File size: 11,597 Bytes
5309153 941f5e0 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 5309153 fe59685 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 |
import pandas as pd
import numpy as np
from datetime import datetime
from data import extract_model_data
from utils import COLORS
import gradio as gr
import plotly.express as px
import plotly.graph_objects as go
def get_time_series_summary_dfs(historical_df: pd.DataFrame) -> dict:
daily_stats = []
dates = sorted(historical_df['date'].unique())
for date in dates:
dd = historical_df[historical_df['date'] == date]
stats = {}
for platform in ['amd', 'nvidia']:
p, f, s = (dd[f'success_{platform}'].sum() if f'success_{platform}' in dd.columns else 0,
(dd[f'failed_multi_no_{platform}'].sum() + dd[f'failed_single_no_{platform}'].sum()) if f'failed_multi_no_{platform}' in dd.columns else 0,
dd[f'skipped_{platform}'].sum() if f'skipped_{platform}' in dd.columns else 0)
tot = p + f + s
stats.update({f'{platform}_passed': p, f'{platform}_failed': f, f'{platform}_skipped': s,
f'{platform}_failure_rate': (f / tot * 100) if tot > 0 else 0})
stats['date'] = date
daily_stats.append(stats)
failure_rate_data = []
for i, s in enumerate(daily_stats):
for p in ['amd', 'nvidia']:
chg = s[f'{p}_failure_rate'] - daily_stats[i-1][f'{p}_failure_rate'] if i > 0 else 0
failure_rate_data.append({'date': s['date'], 'failure_rate': s[f'{p}_failure_rate'],
'platform': p.upper(), 'change': chg})
def build_test_data(platform):
data = []
for i, s in enumerate(daily_stats):
for tt in ['passed', 'failed', 'skipped']:
chg = s[f'{platform}_{tt}'] - daily_stats[i-1][f'{platform}_{tt}'] if i > 0 else 0
data.append({'date': s['date'], 'count': s[f'{platform}_{tt}'],
'test_type': tt.capitalize(), 'change': chg})
return pd.DataFrame(data)
return {'failure_rates_df': pd.DataFrame(failure_rate_data),
'amd_tests_df': build_test_data('amd'),
'nvidia_tests_df': build_test_data('nvidia')}
def get_model_time_series_dfs(historical_df: pd.DataFrame, model_name: str) -> dict:
md = historical_df[historical_df.index.str.lower() == model_name.lower()]
if md.empty:
empty = pd.DataFrame({'date': [], 'count': [], 'test_type': [], 'change': []})
return {'amd_df': empty.copy(), 'nvidia_df': empty.copy()}
dates = sorted(md['date'].unique())
def build_platform_data(platform):
data = []
for i, date in enumerate(dates):
dd = md[md['date'] == date]
if dd.empty:
continue
r = dd.iloc[0]
p = r.get(f'success_{platform}', 0)
f = r.get(f'failed_multi_no_{platform}', 0) + r.get(f'failed_single_no_{platform}', 0)
s = r.get(f'skipped_{platform}', 0)
pr = md[md['date'] == dates[i-1]].iloc[0] if i > 0 and not md[md['date'] == dates[i-1]].empty else None
pc = pr.get(f'success_{platform}', 0) if pr is not None else 0
fc = (pr.get(f'failed_multi_no_{platform}', 0) + pr.get(f'failed_single_no_{platform}', 0)) if pr is not None else 0
sc = pr.get(f'skipped_{platform}', 0) if pr is not None else 0
data.extend([
{'date': date, 'count': p, 'test_type': 'Passed', 'change': p - pc},
{'date': date, 'count': f, 'test_type': 'Failed', 'change': f - fc},
{'date': date, 'count': s, 'test_type': 'Skipped', 'change': s - sc}
])
return pd.DataFrame(data)
return {'amd_df': build_platform_data('amd'), 'nvidia_df': build_platform_data('nvidia')}
def create_time_series_summary_gradio(historical_df: pd.DataFrame) -> dict:
empty_fig = lambda title: go.Figure().update_layout(title=title, height=500,
font=dict(size=16, color='#CCCCCC'), paper_bgcolor='#000000',
plot_bgcolor='#1a1a1a', margin=dict(b=130)) or go.Figure()
if historical_df.empty or 'date' not in historical_df.columns:
ef = empty_fig("No historical data available")
return {'failure_rates': ef, 'amd_tests': ef, 'nvidia_tests': ef}
daily_stats = []
for date in sorted(historical_df['date'].unique()):
dd = historical_df[historical_df['date'] == date]
counts = {'date': date}
for platform in ['amd', 'nvidia']:
tot_tests = tot_fails = p = f = s = 0
for _, row in dd.iterrows():
stats = extract_model_data(row)[0 if platform == 'amd' else 1]
tot = stats['passed'] + stats['failed'] + stats['error']
if tot > 0:
tot_tests += tot
tot_fails += stats['failed'] + stats['error']
p += stats['passed']
f += stats['failed'] + stats['error']
s += stats['skipped']
counts.update({f'{platform}_failure_rate': (tot_fails / tot_tests * 100) if tot_tests > 0 else 0,
f'{platform}_passed': p, f'{platform}_failed': f, f'{platform}_skipped': s})
daily_stats.append(counts)
fr_data = []
for i, s in enumerate(daily_stats):
for p in ['amd', 'nvidia']:
chg = s[f'{p}_failure_rate'] - daily_stats[i-1][f'{p}_failure_rate'] if i > 0 else 0
fr_data.append({'date': s['date'], 'failure_rate': s[f'{p}_failure_rate'],
'platform': p.upper(), 'change': chg})
def build_test_data(platform):
data = []
for i, s in enumerate(daily_stats):
for tt in ['passed', 'failed', 'skipped']:
chg = s[f'{platform}_{tt}'] - daily_stats[i-1][f'{platform}_{tt}'] if i > 0 else 0
data.append({'date': s['date'], 'count': s[f'{platform}_{tt}'],
'test_type': tt.capitalize(), 'change': chg})
return pd.DataFrame(data)
fr_df = pd.DataFrame(fr_data)
fig_fr = go.Figure()
for p, lc, mc in [('NVIDIA', '#76B900', '#FFFFFF'), ('AMD', '#ED1C24', '#404040')]:
d = fr_df[fr_df['platform'] == p]
if not d.empty:
fig_fr.add_trace(go.Scatter(x=d['date'], y=d['failure_rate'], mode='lines+markers',
name=p, line=dict(color=lc, width=3),
marker=dict(size=12, color=mc, line=dict(color=lc, width=2)),
hovertemplate=f'<b>{p}</b><br>Date: %{{x}}<br>Failure Rate: %{{y:.2f}}%<extra></extra>'))
fig_fr.update_layout(title="Overall Failure Rates Over Time", height=500,
font=dict(size=16, color='#CCCCCC'), paper_bgcolor='#000000', plot_bgcolor='#1a1a1a',
title_font_size=20, legend=dict(font=dict(size=16), bgcolor='rgba(0,0,0,0.5)',
orientation="h", yanchor="bottom", y=-0.4, xanchor="center", x=0.5),
xaxis=dict(title='Date', title_font_size=16, tickfont_size=14, gridcolor='#333333', showgrid=True),
yaxis=dict(title='Failure Rate (%)', title_font_size=16, tickfont_size=14, gridcolor='#333333', showgrid=True),
hovermode='x unified', margin=dict(b=130))
def create_line_fig(df, title):
fig = px.line(df, x='date', y='count', color='test_type',
color_discrete_map={"Passed": COLORS['passed'], "Failed": COLORS['failed'], "Skipped": COLORS['skipped']},
title=title, labels={'count': 'Number of Tests', 'date': 'Date', 'test_type': 'Test Type'})
fig.update_traces(mode='lines+markers', marker=dict(size=8), line=dict(width=3))
fig.update_layout(height=500, font=dict(size=16, color='#CCCCCC'), paper_bgcolor='#000000',
plot_bgcolor='#1a1a1a', title_font_size=20, legend=dict(font=dict(size=16),
bgcolor='rgba(0,0,0,0.5)', orientation="h", yanchor="bottom", y=-0.4, xanchor="center", x=0.5),
xaxis=dict(title_font_size=16, tickfont_size=14, gridcolor='#333333', showgrid=True),
yaxis=dict(title_font_size=16, tickfont_size=14, gridcolor='#333333', showgrid=True),
hovermode='x unified', margin=dict(b=130))
return fig
return {'failure_rates': fig_fr,
'amd_tests': create_line_fig(build_test_data('amd'), "AMD Test Results Over Time"),
'nvidia_tests': create_line_fig(build_test_data('nvidia'), "NVIDIA Test Results Over Time")}
def create_model_time_series_gradio(historical_df: pd.DataFrame, model_name: str) -> dict:
def empty_figs():
ef = lambda plat: go.Figure().update_layout(title=f"{model_name.upper()} - {plat} Results Over Time",
height=500, font=dict(size=16, color='#CCCCCC'), paper_bgcolor='#000000',
plot_bgcolor='#1a1a1a', margin=dict(b=130)) or go.Figure()
return {'amd_plot': ef('AMD'), 'nvidia_plot': ef('NVIDIA')}
if historical_df.empty or 'date' not in historical_df.columns:
return empty_figs()
md = historical_df[historical_df.index.str.lower() == model_name.lower()]
if md.empty:
return empty_figs()
dates = sorted(md['date'].unique())
def build_data(platform):
data = []
for i, date in enumerate(dates):
dd = md[md['date'] == date]
if dd.empty:
continue
r = dd.iloc[0]
passed = r.get(f'success_{platform}', 0)
failed = r.get(f'failed_multi_no_{platform}', 0) + r.get(f'failed_single_no_{platform}', 0)
skipped = r.get(f'skipped_{platform}', 0)
pc = fc = sc = 0
if i > 0:
prev_dd = md[md['date'] == dates[i-1]]
if not prev_dd.empty:
pr = prev_dd.iloc[0]
pc = pr.get(f'success_{platform}', 0)
fc = pr.get(f'failed_multi_no_{platform}', 0) + pr.get(f'failed_single_no_{platform}', 0)
sc = pr.get(f'skipped_{platform}', 0)
data.extend([
{'date': date, 'count': passed, 'test_type': 'Passed', 'change': passed - pc},
{'date': date, 'count': failed, 'test_type': 'Failed', 'change': failed - fc},
{'date': date, 'count': skipped, 'test_type': 'Skipped', 'change': skipped - sc}
])
return pd.DataFrame(data)
def create_fig(df, platform):
fig = px.line(df, x='date', y='count', color='test_type',
color_discrete_map={"Passed": COLORS['passed'], "Failed": COLORS['failed'], "Skipped": COLORS['skipped']},
title=f"{model_name.upper()} - {platform} Results Over Time",
labels={'count': 'Number of Tests', 'date': 'Date', 'test_type': 'Test Type'})
fig.update_traces(mode='lines+markers', marker=dict(size=8), line=dict(width=3))
fig.update_layout(height=500, font=dict(size=16, color='#CCCCCC'), paper_bgcolor='#000000',
plot_bgcolor='#1a1a1a', title_font_size=20, legend=dict(font=dict(size=16),
bgcolor='rgba(0,0,0,0.5)', orientation="h", yanchor="bottom", y=-0.4, xanchor="center", x=0.5),
xaxis=dict(title_font_size=16, tickfont_size=14, gridcolor='#333333', showgrid=True),
yaxis=dict(title_font_size=16, tickfont_size=14, gridcolor='#333333', showgrid=True),
hovermode='x unified', margin=dict(b=130))
return fig
return {'amd_plot': create_fig(build_data('amd'), 'AMD'),
'nvidia_plot': create_fig(build_data('nvidia'), 'NVIDIA')} |