Lars Bogner lars
0 Followers · 0 Following
  • Karlsruhe, Germany
  • Joined on 2025-04-20
Repositories
25
 Projects Packages Public Activity Starred Repositories
lars merged pull request lars/giant#26 2026-07-30 17:05:09 +02:00
Add opt-in straight-through Gumbel-softmax combine weights to MoE router
lars pushed to feat/router-gumbel-softmax at lars/giant 2026-07-30 16:46:24 +02:00
8065df896e Scope wandb run config to only-active hyperparameters
lars pushed to feat/router-gumbel-softmax at lars/giant 2026-07-30 16:43:01 +02:00
5eec4c250a Add gumbel/learn_centers/learn_width/learn_temperature to out-dir naming
lars pushed to feat/router-gumbel-softmax at lars/giant 2026-07-30 16:40:09 +02:00
af2ee7c7ce Add gumbel router configs sweeping learnable-knob combinations
lars created pull request lars/giant#26 2026-07-30 16:30:49 +02:00
Add opt-in straight-through Gumbel-softmax combine weights to MoE router
lars pushed to feat/router-gumbel-softmax at lars/giant 2026-07-30 16:29:24 +02:00
b51eafcfa5 Add opt-in straight-through Gumbel-softmax combine weights to MoE router
lars created branch feat/router-gumbel-softmax in lars/giant 2026-07-30 16:29:24 +02:00
lars pushed to master at lars/giant 2026-07-30 16:18:37 +02:00
43cb6dd9ae Merge pull request 'Add learnable per-expert width and shared temperature to EnergyRouter' (#25) from feat/router-learnable-width into master
aa55c407ab Merge branch 'master' into feat/router-learnable-width
da5f54ea1c Add learnable per-expert width and shared temperature to EnergyRouter
Compare 3 commits »
lars merged pull request lars/giant#25 2026-07-30 16:18:35 +02:00
Add learnable per-expert width and shared temperature to EnergyRouter
lars pushed to feat/router-learnable-width at lars/giant 2026-07-30 15:58:57 +02:00
aa55c407ab Merge branch 'master' into feat/router-learnable-width
78297456e3 Merge pull request 'Store a quantile grid instead of a raw reservoir sample in the setup cache' (#24) from perf/quantile-energy-reservoir into master
Compare 2 commits »
lars created pull request lars/giant#25 2026-07-30 15:35:55 +02:00
Add learnable per-expert width and shared temperature to EnergyRouter
lars pushed to feat/router-learnable-width at lars/giant 2026-07-30 15:33:57 +02:00
da5f54ea1c Add learnable per-expert width and shared temperature to EnergyRouter
lars created branch feat/router-learnable-width in lars/giant 2026-07-30 15:33:57 +02:00
lars deleted branch perf/quantile-energy-reservoir from lars/giant 2026-07-30 13:59:20 +02:00
lars pushed to master at lars/giant 2026-07-30 13:59:20 +02:00
78297456e3 Merge pull request 'Store a quantile grid instead of a raw reservoir sample in the setup cache' (#24) from perf/quantile-energy-reservoir into master
d656cf3109 Store a quantile grid instead of a raw reservoir sample in the setup cache
Compare 2 commits »
lars merged pull request lars/giant#24 2026-07-30 13:59:18 +02:00
Store a quantile grid instead of a raw reservoir sample in the setup cache
lars created pull request lars/giant#24 2026-07-30 13:48:14 +02:00
Store a quantile grid instead of a raw reservoir sample in the setup cache
lars pushed to perf/quantile-energy-reservoir at lars/giant 2026-07-30 13:32:50 +02:00
d656cf3109 Store a quantile grid instead of a raw reservoir sample in the setup cache
lars created branch perf/quantile-energy-reservoir in lars/giant 2026-07-30 13:32:50 +02:00
lars pushed to master at lars/giant 2026-07-30 12:56:26 +02:00
de5db25e3f Add giant new-run to scaffold a config.toml + run dir ahead of training