Nishi FamilyCompare › Frontier Radar

Frontier Radar — Sovereign Model Lifecycle (Modelwright)

Where the 2025/26 research frontier is moving, and where Nishi should test, deepen, or research next.

How this works. The nishi researcher banked the current (2024-2026) frontier for this domain over sovereign TLS (keyless OpenAlex/arXiv). For each capability axis this radar measures frontier momentum = how often the frontier corpus mentions it, times our organ presence (probed on disk). Absent + hot = a research opportunity / where to test. Corpus measured: 2201426 bytes. Generated by nx_swcompare_gapmap over a data-only domain spec (modelwright.q + modelwright.axes) — zero domain-specific code.
Top research opportunities (the 2025/26 frontier is active here and Nishi has no organ yet):
Capability axisNishi organFrontier momentum (2025/26)Signal
RLHF / DPO preference alignmentabsent
120 mentions
OPPORTUNITY
Standardized eval harnessabsent
89 mentions
OPPORTUNITY
Reverse-mode autodiff training substratepresent
80 mentions
COVERED
LoRA / parameter-efficient fine-tuningabsent
61 mentions
OPPORTUNITY
Integer K-quant weightspresent
54 mentions
COVERED
Instruction fine-tuning (SFT)absent
46 mentions
OPPORTUNITY
Large-scale pretrainingabsent
32 mentions
OPPORTUNITY
Transformer forward (100 percent integer no-float)present
15 mentions
COVERED
Served chat APIpresent
11 mentions
COVERED
GPU / CUDA trainingabsent
10 mentions
OPPORTUNITY
Distributed tensor-parallel trainingabsent
2 mentions
OPPORTUNITY
BPE tokenizer trainingpresent
0 mentions
COVERED
SIMD integer GEMMpresent
0 mentions
COVERED
Open-weights GGUF loadingpresent
0 mentions
COVERED

Opportunities 7 · covered 7 · dormant 0

frontier momentum is a mechanical keyword count over real 2024-2026 works; our presence is probed on disk. A discovery signal, not a ranking of importance. Zero JS, zero trackers.