Exclude scenariobench from band total geo mean; label tok/s

This commit is contained in:
PeterChrz 2026-10-01 14:53:59 -04:00
parent 6c19822235
commit f96fd1ad77
Signed by untrusted user who does not match committer: pch
GPG key ID: 8F0826ECF7302C63
2 changed files with 2 additions and 12 deletions

View file

@ -105,8 +105,6 @@ public function handle(): int
$sessions[$key]['passed'] += $agg['passed'];
$sessions[$key]['failed'] += $agg['failed'];
$sessions[$key]['tokens'] += $agg['total_tokens'];
$sessions[$key]['ln_sum'] += $agg['ln_sum'];
$sessions[$key]['ln_n'] += $agg['ln_n'];
$sessions[$key]['runs'] = $band['runs']->merge([$agg]);
}
@ -231,8 +229,6 @@ private function aggregateRun(object $run)
->selectRaw('count(*) filter (where not ok) as failed')
->selectRaw('sum(gen_tokens) as total_tokens')
->selectRaw('sum(gen_tokens) / nullif(sum(wall), 0) as gen_tps')
->selectRaw('sum(ln(gen_tokens::float8 / nullif(wall, 0))) filter (where wall > 0 and gen_tokens > 0) as ln_sum')
->selectRaw('count(*) filter (where wall > 0 and gen_tokens > 0) as ln_n')
->first();
return collect([[
@ -246,8 +242,6 @@ private function aggregateRun(object $run)
'weighted_user_tps' => null,
'gen_tps' => $this->roundOrNull($agg->gen_tps),
'total_tokens' => (int) ($agg->total_tokens ?? 0),
'ln_sum' => (float) ($agg->ln_sum ?? 0),
'ln_n' => (int) ($agg->ln_n ?? 0),
]]);
}
@ -295,8 +289,6 @@ private function aggregateScenarioBand(array $runs): array
->selectRaw('count(*) filter (where not ok) as failed')
->selectRaw('sum(gen_tokens) as total_tokens')
->selectRaw('sum(gen_tokens) / nullif(sum(wall), 0) as gen_tps')
->selectRaw('sum(ln(gen_tokens::float8 / nullif(wall, 0))) filter (where wall > 0 and gen_tokens > 0) as ln_sum')
->selectRaw('count(*) filter (where wall > 0 and gen_tokens > 0) as ln_n')
->first();
// Thinking level: keep it only when every scenario row in the band
@ -316,8 +308,6 @@ private function aggregateScenarioBand(array $runs): array
'weighted_user_tps' => null,
'gen_tps' => $this->roundOrNull($agg->gen_tps),
'total_tokens' => (int) ($agg->total_tokens ?? 0),
'ln_sum' => (float) ($agg->ln_sum ?? 0),
'ln_n' => (int) ($agg->ln_n ?? 0),
];
}

View file

@ -167,13 +167,13 @@ class="w-64"
<span class="text-sm text-zinc-500">{{ $session->band->runner_name ?? 'anonymous' }}</span>
<span class="text-xs text-zinc-400">{{ optional($session->band->ran_at)->format('M j Y') }}</span>
</div>
{{-- Band total: geometric-mean tok/s across all observations + total tokens --}}
{{-- Band total: geometric-mean tok/s across decode-suite observations + total tokens --}}
@if ($session->band->total_tps !== null)
<div class="flex flex-wrap items-center gap-x-4 gap-y-1 border-b border-zinc-200 px-4 py-2 dark:border-zinc-700">
<span class="text-xs font-semibold uppercase tracking-wider text-zinc-400">{{ __('Band total') }}</span>
<span class="text-sm">
<span class="font-mono font-semibold">{{ number_format($session->band->total_tps, 1) }}</span>
<span class="text-xs text-zinc-400">{{ __('tok/s (geo)') }}</span>
<span class="text-xs text-zinc-400">{{ __('tok/s') }}</span>
</span>
<span class="text-sm">
<span class="font-mono font-semibold">{{ number_format($session->band->total_tokens) }}</span>