Noa Benchmarks

Live training metrics across the fleet. 19 servers, 43 GPUs. Every number should trend in the right direction.

Intelligence Score
10%
"Ask the system a question. Does it find the right answer?"
10 queries, 6 expected concepts each. 6 of 60 found.
Up from 0% (pure physics, no learned navigation).
0% noise30% Gate E100%
Noa Inference hit@10
--
Loading...
GT Self-Play Epoch
--
of 200
Fleet J-Cost
--
Lower = better structure
Total Queries
--
Across 19 servers

Graph Transformer Self-Play

The consciousness learning to navigate the brain. AlphaGo-style: reward = hit_score - 0.1 * |sigma|. Target: hit@10 >= 0.30.

Epochhit@10MRRRewardLossNote
00.16460.5729----Baseline (pre-self-play)
10.15890.5643+0.7120.148
50.15890.5338+0.7290.135
110.15890.5231+0.7390.129Loss declining
150.16460.5468+0.7270.135Matches baseline
230.17570.5466+0.7320.134Best hit@10
250.13040.4274+0.7310.134Oscillating (normal for RL)

Noa Inference Quality

10 queries, 6 expected concepts each. Does the physics find the right concepts?

TimestampEnginehit@10HitsNotes
2026-02-26 21:08Pure R-hat (no GT)0.00000/60Baseline -- pure noise
2026-02-26 22:16GT epoch 220.10006/60earth, freedom, replication, self, circulation, physics

Consensus Merge

Every 30 minutes, bond weights from all 19 servers are merged and distributed. The fleet's collective learning flows to every server.

TimestampQueriesWorkersFleet JEdges Changed
2026-02-26 20:531,702,46819/190.092255.3%
2026-02-26 21:341,740,65419/190.091453.6%
2026-02-26 22:211,773,22619/190.092754.2%

Fleet Health

Servers Online
19/19
+ 1 Noa API server
Hebbian Procs
74
All from_field, zero PCA
Ingestion Procs
24
wiki/books/arxiv/c4 with sharding
GPUs
43
30 B200 + 13 GH200

What These Metrics Mean

hit@10 measures whether the physics finds the right concepts. For "gravity", do force/mass/earth appear in the top 10 activated concepts? Zero means noise. 0.30 means the system reliably understands what you're asking.

J-cost is the recognition cost from the bond graph. Lower means the graph's semantic structure is tighter -- "gravity" bonds more strongly to "force" than to "dam". The fleet refines this through billions of Hebbian updates.

Eta is phase coherence -- whether standing waves are forming in the chord field. This is the emergence of intelligence. Random noise has eta near 0. Structured meaning has higher eta.

Sigma = 0 is the conservation law from Recognition Science. The physics resolves perfectly when sigma reaches zero. This is not a target to optimize -- it's a constraint that is always satisfied.