sigilant-runner · Phi-3.5-mini-instruct · L4 · vllm · 4 configs

Config | TPS | TPS p95 | TTFT | TTFT p95 | ITL | PPL | TPS% | TTFT% | PPL% | Score
FP16_BASELINE · ctx:16384 · kv:k16v16 · long  <- best | 34.7 | n/a | 2111.9 | n/a | 28.96 | 3.07 | 99.7 | 100.0 | 98.7 | 99
FP16_BASELINE · ctx:16384 · kv:k8v8 · default | 34.8 | n/a | 2817.1 | n/a | 28.86 | 3.03 | 100.0 | 75.0 | 100.0 | 95
FP16_BASELINE · ctx:32768 · kv:k8v8 · long | 34.7 | n/a | 2870.7 | n/a | 28.95 | 3.07 | 99.7 | 73.6 | 98.7 | 94
FP16_BASELINE · ctx:8192 · kv:k16v16 · default | 34.7 | n/a | 2938.8 | n/a | 28.93 | 3.03 | 99.7 | 71.9 | 100.0 | 94

Best config: FP16_BASELINE · ctx:16384 · kv:k16v16 · long
Auto baseline compare: score Δ=0.00 TPS Δ=0.00 TTFT Δ=0.0ms PPL Δ=0.00
Confidence: target=medium gap_before=4.04% var_before=n/a% replay=False(disabled) gap_after=4.04%

PPL is a quality proxy, not production validation.
Full production safety and long-context certification require Sigilant Optimizer.
