slobodanmargetic988
a18db6a3da
Initial population of the weeyuga-benchmarks-public archive (PRIVATE staging visibility — flips public after Miljan + Stevan security audit sign-off per Sloba's 17:34Z dispatch). Contents: - README.md — public-facing intro (warns staging state, schema overview, citation pattern, license split) - LICENSE — CC-BY-4.0 default (auto-init from Gitea) - catalogue.json — schema_version=1.0-draft (locked once Tomas ratifies); 21 benchmarks indexed, 13 complete + 8 meta-only - methodology.md — mirror of WeeyugaWeb docs/BENCHMARKS/HARNESS.md (canonical methodology) - runs/<id>/run.jsonl|run.log|run.md|metadata.json — packaged copies of every run in WeeyugaWeb docs/BENCHMARKS/runs/* Run set covers: - Mission 1 (2026-04-28/29): pavilion-weeyuga-v1 + reconstructed v3 (96 calls, 16 models routed via weeyuga :11435) - Predator trio (2026-05-04): granite-4.1-8B + gemma-4-E4B-it + qwen3.5-9B - Predator qwen rerun (2026-05-04): qwen3.5-9B think500/nothink + qwen3-14B feasibility - A3B campaign (2026-05-04/05): pavilion-a3b + predator-a3b NGL matrix + ctx sweep + NGL+ctx 2D + NGL=6 deep dive - VPS50 CPU matrix + gemma-e4b CPU lane (2026-05-04/05) Visibility GATE: this repo stays private until Miljan G1-G4 audit and Stevan G3 credential audit both green. After sign-off, single API call flips visibility=public, anonymous read on, push-protection requires auth, issues moderate by default. No raw IPs, no SSH user@host strings, no /Users/ paths, no whisper transcripts in any of these files. Hardware names (pavilion, predator, vps50) are intentional and fine to share. Builder: WeeyugaWeb/scripts/benchmarks/build_catalogue.py (deterministic, idempotent, ~5s wall on 21 runs). Publish flow: WeeyugaWeb/scripts/benchmarks/publish_bench_run.py (builds packaged dirs, regenerates catalogue, optional --push to mirror into this repo, optional --deploy stub for cicd rsync). Owner: mac/benchmark-tester-ben (Ben). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
66 lines
3.6 KiB
Plaintext
66 lines
3.6 KiB
Plaintext
[21:24:02Z] === pavilion-a3b-bench run_id=487c8a51-7c5c-417d-a0c5-4b6e8491e11f ===
|
|
[21:24:02Z] Output: /Users/slobodan/projects/WeeyugaWeb/docs/BENCHMARKS/runs/487c8a51-7c5c-417d-a0c5-4b6e8491e11f.jsonl
|
|
[21:24:02Z] Creating run-mirror dir on Pavilion: D:\WeeyugaBench\runs\487c8a51-7c5c-417d-a0c5-4b6e8491e11f
|
|
[21:24:10Z] Polling for Qwen3-30B-A3B-UD-IQ2_XXS.gguf on Pavilion...
|
|
[21:24:13Z] download progress: 5.67 GB (poll 1/80)
|
|
[21:24:45Z] download progress: 5.75 GB (poll 2/80)
|
|
[21:25:19Z] download progress: 5.83 GB (poll 3/80)
|
|
[21:25:53Z] download progress: 5.91 GB (poll 4/80)
|
|
[21:26:25Z] download progress: 5.97 GB (poll 5/80)
|
|
[21:26:58Z] download progress: 6.05 GB (poll 6/80)
|
|
[21:27:31Z] download progress: 6.14 GB (poll 7/80)
|
|
[21:28:04Z] download progress: 6.24 GB (poll 8/80)
|
|
[21:28:36Z] download progress: 6.33 GB (poll 9/80)
|
|
[21:29:10Z] download progress: 6.41 GB (poll 10/80)
|
|
[21:29:42Z] download progress: 6.51 GB (poll 11/80)
|
|
[21:30:16Z] download progress: 6.61 GB (poll 12/80)
|
|
[21:30:49Z] download progress: 6.68 GB (poll 13/80)
|
|
[21:31:23Z] download progress: 6.75 GB (poll 14/80)
|
|
[21:31:56Z] download progress: 6.83 GB (poll 15/80)
|
|
[21:32:29Z] download progress: 6.9 GB (poll 16/80)
|
|
[21:33:03Z] download progress: 6.99 GB (poll 17/80)
|
|
[21:33:35Z] download progress: 7.07 GB (poll 18/80)
|
|
[21:34:09Z] download progress: 7.17 GB (poll 19/80)
|
|
[21:34:42Z] download progress: 7.26 GB (poll 20/80)
|
|
[21:35:16Z] download progress: 7.33 GB (poll 21/80)
|
|
[21:35:49Z] download progress: 7.4 GB (poll 22/80)
|
|
[21:36:22Z] download progress: 7.47 GB (poll 23/80)
|
|
[21:36:56Z] download progress: 7.55 GB (poll 24/80)
|
|
[21:37:29Z] download progress: 7.62 GB (poll 25/80)
|
|
[21:38:02Z] download progress: 7.69 GB (poll 26/80)
|
|
[21:38:35Z] download progress: 7.73 GB (poll 27/80)
|
|
[21:39:09Z] download progress: 7.81 GB (poll 28/80)
|
|
[21:39:42Z] download progress: 7.89 GB (poll 29/80)
|
|
[21:40:15Z] download progress: 7.97 GB (poll 30/80)
|
|
[21:40:50Z] download progress: 8.03 GB (poll 31/80)
|
|
[21:41:23Z] download progress: 8.11 GB (poll 32/80)
|
|
[21:41:57Z] download progress: 8.2 GB (poll 33/80)
|
|
[21:42:31Z] download progress: 8.29 GB (poll 34/80)
|
|
[21:43:05Z] download progress: 8.36 GB (poll 35/80)
|
|
[21:43:38Z] download progress: 8.44 GB (poll 36/80)
|
|
[21:44:11Z] download progress: 8.52 GB (poll 37/80)
|
|
[21:44:43Z] download progress: 8.59 GB (poll 38/80)
|
|
[21:45:17Z] download progress: 8.68 GB (poll 39/80)
|
|
[21:45:52Z] download progress: 8.73 GB (poll 40/80)
|
|
[21:46:25Z] download progress: 8.79 GB (poll 41/80)
|
|
[21:46:59Z] download progress: 8.87 GB (poll 42/80)
|
|
[21:47:32Z] download progress: 8.94 GB (poll 43/80)
|
|
[21:48:06Z] download progress: 9.01 GB (poll 44/80)
|
|
[21:48:39Z] download progress: 9.05 GB (poll 45/80)
|
|
[21:49:12Z] download progress: 9.13 GB (poll 46/80)
|
|
[21:49:45Z] download progress: 9.21 GB (poll 47/80)
|
|
[21:50:18Z] download progress: 9.28 GB (poll 48/80)
|
|
[21:50:51Z] download progress: 9.35 GB (poll 49/80)
|
|
[21:51:24Z] download progress: 9.42 GB (poll 50/80)
|
|
[21:51:57Z] GGUF ready: 9.51 GB
|
|
[21:51:57Z] Stopping any llama-server on Pavilion :11437...
|
|
[21:52:02Z] Starting llama-server with Qwen3-30B-A3B-UD-IQ2_XXS.gguf on :11437...
|
|
[21:52:10Z] Waiting for llama-server (give up to 6 min — IQ2_XXS mmap on Pavilion is slower)...
|
|
[21:59:22Z] TIMEOUT — llama-server didn't come up within 360s
|
|
[21:59:22Z] Server failed to start
|
|
[21:59:22Z] Stopping any llama-server on Pavilion :11437...
|
|
[21:59:26Z] === bench complete ===
|
|
[21:59:26Z] Synthesizing report...
|
|
[21:59:26Z] Report: /Users/slobodan/projects/WeeyugaWeb/docs/BENCHMARKS/runs/487c8a51-7c5c-417d-a0c5-4b6e8491e11f.md
|
|
[21:59:26Z] Mirroring run-dir to Pavilion...
|