3, "Response Characters": 6198, "Input Tokens": 468, "Output Tokens": 1799, "Cost": 0.12}
+{"Model": "Google/Gemini-2.5-Flash", "Category": "Closed-source Instruct", "Overall": 47.8, "Physics": 37.0, "Chemistry": 57.6, "Finance": 36.7, "Consulting": 60.0, "Extraction": 41.8, "Reasoning": 50.3, "Style": 53.6, "Response Characters": 24479, "Input Tokens": 480, "Output Tokens": 6255, "Cost": 2.53}
+{"Model": "Google/Gemini-2.5-Flash-Lite", "Category": "Closed-source Instruct", "Overall": 41.8, "Physics": 28.8, "Chemistry": 51.2, "Finance": 25.0, "Consulting": 62.1, "Extraction": 35.2, "Reasoning": 41.3, "Style": 52.1, "Response Characters": 26746, "Input Tokens": 480, "Output Tokens": 8723, "Cost": 0.57}
+{"Model": "Anthropic/claude-sonnet-4", "Category": "Closed-source Instruct", "Overall": 41.2, "Physics": 34.8, "Chemistry": 47.1, "Finance": 18.9, "Consulting": 63.9, "Extraction": 32.6, "Reasoning": 40.9, "Style": 58.3, "Response Characters": 4047, "Input Tokens": 531, "Output Tokens": 1375, "Cost": 3.55}
+{"Model": "Anthropic/claude-3.5-haiku", "Category": "Closed-source Instruct", "Overall": 21.2, "Physics": 6.9, "Chemistry": 25.2, "Finance": 8.5, "Consulting": 44.3, "Extraction": 16.6, "Reasoning": 19.7, "Style": 41.7, "Response Characters": 1618, "Input Tokens": 531, "Output Tokens": 519, "Cost": 0.4}
+{"Model": "Qwen/Qwen3-235B-A22B-Instruct-2507", "Category": "Open-weight Instruct", "Overall": 45.0, "Physics": 31.9, "Chemistry": 52.4, "Finance": 32.9, "Consulting": 63.0, "Extraction": 38.1, "Reasoning": 48.4, "Style": 54.5, "Response Characters": 14314, "Input Tokens": 487, "Output Tokens": 5000, "Cost": 0.45}
+{"Model": "Qwen/Qwen3-30B-A3B-Instruct-2507", "Category": "Open-weight Instruct", "Overall": 40.0, "Physics": 26.6, "Chemistry": 47.2, "Finance": 24.9, "Consulting": 61.4, "Extraction": 31.2, "Reasoning": 40.1, "Style": 50.7, "Response Characters": 10488, "Input Tokens": 487, "Output Tokens": 3654, "Cost": 0.2}
+{"Model": "DeepSeek-AI/DeepSeek-V3.1", "Category": "Open-weight Instruct", "Overall": 46.6, "Physics": 37.1, "Chemistry": 51.7, "Finance": 34.4, "Consulting": 63.1, "Extraction": 40.8, "Reasoning": 47.2, "Style": 55.9, "Response Characters": 8740, "Input Tokens": 456, "Output Tokens": 2717, "Cost": 0.36}
+{"Model": "MoonshotAI/Kimi-K2-Instruct-0905", "Category": "Open-weight Instruct", "Overall": 44.2, "Physics": 33.7, "Chemistry": 47.3, "Finance": 32.2, "Consulting": 63.6, "Extraction": 36.6, "Reasoning": 43.2, "Style": 58.8, "Response Characters": 5322, "Input Tokens": 481, "Output Tokens": 1709, "Cost": 0.55}
+{"Model": "Meta/llama-4-maverick", "Category": "Open-weight Instruct", "Overall": 34.4, "Physics": 31.3, "Chemistry": 35.4, "Finance": 22.1, "Consulting": 48.8, "Extraction": 27.2, "Reasoning": 34.6, "Style": 32.6, "Response Characters": 4532, "Input Tokens": 479, "Output Tokens": 1292, "Cost": 0.14}
+{"Model": "Meta/llama-4-scout", "Category": "Open-weight Instruct", "Overall": 31.2, "Physics": 19.2, "Chemistry": 30.0, "Finance": 19.5, "Consulting": 55.9, "Extraction": 26.2, "Reasoning": 29.9, "Style": 38.9, "Response Characters": 4200, "Input Tokens": 457, "Output Tokens": 1197, "Cost": 0.06}