Update submit tab

.gitignore

@@ -0,0 +1,24 @@
+# Ignore compiled Python files
+*.pyc
+*.pyo
+__pycache__/
+
+# Ignore virtual environment files
+venv/
+env/
+.env/
+
+# Ignore package build files
+build/
+dist/
+*.egg-info/
+
+# Ignore IDE and editor files
+.vscode/
+.idea/
+*.sublime-project
+*.sublime-workspace
+
+# Ignore system-specific files
+.DS_Store
+Thumbs.db

app.py

@@ -1,8 +1,7 @@
-import gradio as gr
-import pandas as pd
-
 from utils import *
 
+global data_component
+
 block = gr.Blocks()
 
 with block:
@@ -10,7 +9,7 @@ with block:
         LEADERBORAD_INTRODUCTION
     )
     with gr.Tabs(elem_classes="tab-buttons") as tabs:
-        # Table 0
+        # Table 1
         with gr.TabItem("📊 TheoremQA", elem_id="theoremqa-tab-table1", id=1):
             with gr.Row():
                 with gr.Accordion("Citation", open=False):
@@ -24,67 +23,39 @@ with block:
                 TABLE_INTRODUCTION
             )
 
-            gr.components.Dataframe(
-                value=pd.read_csv(CSV_DIR), 
+            data_component=gr.components.Dataframe(
+                value=get_df(), 
                 headers=COLUMN_NAMES,
                 type="pandas", 
                 datatype=DATA_TITILE_TYPE,
                 interactive=False,
                 visible=True,
                 )
+            refresh_button = gr.Button("Refresh")
+            refresh_button.click(fn=refresh_data, outputs=data_component)
 
+        # table 2
         with gr.TabItem("📝 About", elem_id="theoremqa-tab-table2", id=2):
             gr.Markdown(LEADERBORAD_INFO, elem_classes="markdown-text")
         
         # table 3 
-        # with gr.TabItem("🚀 Submit here! ", elem_id="mtheoremqa-tab-table", id=3):
-        #     gr.Markdown(LEADERBORAD_INTRODUCTION, elem_classes="markdown-text")
-
-        #     with gr.Row():
-        #         gr.Markdown(SUBMIT_INTRODUCTION, elem_classes="markdown-text")
-
-        #     with gr.Row():
-        #         gr.Markdown("# ✉️✨ Submit your model evaluation json file here!", elem_classes="markdown-text")
-
-        #     with gr.Row():
-        #         with gr.Column():
-        #             model_name_textbox = gr.Textbox(
-        #                 label="Model name", placeholder="LaVie"
-        #                 )
-        #             revision_name_textbox = gr.Textbox(
-        #                 label="Revision Model Name", placeholder="LaVie"
-        #             )
-
-        #         with gr.Column():
-        #             model_link = gr.Textbox(
-        #                 label="Model Link", placeholder="https://huggingface.co/decapoda-research/llama-7b-hf"
-        #             )
+        with gr.TabItem("🚀 Submit here! ", elem_id="mtheoremqa-tab-table", id=3):
+            with gr.Row():
+                gr.Markdown(SUBMIT_INTRODUCTION, elem_classes="markdown-text")
 
+            with gr.Row():
+                gr.Markdown("# ✉️✨ Submit your json file here!", elem_classes="markdown-text")
 
-            # with gr.Column():
+            with gr.Column():
 
-            #     input_file = gr.components.File(label = "Click to Upload a json File", file_count="single", type='binary')
-            #     submit_button = gr.Button("Submit Eval")
+                input_file = gr.components.File(label = "Click to Upload a json File", file_count="single", type='binary')
+                submit_button = gr.Button("Submit Results")
     
-            #     submission_result = gr.Markdown()
-            #     submit_button.click(
-            #         add_new_eval,
-            #         inputs = [
-            #             input_file,
-            #             model_name_textbox,
-            #             revision_name_textbox,
-            #             model_link,
-            #         ],
-            #     )
-
-
-    # def refresh_data():
-    #     value1 = get_baseline_df()
-    #     return value1
-
-    # with gr.Row():
-    #     data_run = gr.Button("Refresh")
-    #     data_run.click(on_filter_model_size_method_change, inputs=[checkbox_group], outputs=data_component)
+                submission_result = gr.Markdown()
+                submit_button.click(
+                    add_new_eval,
+                    inputs=[input_file],
+                )
 
 
 block.launch()

leaderboard/results.csv

@@ -1,25 +0,0 @@
-Model (CoT),TheoremQA,MATH,GSM
-Mistral-v0.2-base,19.2,10.2,36.2
-Mixtral-7x8B-base,23.2,22.1,58.4
-Qwen-1.5-7B,14.2,13.3,54.1
-Qwen-1.5-14B,14,25.2,61.6
-Qwen-1.5-72B,29.3,35.1,77.6
-Yi-6B,12,5.8,32.6
-Yi-34B,23.2,15.9,67.9
-ChatGLM3-6B,11.3,25.7,72.3
-Gemma-7B,21.5,24.3,46.4
-LLaMA-2-13B,10.9,5,29.6
-LLeMMA-7B,17.2,18,36.4
-LLeMMA-34B,21.1,25,71.9
-InternLM2-7B,7.8,20.2,70.8
-InternLM2-20B,19.5,25.5,76.1
-Deepseek-7B,15.7,6.4,17.4
-Deepseek-67B,25.3,15.9,66.5
-GPT-4-0409,0,69.2,94.5
-InternLM-Math-20B,17.1,37.7,82.9
-Deepseek-Math-7B,27.1,36.2,64.2
-Deepseek-Math-7B-Instruct,23.7,46.8,82.9
-WizardMath-7B-1.1,11.7,33,83.2
-MetaMath-Mistral-7B,16.5,28.2,77.7
-Abel-7B-002,19.3,29.5,83.2
-OpenMath-Mistral-7B,13.1,44.5,80.2

requirements.txt

@@ -1,2 +1,2 @@
-gradio==3.23.0
+gradio==4.26.0
 pandas==2.0.0

utils.py

@@ -1,20 +1,26 @@
+import pandas as pd
+import gradio as gr
+import csv
+import json
+import os
+import shutil
+from huggingface_hub import Repository
+
+HF_TOKEN = os.environ.get("HF_TOKEN")
+
 MODEL_INFO = [
-    "Model Name (clickable)",
+    "Model (CoT)",
+    "Avg",
     "TheoremQA",
     "MATH",
     "GSM",
     ]
 
-MODEL_INFO_TAB_QUALITY = [
-    "Model Name (clickable)",
-    "Quality Score",
-    "Selected Score"
-]
-
-
-DATA_TITILE_TYPE = ['markdown', 'number', 'number', 'number']
+DATA_TITILE_TYPE = ['markdown', 'number', 'number', 'number', 'number']
 
-CSV_DIR = "./leaderboard/results.csv"
+SUBMISSION_NAME = "science_leaderboard_submission"
+SUBMISSION_URL = os.path.join("https://huggingface.co/datasets/TIGER-Lab/", SUBMISSION_NAME)
+CSV_DIR = "./science_leaderboard_submission/results.csv"
 
 COLUMN_NAMES = MODEL_INFO
 
@@ -47,3 +53,46 @@ CITATION_BUTTON_TEXT = r"""@inproceedings{chen2023theoremqa,
   booktitle={The 2023 Conference on Empirical Methods in Natural Language Processing},
   year={2023}
 }"""
+
+SUBMIT_INTRODUCTION = """# Submit on TheoremQA Leaderboard Introduction
+
+## ⚠ Please note that you need to submit the json file with following format:
+
+```json
+{
+    "Model Name": "Model X",
+    "TheoremQA": 0.5,
+    "MATH": 0.5,
+    "GSM": 0.5
+}
+```
+After submitting, you can click the "Refresh" button to see the updated leaderboard(it may takes few seconds).
+
+"""
+def get_df():
+    repo = Repository(local_dir=SUBMISSION_NAME, clone_from=SUBMISSION_URL, use_auth_token=HF_TOKEN)
+    repo.git_pull()
+    df = pd.read_csv(CSV_DIR)
+    df['Avg'] = df[['TheoremQA', 'MATH', 'GSM']].mean(axis=1).round(1)
+    return df[COLUMN_NAMES]
+
+def add_new_eval(
+    input_file,
+):
+    if input_file is None:
+        return "Error! Empty file!"
+
+    upload_data=json.loads(input_file)
+    data_row = [upload_data['ModelName'], upload_data['TheoremQA'], upload_data['MATH'], upload_data['GSM']]
+
+    submission_repo = Repository(local_dir=SUBMISSION_NAME, clone_from=SUBMISSION_URL, use_auth_token=HF_TOKEN, repo_type="dataset")
+    submission_repo.git_pull()
+
+    with open(CSV_DIR, mode='a', newline='') as file:
+        writer = csv.writer(file)
+        writer.writerow(data_row)
+    
+    submission_repo.push_to_hub()
+
+def refresh_data():
+    return get_df()