-
Notifications
You must be signed in to change notification settings - Fork 3.4k
Pull requests: openai/parameter-golf
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Record: SP8192 + NEFTune + Z-Loss + Phased-TTT (4 phases, prefix=3000, LoRA-128) — val_bpb 1.06035 (3-seed mean)
#2163
opened May 9, 2026 by
uniagent-alpha
Loading…
5 tasks done
Swiglu gating_ QAT_Residual Attention Scalin _EMA- Sliding window_Optimizations for 10min/16MB Track
#2159
opened May 6, 2026 by
visin109
Loading…
feat(non_record): add SP8192 BPE Mamba3 SSM hybrid 16MB non-record submission
#2155
opened May 4, 2026 by
divagr18
Loading…
Non-record: SP8192 + RandProj384 tied embeddings + Pairwise-QK Muon -- Single-seed negative result
#2149
opened May 3, 2026 by
YaseenHQ
Loading…
[Non-record] MHALM V2 non-record submission (1.3477 bpb)
#2145
opened May 2, 2026 by
aquemy
Loading…
Non record: Progressive context growth precursor to PR 2014, 12 hours on RTX 4090, val_bpb 0.9697 pre-quant
#2144
opened May 2, 2026 by
simonbissonnette
Loading…
Non-record submission: post-deadline CaseOps + SparseAttnGate + Phased TTT (1.07134 BPB)
#2143
opened May 2, 2026 by
upascal
Loading…
4 tasks done
records(non-record-16mb): JEPA-on-LM 14-run ablation (negative result)
#2142
opened May 2, 2026 by
eren23
Loading…
3 tasks
Corrected: PR #2014 stack + LeakyReLU 0.3 + token-only in-timer n-gram TTT (val_bpb 1.0570)
#2140
opened May 1, 2026 by
simon-marcus
Loading…
Record: SP8192 + Sliding-Window Eval + Lock-In Byte Mixer - val_bpb 1.067219
#2138
opened May 1, 2026 by
anmarhindi
Loading…
Non-record: notes on the recurrence band (mixing parameters, MLP sizing, loop sizing)
#2137
opened May 1, 2026 by
leon2k2k2k
Loading…
Record candidate: PR #2130 base + GPTQ_CALIBRATION_BATCHES=32 — val_bpb 1.05651 (3-seed mean)
#2135
opened May 1, 2026 by
codemath3000
Contributor
Loading…
Record candidate: 1.05670 BPB — token-only n-gram tilt + AsymLogit + #2060 levers + NUM_PHASES=1
#2130
opened May 1, 2026 by
TanishGudise
Loading…
Non-record: Confidence-Adaptive N-gram Boost on PR #2018 stack, val_bpb=1.05874
#2129
opened May 1, 2026 by
okezue
Loading…
Non-record: Post-Quantization LoRA Distillation (LCQ) on PR #1855 stack, val_bpb=1.06767
#2128
opened May 1, 2026 by
okezue
Loading…
Non-record: Redoing ZerO initialization + Follow-up to PR 2104
#2126
opened May 1, 2026 by
AlstonTang
Loading…
Record : CaseOps Gated XSA NgramTilt LQER | val_bpb=1.05933439
#2124
opened May 1, 2026 by
vaibhavmishra1
Loading…
Previous Next
ProTip!
Exclude everything labeled
bug with -label:bug.