Skip to content

Pull requests: wandb/senpai

New pull request New
16 Open 1,238 Closed
16 Open 1,238 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Regime L: lr=2e-3, n_hidden=224, mlp_ratio=1, no noise, EMA from ep30 (radical combo) noam Noam advisor branch experiments status:wip Student is working on it student:kohaku Assigned to kohaku
#1254 opened Mar 19, 2026 by tcapelle • Draft
1
Regime K: n_head=8, slice_num=64 (more attention heads + slices) noam Noam advisor branch experiments status:wip Student is working on it student:senku Assigned to senku
#1253 opened Mar 19, 2026 by tcapelle • Draft
1
Regime J: mlp_ratio=4 (wider FFN at same hidden dim) noam Noam advisor branch experiments status:wip Student is working on it student:gilbert Assigned to gilbert
#1252 opened Mar 19, 2026 by tcapelle • Draft
1
Regime I: warmup=5, cosine restart at ep40 (fast start + second phase) noam Noam advisor branch experiments status:wip Student is working on it student:violet Assigned to violet
#1251 opened Mar 19, 2026 by tcapelle • Draft
1
Regime H: slice_num=48, n_hidden=160 (finer routing, narrower) noam Noam advisor branch experiments status:wip Student is working on it student:askeladd Assigned to askeladd
#1250 opened Mar 19, 2026 by tcapelle • Draft
1
Regime G: Remove hard-mining, no noise, surf_weight fixed at 30 (clean training) noam Noam advisor branch experiments status:wip Student is working on it student:thorfinn Assigned to thorfinn
#1249 opened Mar 19, 2026 by tcapelle • Draft
1
Regime F: n_hidden=160, n_layers=2 (trade width for depth) noam Noam advisor branch experiments status:wip Student is working on it student:edward Assigned to edward
#1248 opened Mar 19, 2026 by tcapelle • Draft
1
Regime E: EMA from ep25 + decay=0.997 + T_max=72 (longer EMA window) noam Noam advisor branch experiments status:wip Student is working on it student:alphonse Assigned to alphonse
#1247 opened Mar 19, 2026 by tcapelle • Draft
1
Regime D: batch_size=8, lr=2e-3 (larger batch, adjusted LR) noam Noam advisor branch experiments status:wip Student is working on it student:nezuko Assigned to nezuko
#1246 opened Mar 19, 2026 by tcapelle • Draft
1
Regime C: Remove Lookahead, lr=4e-3 (pure AdamW) noam Noam advisor branch experiments status:wip Student is working on it student:tanjiro Assigned to tanjiro
#1245 opened Mar 19, 2026 by tcapelle • Draft
1
Regime B: lr=5e-3, temp clamp 0.15 at ep40 (aggressive exploration) noam Noam advisor branch experiments status:wip Student is working on it student:fern Assigned to fern
#1244 opened Mar 19, 2026 by tcapelle • Draft
1
Regime A: lr=1.5e-3, T_max=80, warmup=15 (slower deeper convergence) noam Noam advisor branch experiments status:wip Student is working on it student:frieren Assigned to frieren
#1243 opened Mar 19, 2026 by tcapelle • Draft
1
Seed sweep: seed=137 (frontier verification) status:review Ready for advisor review
#1220 opened Mar 19, 2026 by tcapelle Loading…
1
Physics prior residual learning (predict deviation from potential flow Cp) status:review Ready for advisor review
#1208 opened Mar 19, 2026 by tcapelle Loading…
1
Noam lab's work
#789 opened Mar 17, 2026 by tcapelle Loading…
1
Skillz based flow
#457 opened Mar 16, 2026 by tcapelle Loading…
ProTip! Adding no:label will show everything without a label.