Add duplicate submission graph

src/app.py

@@ -3,7 +3,7 @@ import gradio as gr
 from chain_data import sync_chain
 from leaderboard import create_leaderboard, create_dropdown, create_baseline
 from model_demo import create_demo
-from submissions import create_submissions, DROPDOWN_OPTIONS
+from submissions import create_submissions, create_duplicate_submissions_plot, DROPDOWN_OPTIONS
 from validator_states import create_validator_states
 from validator_weights import create_weights
 from wandb_data import sync
@@ -49,6 +49,9 @@ def main():
 
             filter_dropdown.change(lambda submission_filter: create_submissions(submission_filter), [filter_dropdown], [submissions_dataframe])
 
+            duplicate_submissions_plot = gr.Plot()
+            duplicate_submissions_plot.attach_load_event(lambda: create_duplicate_submissions_plot(), None)
+
         with gr.Tab("Model Demo"):
             create_demo()
     app.launch()

src/submissions.py

@@ -1,7 +1,9 @@
+from collections import defaultdict
 from enum import Enum
 
 import gradio as gr
 import pandas as pd
+import plotly.graph_objects as go
 
 from chain_data import sync_chain, fetch_commitments
 from src import Key
@@ -18,14 +20,10 @@ class SubmissionStatus(Enum):
 
     @staticmethod
     def get_status(run: Run, hotkey: Key, coldkey: Key, block: int, revision: str) -> "SubmissionStatus":
-        blacklisted_keys = get_blacklisted_keys()
-        if hotkey in blacklisted_keys.hotkeys or coldkey in blacklisted_keys.coldkeys:
+        if is_blacklisted(hotkey, coldkey):
             return SubmissionStatus.BLACKLISTED
 
-        if not any(
-            submission.hotkey == hotkey and submission.revision == revision
-            for submission in blacklisted_keys.duplicate_selection.safe_submissions
-        ):
+        if is_duplicate_submission(hotkey, revision):
             return SubmissionStatus.DUPLICATE
 
         if hotkey in run.submissions and block > run.submissions[hotkey].info.block and hotkey not in run.invalid_submissions:
@@ -40,9 +38,85 @@ class SubmissionStatus(Enum):
         return SubmissionStatus.PENDING
 
 
+def is_blacklisted(hotkey: Key, coldkey: Key) -> bool:
+    return hotkey in get_blacklisted_keys().hotkeys or coldkey in get_blacklisted_keys().coldkeys
+
+
+def is_duplicate_submission(hotkey: Key, revision: str) -> bool:
+    return not any(
+        submission.hotkey == hotkey and submission.revision == revision
+        for submission in get_blacklisted_keys().duplicate_selection.safe_submissions
+    )
+
+
 DROPDOWN_OPTIONS = [status.value[0] for status in SubmissionStatus]
 
 
+def create_duplicate_submissions_plot() -> gr.Plot:
+    sync_chain()
+
+    submissions_by_coldkey = defaultdict(lambda: ([], []))
+
+    for hotkey, commitment in fetch_commitments().items():
+        neuron = get_neurons().get(hotkey)
+        if not neuron:
+            continue
+
+        coldkey = neuron.coldkey
+
+        if is_blacklisted(hotkey, coldkey):
+            continue
+
+        if is_duplicate_submission(hotkey, commitment.revision):
+            submissions_by_coldkey[coldkey][0].append(commitment)
+        else:
+            submissions_by_coldkey[coldkey][1].append(commitment)
+
+    submissions_by_coldkey = dict(sorted(
+        submissions_by_coldkey.items(),
+        key=lambda x: len(x[1][0]) + len(x[1][1]),
+        reverse=True
+    ))
+
+    figure = go.Figure()
+
+    for coldkey, (duplicate_commitments, safe_commitments) in submissions_by_coldkey.items():
+        urls_safe = [c.url for c in safe_commitments]
+        figure.add_trace(go.Bar(
+            x=[f"{coldkey[:6]}..."],
+            y=[len(safe_commitments)],
+            name="Safe",
+            marker_color="green", # type: ignore
+            hovertemplate="<br>".join([
+                "Submissions (%{y}):<br>" + "<br>".join(urls_safe)
+            ]) + "<extra></extra>"
+        ))
+
+        urls_duplicate = [c.url for c in duplicate_commitments]
+        figure.add_trace(go.Bar(
+            x=[f"{coldkey[:6]}..."],
+            y=[len(duplicate_commitments)],
+            name="Duplicate",
+            marker_color="red", # type: ignore
+            hovertemplate="<br>".join([
+                "Submissions (%{y}):<br>" + "<br>".join(urls_duplicate)
+            ]) + "<extra></extra>"
+        ))
+
+    figure.update_layout(
+        title="Duplicate Submissions",
+        xaxis_title="Coldkey",
+        yaxis_title="# of Submissions",
+        barmode="stack",
+        showlegend=False,
+        autosize=True,
+        margin=dict(l=50, r=50, t=50, b=50),
+        template="plotly_dark"
+    )
+
+    return gr.Plot(figure)
+
+
 def create_submissions(submission_filters: list[str]) -> gr.Dataframe:
     data: list[list] = []
     sync_chain()