Gahow Wang 0794efa249 Reduce ablation probe budget to 3 per trial for tractability ...

First TP1 baseline probe under scale=0.2 ran ~6min (severe overload, 260
preemptions on the lighter half of the trace; TP1 is decode-bound and the
arrival-lag early-stop does not cut a decode-drain-bound probe). Cut
search.max_probes 5->3 to bound binary-search steps per trial. Caps stay
at elapsed=180/lag=30. Both configs still differ only in use_harness +
study_id.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

2026-06-16 20:01:19 +08:00

..

dash0_smoke_proposals

Add study tune loop and smoke configs

2026-04-04 22:29:59 +08:00

stopb_27b_ab

Add 27B TP A/B (deterministic ground-truth: does TP2 beat TP1 per-GPU)

2026-06-15 18:39:54 +08:00

trace_windows

Add study tune loop and smoke configs

2026-04-04 22:29:59 +08:00

capability.example.json

Initial AITuner study orchestrator

2026-04-04 21:26:37 +08:00

compare.example.json

Add multi-window baseline vs tuned compare flow

2026-04-11 13:51:54 +08:00

dash0_llm_10min_study_run1f.json

Tighten LLM proposal schema

2026-04-04 23:24:32 +08:00

dash0_manual_trial2_maxprobes6.json

Add deeper infeasible probe diagnostics

2026-04-05 01:44:38 +08:00

dash0_manual_trial2_probe_003125.json

Add targeted low-threshold probe specs

2026-04-05 02:08:27 +08:00

dash0_manual_trial2_probe_0015625.json

Add targeted low-threshold probe specs

2026-04-05 02:08:27 +08:00

dash0_manual_trial2_proposal.json

Add deeper infeasible probe diagnostics

2026-04-05 01:44:38 +08:00

dash0_qwen27b_ablation_harness_on.json

Reduce ablation probe budget to 3 per trial for tractability

2026-06-16 20:01:19 +08:00

dash0_qwen27b_ablation_naive_off.json

Reduce ablation probe budget to 3 per trial for tractability

2026-06-16 20:01:19 +08:00

dash0_qwen27b_stopB_loop.json

Add 27B Stop-B agentic-loop config (harness-driven, GPUs 2-7)

2026-06-16 09:08:46 +08:00

dash0_qwen27b_tight_slo_baseline_proposal.json

Align qwen27b baseline proposal to TP1 run script

2026-04-11 00:40:05 +08:00

dash0_qwen27b_tight_slo_run1.json

Add trace length bucket tuning support

2026-04-07 11:03:16 +08:00

dash0_qwen27b_tight_slo_run2.json

Add trace length bucket tuning support

2026-04-07 11:03:16 +08:00

dash0_qwen27b_tight_slo_run3.json

Add trace length bucket tuning support

2026-04-07 11:03:16 +08:00

dash0_qwen27b_tight_slo_run4_0_8k.json

Fix topology-aware incumbents for qwen27b tuning

2026-04-11 00:32:41 +08:00

dash0_qwen27b_tp_ab.json

Pin 27B A/B to GPUs 2-7 (route around leaked GPU0/1 memory)

2026-06-15 23:01:22 +08:00

dash0_qwen30b_a3b_community_vllm020_harness.json

Use time-compressed community vllm ablation

2026-05-02 10:03:59 +08:00

dash0_qwen30b_a3b_community_vllm020_noharness.json

Use time-compressed community vllm ablation

2026-05-02 10:03:59 +08:00

dash0_qwen30b_a3b_stopA_fulldata.json

Fix Stop-A validation config: system vllm, cap max-model-len

2026-06-15 15:22:48 +08:00

dash0_qwen30b_a3b_stopA_on.json

Add Stop-A ON config (adaptive_stop enabled + boundary guard) for A/B

2026-06-15 16:25:24 +08:00

dash0_qwen30b_a3b_stopB_e2e.json

Phase 5: widen search.high to 1.0 to force multi-iteration Stop-B convergence

2026-06-15 17:12:32 +08:00

dash0_qwen235b_decode_thinking_baseline.json

Add decode-only study mode support

2026-04-09 11:23:17 +08:00

dash0_qwen235b_decode_thinking_run1.json

Document decode harness one-shot mechanism

2026-05-02 06:25:06 +08:00

dash0_qwen235b_decode_thinking_run2_tpot40.json

Document decode harness one-shot mechanism

2026-05-02 06:25:06 +08:00

dash0_qwen235b_prefill_thinking_baseline.json

Add qwen235b prefill-only tuning support

2026-04-11 21:00:02 +08:00

dash0_qwen235b_prefill_thinking_run1_ttft.json

Add qwen235b prefill-only tuning support

2026-04-11 21:00:02 +08:00

dash0_qwen235b_prefill_thinking_run2_ttft_tight.json

Add qwen235b prefill docs and tight TTFT spec

2026-04-12 11:24:23 +08:00

dash0_qwen235b_prefill_thinking_run3_ttft_tight_0323.json

configs: add qwen235b prefill tight ttft 0323 study

2026-04-13 09:39:32 +08:00

dash0_smoke_study.json

Add probe early stop guards

2026-04-04 22:58:33 +08:00

dash1_qwen235b_prefill_thinking_7day_compare.json

compare: add multi-candidate runner

2026-04-13 20:50:39 +08:00

dash1_qwen235b_prefill_thinking_run4_ttft_tight_0_32k.json

configs: add qwen235b prefill 0-32k study

2026-04-17 19:20:44 +08:00

dash1_qwen235b_prefill_thinking_run5_ttft_1s_2s_0_32k.json

configs: add q235b prefill 1s 2s 0-32k study

2026-04-17 19:25:32 +08:00

study.example.json

Add codex and bailian LLM provider presets

2026-04-07 11:31:26 +08:00