upload gpt-4o-mini battles

.gitattributes

@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.md filter=lfs diff=lfs merge=lfs -text
+*.py filter=lfs diff=lfs merge=lfs -text
+*.jsonl filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,13 +1,3 @@
----
-title: Arena Example
-emoji: 📚
-colorFrom: indigo
-colorTo: green
-sdk: gradio
-sdk_version: 3.40.0
-app_file: app.py
-pinned: false
-license: apache-2.0
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+version https://git-lfs.github.com/spec/v1
+oid sha256:826aceaffa9b514858b128b5478a9568e89ac9e596fe1c379cff219e9465b62d
+size 254

app.py

@@ -1,184 +1,3 @@
-"""
-Usage:
-python3 qa_browser.py --share
-"""
-
-import argparse
-from collections import defaultdict
-from glob import glob
-import re
-import gradio as gr
-import pandas as pd
-
-questions = []
-
-question_selector_map = {}
-category_selector_map = defaultdict(list)
-
-
-def display_question(category_selector, request: gr.Request):
-    # print(category_selector)
-    choices = category_selector_map[category_selector]
-    return gr.Dropdown.update(
-        value=choices[0],
-        choices=choices,
-    )
-
-
-def display_pairwise_answer(
-    question_selector, model_selector1, model_selector2, request: gr.Request
-):
-    q = question_selector_map[question_selector]
-    qid = q["question_id"]
-
-    ans1 = q["conversation_a"]
-    ans2 = q["conversation_b"]
-
-    chat_mds = pairwise_to_gradio_chat_mds(q, ans1, ans2)
-    
-    return chat_mds
-
-
-newline_pattern1 = re.compile("\n\n(\d+\. )")
-newline_pattern2 = re.compile("\n\n(- )")
-
-
-def post_process_answer(x):
-    """Fix Markdown rendering problems."""
-    x = x.replace("\u2022", "- ")
-    x = re.sub(newline_pattern1, "\n\g<1>", x)
-    x = re.sub(newline_pattern2, "\n\g<1>", x)
-    return x
-
-
-def pairwise_to_gradio_chat_mds(question, ans_a, ans_b, turn=None):
-    end = question["turn"] * 3
-
-    mds = [""] * end
-    base = 0
-    for i in range(0, end, 3):
-        mds[i] = "##### `User`\n" + question["conversation_a"][base]["content"].strip()
-        mds[i + 1] = f"##### `{question['model_a']}`\n" + post_process_answer(
-            ans_a[base + 1]["content"].strip()
-        )
-        mds[i + 2] = f"##### `{question['model_b']}`\n" + post_process_answer(
-            ans_b[base + 1]["content"].strip()
-        )
-        base += 2
-        
-    winner = question["winner"] if "tie" in question["winner"] else question[question["winner"]]
-    mds += [f"##### Vote: {winner}"]
-    mds += [""] * (16 - len(mds))
-    
-    return mds
-
-
-def build_question_selector_map():
-    global question_selector_map, category_selector_map
-
-    # Build question selector map
-    for q in questions:
-        preview = q["conversation_a"][0]["content"][:128] + "..."
-        question_selector_map[preview] = q
-        category_selector_map[q["category"]].append(preview)
-
-
-def build_pairwise_browser_tab():
-    global question_selector_map, category_selector_map
-
-    num_sides = 2
-    num_turns = 5
-    side_names = ["A", "B"]
-
-    question_selector_choices = list(question_selector_map.keys())
-    category_selector_choices = list(category_selector_map.keys())
-    print(category_selector_choices)
-    # Selectors
-    with gr.Row():
-        with gr.Column(scale=1, min_width=200):
-            category_selector = gr.Dropdown(
-                choices=category_selector_choices,
-                # value="Instruction Following",
-                label="Category", container=False
-            )
-        with gr.Column(scale=100):
-            question_selector = gr.Dropdown(
-                choices=question_selector_choices, label="Question", container=False
-            )
-
-    # Conversation
-    chat_mds = []
-    for i in range(num_turns):
-        chat_mds.append(gr.Markdown(elem_id=f"user_question_{i+1}"))
-        with gr.Row():
-            for j in range(num_sides):
-                with gr.Column(scale=100):
-                    chat_mds.append(gr.Markdown())
-
-                if j == 0:
-                    with gr.Column(scale=1, min_width=8):
-                        gr.Markdown()
-    chat_mds.append(gr.Markdown())
-
-    # Callbacks
-    category_selector.change(display_question, [category_selector], [question_selector])
-    question_selector.change(
-        display_pairwise_answer,
-        [question_selector],
-        chat_mds,
-    )
-
-    return (category_selector,)
-
-
-def load_demo():
-    dropdown_update = gr.Dropdown.update(value="Math")
-    return dropdown_update
-
-
-def build_demo():
-    build_question_selector_map()
-
-    with gr.Blocks(
-        title="Chatbot Arena Samples",
-        theme=gr.themes.Base(text_size=gr.themes.sizes.text_lg),
-    ) as demo:
-        gr.Markdown(
-            """
-# Chatbot Arena Samples
-We randomly sample 20 battles from each category using seed 42.
-| [Paper](https://arxiv.org/abs/2403.04132) | [Leaderboard](https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard) |
-"""
-        )
-        (category_selector,) = build_pairwise_browser_tab()
-        demo.load(load_demo, [], [category_selector])
-
-    return demo
-
-
-def load_questions(directory: str):
-    import json
-    """Load questions from a file."""
-    questions = []
-    for file in glob(directory):
-        with open(file, "r") as ques_file:
-            for line in ques_file:
-                if line:
-                    questions.append(json.loads(line))
-    return questions
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--host", type=str, default="0.0.0.0")
-    parser.add_argument("--port", type=int)
-    parser.add_argument("--share", action="store_true")
-    args = parser.parse_args()
-    print(args)
-    
-    questions = load_questions("data/*.jsonl")
-
-    demo = build_demo()
-    demo.launch(
-        server_name=args.host, server_port=args.port, share=args.share, max_threads=200
-    )
+version https://git-lfs.github.com/spec/v1
+oid sha256:6be9db2a4da8b10357d2cf03690e57af8c49b49620bc452325d94a8b3d134ce4
+size 5183

data/sample_gpt-4o-mini.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a66497ae13a735ef8acaa4a95c2bbcad16233f48637e3f4f57afff080d3ff529
+size 11210815