throsturx
/

bihmoe-poc

mixture-of-experts

interpretability

Model card Files Files and versions

104 kB

Ctrl+K

Ctrl+K

1 contributor

History: 33 commits

Throstur

meta: add contributing/roadmap/citation/architecture notes

61cc313 5 months ago

configs
docs: add preliminary results + reproducible 4k panel config/scripts 5 months ago
docs
meta: add contributing/roadmap/citation/architecture notes 5 months ago
runs
docs: freeze PoC skeleton (arch/tasks/compute/protocol) + artifact policy 5 months ago
scripts
docs: add preliminary results + reproducible 4k panel config/scripts 5 months ago
src
arch: add chiasm_mode=bias to avoid global-branch starvation 5 months ago
.gitignore

308 Bytes
head: make models query-aware (last-token readout) + trainer update 5 months ago
CITATION.cff

346 Bytes
meta: add contributing/roadmap/citation/architecture notes 5 months ago
CONTRIBUTING.md

1.02 kB
meta: add contributing/roadmap/citation/architecture notes 5 months ago
README.md

1.03 kB
hf: add model card metadata (license/tags) 5 months ago
ROADMAP.md

725 Bytes
meta: add contributing/roadmap/citation/architecture notes 5 months ago
pyproject.toml

418 Bytes
bag: bihmoe-bag_20260222-030013 (snapshot) 5 months ago
requirements-core.txt

30 Bytes
probe: add minimal dense + structured models and VRAM probe script 5 months ago
uv.lock

18.7 kB
bag: bihmoe-bag_20260222-030013 (snapshot) 5 months ago