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>
2.8 KiB
2.8 KiB
Predator Qwen rerun — fba9d9b1-cc5d-40bc-9e21-beafbb72c65d
- Started: 2026-05-04T17:49:07Z
- Driver host: Slobodans-MacBook-Air.local
- Node: predator (http://10.8.0.7:11436)
- Engine: llamacpp
- Harness: predator-qwen-rerun-1
- Purpose: Sloba's follow-up to trio-bench: budget-capped thinking, thinking-off comparison, 14B-on-6GB-GPU feasibility test
Configurations
| Key | Cell | GGUF | Reasoning budget | Sampling | System prompt |
|---|---|---|---|---|---|
| qwen-9b-think500 | predator:llamacpp:qwen3.5:9b-q4km-think500 |
Qwen3.5-9B-Q4_K_M.gguf |
500 | T=1.0 top_p=0.95 top_k=20 (thinking-mode) | — |
| qwen-9b-nothink | predator:llamacpp:qwen3.5:9b-q4km-nothink |
Qwen3.5-9B-Q4_K_M.gguf |
0 | T=0.7 top_p=0.8 top_k=20 (instruct-mode) | Answer directly without showing reasoning. |
| qwen3-14b | predator:llamacpp:qwen3:14b-q4km |
Qwen3-14B-Q4_K_M.gguf |
1024 | T=0.7 top_p=0.8 top_k=20 (Qwen3 default) | — |
llama-bench synthetic throughput (pp512 + tg128)
| Cell | pp512 (t/s) | tg128 (t/s) |
|---|---|---|
| qwen-9b-think500 | 61.32 | 7.68 |
| qwen-9b-nothink | 65.19 | 7.56 |
| qwen3-14b | 52.96 | 0.86 |
VRAM snapshot (used / free / total MB, post-load)
| Cell | Used | Free | Total |
|---|---|---|---|
| qwen-9b-think500 | 5887 | 143 | 6144 |
| qwen-9b-nothink | 5918 | 112 | 6144 |
| qwen3-14b | 5883 | 147 | 6144 |
Per-prompt latency, throughput, and thinking-vs-answer split
| Cell | Prompt | Cold dur (ms) | Cold tok/s | Warm tok/s mean | Warm dur p50 (ms) | Reasoning chars (warm mean) | Answer chars (warm mean) | Errors |
|---|---|---|---|---|---|---|---|---|
| qwen-9b-think500 | hello | 6773 | 9.4 | 13.8 | 4632 | 222 | 0 | 0 |
| qwen-9b-think500 | P-MEDIUM | 35949 | 14.2 | 15.2 | 33586 | 2151 | 45 | 0 |
| qwen-9b-think500 | P-HARD | 53568 | 14.9 | 15.2 | 51802 | 1938 | 1243 | 0 |
| qwen-9b-nothink | hello | 2372 | 5.5 | 9.5 | 1350 | 0 | 30 | 0 |
| qwen-9b-nothink | P-MEDIUM | 7338 | 14.4 | 14.3 | 7062 | 0 | 539 | 0 |
| qwen-9b-nothink | P-HARD | 20870 | 14.8 | 14.9 | 23327 | 0 | 1493 | 0 |
| qwen3-14b | hello | 98037 | 0.7 | 1.1 | 58891 | 264 | 11 | 0 |
| qwen3-14b | P-MEDIUM | 299410 | 1.1 | 1.3 | 210410 | 839 | 554 | 0 |
| qwen3-14b | P-HARD | — | — | — | — | — | — | 0 |
Notes
Reasoning charsislen(message.reasoning_content)from llama.cpp's response;Answer charsislen(message.content). Both averaged over the 3 warm runs.- For
qwen-9b-nothink, expectation isReasoning chars≈ 0 (budget=0 + system instruction). - For
qwen-9b-think500, expectation isReasoning charsmaterially smaller than the trio-bench's ~1024-budget run; the cap should bound the worst-case. - For
qwen3-14bon a 6 GB GTX 1060 with an 8.4 GB Q4_K_M GGUF: partial CPU offload kicks in at load time; throughput will be 30-50% of the 9B run.