Gahow Wang gahow

0 Followers · 0 Following

• Joined on 2026-04-03

Repositories

18

Projects Packages Public Activity Starred Repositories

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 18:48:53 +00:00

3e32ea609f Cover Qwen235 MNS128 graph buckets

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 18:47:01 +00:00

40bac6dbf4 Add Qwen235 same-stack Frontier profiling campaign

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 18:44:58 +00:00

e631f6a269 Add Qwen235 Frontier latency surface runner

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 18:42:31 +00:00

a3c8cb5808 Add Qwen235 Frontier MoE portability smoke

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 18:39:28 +00:00

f4813cf537 Add Qwen235 vLLM 0.20 real surface runner

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 18:25:20 +00:00

c8383c9c4c Launch Qwen30 fixed pressure comparison

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 17:45:24 +00:00

7ea9635878 Record fixed PD pressure calibration result

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 17:26:17 +00:00

33b73afe9b Narrow fixed PD pressure sweep around load knee

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 17:13:55 +00:00

909a80f0a6 Reuse validated FlashInfer cache for pressure probe

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 17:08:04 +00:00

9837fa5133 Match Qwen30 probe file descriptor limit

gahow pushed to codex/fixed-pd-pressure-probe-20260719 at gahow/aituner 2026-07-18 16:54:55 +00:00

6726318792 Profile fixed PD workload pressure against trace anchor

gahow created branch codex/fixed-pd-pressure-probe-20260719 in gahow/aituner 2026-07-18 16:54:55 +00:00

gahow pushed to codex/fidelity-prefix-pilot-20260714 at gahow/aituner 2026-07-18 09:27:18 +00:00

e98608911e Resume valid Qwen30 latency cells safely

gahow pushed to codex/fidelity-prefix-pilot-20260714 at gahow/aituner 2026-07-18 03:56:53 +00:00

f11daa6776 Fix Qwen30 exact trace served-model routing

gahow pushed to codex/fidelity-prefix-pilot-20260714 at gahow/aituner 2026-07-18 03:15:29 +00:00

b9523cef5c Pin DeepGEMM nvcc for Qwen235 smoke

gahow pushed to codex/fidelity-prefix-pilot-20260714 at gahow/aituner 2026-07-18 03:11:13 +00:00

b7f9cef9c5 Keep Qwen30 JIT cache scoped to experiment

gahow pushed to codex/fidelity-prefix-pilot-20260714 at gahow/aituner 2026-07-18 03:09:30 +00:00

ecb45cf762 Extend Qwen30 JIT readiness for latency surface

gahow pushed to codex/fidelity-prefix-pilot-20260714 at gahow/aituner 2026-07-18 02:57:53 +00:00

686b050517 Isolate Qwen235 DeepGEMM JIT cache

gahow pushed to codex/fidelity-prefix-pilot-20260714 at gahow/aituner 2026-07-18 02:50:15 +00:00

28ffb22bee Create Qwen30 case output before launch logging

gahow pushed to codex/fidelity-prefix-pilot-20260714 at gahow/aituner 2026-07-18 02:48:52 +00:00

a01c1206ab Launch frozen Qwen30 latency case surfaces