-
Notifications
You must be signed in to change notification settings - Fork 2.5k
Pull requests: openai/parameter-golf
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Record: Order-Adaptive Entropy Gating + BackoffNgramMixer (val_bpb=0.5466)
#798
opened Mar 26, 2026 by
travispchen
Loading…
Record: 0.6567 BPB — Prefill Cache + 7-Gram Entropy-Adaptive + EBLS
#796
opened Mar 26, 2026 by
Robby955
Loading…
10 tasks done
Record: 11L + order-adaptive 11-gram (mean val_bpb=0.8881)
#795
opened Mar 26, 2026 by
hypery11
Loading…
4 tasks done
Muon Optimizer Tuning: val_bpb 1.3346 by jeremyschied
#794
opened Mar 26, 2026 by
jeremyschied
Loading…
11L LeakyReLU² + XSA-all + Full GPTQ + 5-gram Backoff (1.0340 BPB)
#792
opened Mar 26, 2026 by
xexyz
Loading…
Record: Residual Input Mixing + mixed int6 GPTQ + grouped TTT + MLP 3.5x
#790
opened Mar 26, 2026 by
danialht
Loading…
Add non-record submission: Depth Recurrence + TTT + Gradient Checkpointing
#789
opened Mar 26, 2026 by
ab2891
Loading…
Record: 11L + order-adaptive 9-gram backoff (mean val_bpb=0.9059)
#788
opened Mar 26, 2026 by
hypery11
Loading…
4 tasks done
0.8128 BPB: Classical Compression Eval + N-gram Backoff on PR #549 Base
#786
opened Mar 26, 2026 by
shinegami-2002
•
Draft
3 of 5 tasks
Applied Async Prefetching Boost Performance of Any Approach
#785
opened Mar 26, 2026 by
SirSaltySalmon
•
Draft
Non-record: Depth Recurrence + XSA + LeakyReLU² (val_bpb 1.2065)
#784
opened Mar 25, 2026 by
iverbovoy
Loading…
Non-record: PR703 + shard-order curriculum + GPTQ cache-backout (1.1171)
#783
opened Mar 25, 2026 by
petergpt
Loading…
Non-record WIP: Systematic Eval-Time Prediction Mixing | Requesting compute credits for full 8xH100 evaluation.
#780
opened Mar 25, 2026 by
sargonxg
Loading…
Record: BackoffNgramMixer + Drift-Free TTT (3-seed mean val_bpb=0.6683)
#779
opened Mar 25, 2026 by
deanbrr
Loading…
Record: 11L Full GPTQ + Multi-Order N-gram Backoff (fixed-alpha 0.9757 / entropy-adaptive 0.9605, 3-seed)
#778
opened Mar 25, 2026 by
raahilshah
Loading…
Record: 0.9623 BPB — 7-Gram Entropy Cache + XSA-all + EBLS
#777
opened Mar 25, 2026 by
Robby955
Loading…
8 tasks done
Record Submission: 0.9258 BPB — Kitchen Sink (7-gram + XSA6 + BigramHash4K + Cosine TTT)
#776
opened Mar 25, 2026 by
agalimova
Loading…
3 of 4 tasks
Record: Order-Adaptive Entropy Gating + XSA-All (val_bpb=0.9370)
#774
opened Mar 25, 2026 by
travispchen
Loading…
Add non-record shared-weight Frugendorff submission
#773
opened Mar 25, 2026 by
siddhantparadox
Loading…
Non-record: Data ordering & selection — negative result on FineWeb
#772
opened Mar 25, 2026 by
abaybektursun
Loading…
Record: AdamW TTT 30ep Cosine + Per-Layer LR (val_bpb: 1.0705)
#771
opened Mar 25, 2026 by
sunnypatneedi
Loading…
Previous Next
ProTip!
Follow long discussions with comments:>50.