id stringlengths 36 36 | tier stringclasses 1
value | github_login stringclasses 179
values | created_at stringlengths 32 32 | status stringclasses 2
values | sha256 stringlengths 64 64 | source_bytes int64 341 260k | leaderboard_rank int64 | source large_stringlengths 341 260k | metadata_json large_stringlengths 645 734 | result_json large_stringlengths 5 4.8k |
|---|---|---|---|---|---|---|---|---|---|---|
4ebd11c0-28a4-4d49-a931-bf57a76f6469 | easy | erdavis0 | 2026-08-25 01:23:24.083740+00:00 | succeeded | 441829d198f853ac2efa0065a0a953215d79f7cb90ee2b49280f23481a092acf | 24,085 | null | """Matched categorical-boundary panel: C1 continuous endpoint auxiliary.
Public delimiter syntax is used only to route opaque token identities onto
separate right-aligned tapes. One T-blind learned macro maps a categorical X
tape and a digitwise N tape to a new categorical tape through freshly reset
interaction, carr... | {
"id": "4ebd11c0-28a4-4d49-a931-bf57a76f6469",
"created_at": "2026-08-25 01:23:24.083740+00:00",
"db_md5": "18bcf63268fc44cb8c6a3e621ae926dd",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "2326b61d-c842-4a16-9af1-b94dc60c3f42",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.166034840425713,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000037252903
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1715543676384894,
"example_count": 600... |
4ebfff8b-cbe6-48f3-923d-0ee4b91571e0 | easy | Laz4rz | 2026-08-08 00:43:23.621903+00:00 | succeeded | 0264a1019c5981c58b05bb405982b9a5fc0c5f22e4f79192bba565104fe7bf50 | 27,691 | null | """PDiff DiT + masked GIDD, adapted to the One Layer Deeper API.
The model, constant-pi objective, and adaptive sampler follow the current
PDiff working tree's ``gidd_brevo1_paper_aligned_nplus2`` path. The benchmark
adapter keeps the prompt and completion canvas separate internally, then
converts sampled answer toke... | {
"id": "4ebfff8b-cbe6-48f3-923d-0ee4b91571e0",
"created_at": "2026-08-08 00:43:23.621903+00:00",
"db_md5": "de76380160fb8b62f20fcfe661a8b98a",
"submitter": "Laz4rz",
"github_login": "Laz4rz",
"run_id": "b6a007bc-8228-4c5d-a5ae-80aa39d4454d",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.8091925135272304,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013958333569268386
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.809038609716258,
"example_count": 300... |
4eca0f94-01d9-45c4-b69f-6375be274873 | easy | jordanrubin | 2026-08-14 01:40:14.200995+00:00 | succeeded | 69ece06bd94846b5a380dc94d4c00c6798b85f8c1e9f0ba91ba31bcdb9376a77 | 29,580 | null | """Weight-tied MLP-cell loop over soft digit states (mlploop).
Cell = MLP over [expected digits, pairwise digit products, RNS residue
simplices (fixed differentiable mixing over Z_p), N digits]. The cell that
learns one-step modular squaring from direct pairs (52% unseen at 14k rows,
day-20 screen) inside the exact-T ... | {
"id": "4eca0f94-01d9-45c4-b69f-6375be274873",
"created_at": "2026-08-14 01:40:14.200995+00:00",
"db_md5": "0cdede812da1d9957f86c9ea89ed1f0f",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "bb0bfbab-65a2-4834-8e9d-e6118b66dd34",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.5613181669395515,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08666666542800763
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1578400211216624,
"example_count": 600... |
4ecd3fa9-13a0-4187-81d0-ab5ab908b01e | easy | sapient-sapiens | 2026-08-19 10:59:20.121930+00:00 | succeeded | 9560ac8130059082a60b88f55aebdad778fa1994b9d024888cb7fe7258b3d226 | 31,889 | null | """Depth-quantized recurrence with a mixture-over-depth objective.
The recurrent state is a bank of right-aligned digit slots. Every operator
application is followed by a soft quantization back onto the token simplex, and
the digit logits produced by that quantization are the answer logits at that
depth. There is no s... | {
"id": "4ecd3fa9-13a0-4187-81d0-ab5ab908b01e",
"created_at": "2026-08-19 10:59:20.121930+00:00",
"db_md5": "2c67e44716a742321c8bfe837e7f35c7",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "6f086a14-d528-4e03-9282-1401ecdb3e62",
"tier": "easy",
"dataset_id": "e9",
"status": "su... | {
"score": {
"mean_loss": 4.508772373199463,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03222222253680229
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.341947555541992,
"example_count": 90,
... |
4ecd6150-8291-48bd-884d-a2015ea270b9 | easy | shirvani-jr | 2026-08-29 00:48:26.767480+00:00 | succeeded | 436d289f73f7711eb5f4f83e247a24c8ee80a5dfcded11d6c89cbbeb4d25f660 | 50,044 | null | """Adaptive learned discrete recurrent fabric for One Layer Deeper.
A gradient-trained, operation-free recurrent computational machine. The
model learns, from the competition's endpoint supervision: how to parse the
prompt into a bank of categorical digit registers, a reusable transition
circuit applied serially, a l... | {
"id": "4ecd6150-8291-48bd-884d-a2015ea270b9",
"created_at": "2026-08-29 00:48:26.767480+00:00",
"db_md5": "ea7cfed6998d1a1271c3a4e1a38bf159",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "4d4cb5d0-2e09-48c8-a91c-c1a067ccde07",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.2597556114196777,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00874999980442226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2554163932800293,
"example_count": 800... |
4edc6cf8-fbc7-409b-8711-0a81e64c0fad | easy | DDanlov | 2026-08-16 19:53:45.488936+00:00 | failed | ba3b4d1122f02497a131c4eb962ce4ae0c646c7f18eafcea30be16d426d1a46a | 16,028 | null | """
Production Model E1 Breakthrough Submission (Hardened v2)
=========================================================
Architecture:
- 64-bit Vectorized Place-Value Prompt Parser
- Multi-Scale Dyadic & Coprime Fourier Torus Arithmetic Coupler (CRT for N=323: 17 x 19)
- Recurrent Bilinear Multiplicative Squaring Unit (... | {
"id": "4edc6cf8-fbc7-409b-8711-0a81e64c0fad",
"created_at": "2026-08-16 19:53:45.488936+00:00",
"db_md5": "c6b5b696e776ec593e7c29b122960a9d",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "4288de52-fd85-4a57-95b7-e221493a92c0",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
4ee34141-f50f-41e6-b4b9-f65a12da7eb3 | easy | yashkant | 2026-08-16 00:42:13.401866+00:00 | succeeded | 2584011f05bbc145175fe808a5f2c287405d2780b0931910b585d1d19e4990a0 | 34,958 | null | """Round1776 anisotropy-only centered diagonal-metric Dykstra.
The proven Round1766 W8/D2 recurrent carrier is retained. Before each named
set projection, a shared network maps the immutable input anchor and that set's
code to a positive diagonal metric. Its bounded log-diagonal is mean-centered
so the metric has un... | {
"id": "4ee34141-f50f-41e6-b4b9-f65a12da7eb3",
"created_at": "2026-08-16 00:42:13.401866+00:00",
"db_md5": "3c1f0b2a1ae90d8ea343a18948370c09",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "dc399fe1-8e56-4f3e-a59a-35909ee40187",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 3.0841974020004272,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07625000551342964
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.6939196586608887,
"example_count": 300... |
4ee6b3a1-ffdd-4b56-9dd7-647e22c60e50 | easy | chad-atexpedient | 2026-08-05 04:20:31.281200+00:00 | succeeded | 3bcaebaa97a7c9426b8a015f544a235d3dae387ad3a49fc71e90913d42d7401c | 11,222 | null | """R029MH: soft behavior routing with hard retain/write choice."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
LANES = 6
CHANNELS... | {
"id": "4ee6b3a1-ffdd-4b56-9dd7-647e22c60e50",
"created_at": "2026-08-05 04:20:31.281200+00:00",
"db_md5": "3159cb859b18c465f753a8a31251f320",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "ffcebc6d-e90a-4da9-89f2-cc657087daca",
"tier": "easy",
"dataset_id": "e3",
"st... | {
"score": {
"mean_loss": 2.1767986460216378,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.016250000121071934
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1820119716161233,
"example_count": 80... |
4ef89339-ea4b-41c9-a710-83576bf4b9c1 | easy | oupadhyay | 2026-08-12 06:28:50.267106+00:00 | succeeded | 255b16e4d111fa357b7de34447718c5e2be6fce52787933e8f2f0c5777137b91 | 7,888 | null | """One-block RMS-normalized spatial-gating register for the Easy tier."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
PAD, N_MARK, X_MARK, T_MARK, ... | {
"id": "4ef89339-ea4b-41c9-a710-83576bf4b9c1",
"created_at": "2026-08-12 06:28:50.267106+00:00",
"db_md5": "e777d62b4feadef11d81165d6c539cf6",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "7c8f9b9e-5e88-4bbf-bdc9-dd4d636eef44",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.2544739129268505,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0054166666790843015
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.16115422419903,
"example_count": 600... |
4efce47b-5295-4c2d-99b9-87dd715a514a | easy | Maxwells-Demons | 2026-08-06 10:02:35.080929+00:00 | succeeded | dee439a29fd8d1c21097517dcc8d0ced413eae08af5707f9cfa0562f2b20bf9f | 12,594 | null | """v19: hash-routed per-modulus value tables.
Every modulus appearing in e3/e4/e5 train also appears at eval time, so the
transition can be a bank of full V x V lookup tables routed by a fixed hash
of N (a mixture-of-experts with deterministic routing; nothing about the task
arithmetic is hardcoded). The recurrent st... | {
"id": "4efce47b-5295-4c2d-99b9-87dd715a514a",
"created_at": "2026-08-06 10:02:35.080929+00:00",
"db_md5": "15aba5fe45cf919e92b3708d0c9d2387",
"submitter": "Maxwells_Ayakashi",
"github_login": "Maxwells-Demons",
"run_id": "faf6bc91-d52b-4380-9706-676229c17dbc",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 7.736619472503662,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02166666742414236
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.008681297302246,
"example_count": 600,
... |
4efdbd72-4ffe-4727-9596-70f92789e914 | easy | DDanlov | 2026-08-08 10:22:53.476360+00:00 | succeeded | 750332aeb8b7a1ebb426a0b0c8fb24c8e179da1fa831d92b1218df84c5bead13 | 15,561 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class RohanShampoo(Optimizer):
"""Self-contained RohanShampoo optimizer with eigenvalue ma... | {
"id": "4efdbd72-4ffe-4727-9596-70f92789e914",
"created_at": "2026-08-08 10:22:53.476360+00:00",
"db_md5": "79da4a79d93d548dd005ba5d258f04e6",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "ccad59ff-ee0a-4793-b315-1f2b827641e6",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 206.12353884940222,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 222.51751097435374,
"example_count": 6... |
4f00ad9d-b1ec-442c-a329-03947a713c9c | easy | MarcoF1 | 2026-08-13 20:24:18.981228+00:00 | succeeded | 28e62b30eccadb618d7693bcaf95ad175e4f85101e611f8b4d3cdff6220fafa8 | 17,271 | null | """Population-of-clocks: K independent residue-clock models trained in
parallel as batched weight tensors, ensembled at evaluation.
Rationale: hosted training is kernel-launch-bound, not FLOP-bound, so running
K=8 models costs almost no wall-clock. Each member gets its own init and its
own table-dropout probability (s... | {
"id": "4f00ad9d-b1ec-442c-a329-03947a713c9c",
"created_at": "2026-08-13 20:24:18.981228+00:00",
"db_md5": "6557b63c1fa3647adcaef172894411da",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "5cadb010-4f86-405c-a0a0-57c1d75c7ee2",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.8919963248055476,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07499999836087226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9963338375091553,
"example_count": 100... |
4f055149-9b61-4cba-8803-828a8e824844 | easy | chad-atexpedient | 2026-08-04 21:45:00.835602+00:00 | succeeded | b6b0a57109f09e8f32900bcc1500b2d85ee9938c1c105ab56c6f3389f9db41a2 | 4,423 | null | """R015: five tied Transformer passes with rotary position encoding."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
WIDTH = 128
H... | {
"id": "4f055149-9b61-4cba-8803-828a8e824844",
"created_at": "2026-08-04 21:45:00.835602+00:00",
"db_md5": "63f007b1be010592ad7b519b376d0ca5",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "00706c74-7c2a-4ccc-9e69-a391bf641e5f",
"tier": "easy",
"dataset_id": "e3",
"st... | {
"score": {
"mean_loss": 7.9844466501662605,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000018626451
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.120350566088877,
"example_count": 800... |
4f069ae0-7466-4287-b732-1b3f0ceb5eff | easy | AFeinfield | 2026-08-25 00:01:20.531874+00:00 | succeeded | b744d91db4bfeeb3abee63150ce0e427c83d5d29abc4b0c355618d3c622de853 | 17,427 | null | """108: ABACUS digit-place embeddings on the 094 instrumented base.
ONE change vs 094_sub: every digit token gets a learned embedding indexed
by its PLACE VALUE WITHIN ITS OWN NUMBER FIELD (distance from that field's
least-significant digit; units = 0), plus a small per-field embedding
(N / x / T). During training a ... | {
"id": "4f069ae0-7466-4287-b732-1b3f0ceb5eff",
"created_at": "2026-08-25 00:01:20.531874+00:00",
"db_md5": "adaa4697b4ee32795963a6a39e678967",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "494b7e9f-f797-4a13-8bdb-c144a1683e98",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1494750015734687,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333333457509677
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.158694698137018,
"example_count": 60... |
4f087101-ab66-43bc-a93f-b2916fff2ad4 | easy | velocizapkar | 2026-08-13 09:41:38.991028+00:00 | succeeded | 28101652d70a0efca5dea93e051dc1d9b6ca798f7e86f6f81eb0fc607b056577 | 17,950 | null | """Faithful completely skipless Transformer with the official SOAP optimizer.
Every one of eight untied blocks is the strict composition
``x = attention(norm(x)); x = mlp(norm(x))``. There is no additive state
path, attention residual, recurrence, input reinjection, depth bank, or raw
embedding decoder. Attention us... | {
"id": "4f087101-ab66-43bc-a93f-b2916fff2ad4",
"created_at": "2026-08-13 09:41:38.991028+00:00",
"db_md5": "f156d43a65ce1ae9b3bad19551231794",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "adcda13d-23f2-4aa2-ac11-4e1c305eae2c",
"tier": "easy",
"dataset_id": "e2",
"statu... | {
"score": {
"mean_loss": 6.923438549041748,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000558793545
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.043040752410889,
"example_count": 300,... |
4f0ad9c6-6133-4788-86a2-3afeca096a44 | easy | DDanlov | 2026-08-16 23:06:14.788766+00:00 | succeeded | a8186b80178fb17882da6262a429e255de0274830ab319d2c042024905b83381 | 16,064 | null | """
Dual-Origin Symplectic Lie Deliberation Net (DOS-DeliberationNet)
=============================================================================
Architecture & Breakthrough Dynamics:
1. Single Unit RMSNorm strictly on input token embedding boundary.
2. Dual-Origin Dynamic RoPE:
- First half of head dimensions: Fo... | {
"id": "4f0ad9c6-6133-4788-86a2-3afeca096a44",
"created_at": "2026-08-16 23:06:14.788766+00:00",
"db_md5": "c3fe128c6996cdb11cba0bfa9032eb49",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "2e3bb40b-2dc7-4ab4-9042-c6296fd0358f",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.4204069708852445,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0104166666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.471791985590896,
"example_count": 600,
... |
4f1d79c6-f783-4b6d-bb8a-1d4ebe404332 | easy | sapient-sapiens | 2026-08-17 19:11:10.389346+00:00 | succeeded | 1600b89424131b808a122dfc02d4a48d86c59b074b9a93e3e40cba04469ad734 | 28,654 | null | """v40 lineage: T-free operator, N memory, PonderNet mix, no T in the loss.
Field-local encoder, T-free core, countable halt, answer-slot readout.
N is seeded and restored every step. T is never parsed in forward or loss.
"""
from __future__ import annotations
import math
import time
import torch
import torch.nn.fu... | {
"id": "4f1d79c6-f783-4b6d-bb8a-1d4ebe404332",
"created_at": "2026-08-17 19:11:10.389346+00:00",
"db_md5": "4d0925bb07bf157790930053f7c39979",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "bb34cefe-dd5b-42b2-b8e1-0dd10c4852a9",
"tier": "easy",
"dataset_id": "e9",
"status": "su... | {
"score": {
"mean_loss": 2.0245218730393226,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.017777778146167596
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0502331256866455,
"example_count": 90... |
4f21b295-729f-4554-a2dc-260fc50e1df5 | easy | liam-gb | 2026-08-13 02:29:19.986541+00:00 | succeeded | 7bf716e950d5d859ca853cba6f8b5faa423a995e823863cab505a20f4783d9e3 | 16,173 | null | """rns_pm_gather_ct_widet_coxlrt: scale channel + tables at lr 3e-2.
Base: widet plus a global scale/rank channel.
The wide readout is what makes long answers reachable; the trim is what pays
for it, by not running loop iterations whose update is multiplied by zero.
Base docstring:
Single change vs rns_pm_gather_ct... | {
"id": "4f21b295-729f-4554-a2dc-260fc50e1df5",
"created_at": "2026-08-13 02:29:19.986541+00:00",
"db_md5": "37f1f874e55e0d818bb16846c93a8ea4",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "83db13de-91c9-48c7-a96b-542d483fe494",
"tier": "easy",
"dataset_id": "e4",
"status": "succeeded"... | {
"score": {
"mean_loss": 5.063740197507409,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.025208333631356557
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.351014237529791,
"example_count": 1200... |
4f23ab9f-cddb-4bb2-8b9d-994ffe0430ff | easy | alirezashirvani-jr | 2026-08-30 00:11:51.879704+00:00 | succeeded | 672293a76323da62e296a3ee3f10c6ca38ac913874438ad7921545a2384944f0 | 31,682 | null | from __future__ import annotations
import math
import time
from contextlib import nullcontext
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
MODEL_... | {
"id": "4f23ab9f-cddb-4bb2-8b9d-994ffe0430ff",
"created_at": "2026-08-30 00:11:51.879704+00:00",
"db_md5": "310e8a85d3554c9bd27bdd0854d43589",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "d6c63f4e-49e4-4f1d-8ca2-f648fbb50547",
"tier": "easy",
"dataset_id": "e2",
... | {
"score": {
"mean_loss": 1.178758829832077,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.7443750500679016
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.8190068602561951,
"example_count": 300,
... |
4f2fb30b-3e97-46cb-afcd-fa85bf209bdf | easy | erdavis0 | 2026-08-27 01:19:01.983835+00:00 | succeeded | dea5a50d1a7e1542b606c505bb2e064069826b4fcc486a84827b8cbefc9b781f | 31,293 | null | """Typed-prefix reversible transition B0: ordinary evaluator token CE.
Public delimiter syntax routes opaque N, X, and T fields into three strict
functional roles. X initializes only an evolving auxiliary tape, N initializes
only an invariant environment lane, and T reaches only a learned monotone
prefix controller. ... | {
"id": "4f2fb30b-3e97-46cb-afcd-fa85bf209bdf",
"created_at": "2026-08-27 01:19:01.983835+00:00",
"db_md5": "9a82ec9520ce6d08ad83d5dbf700b0ab",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "1702c951-f3ae-4202-afe0-dc3bac850dea",
"tier": "easy",
"dataset_id": "e10",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.750487804412842,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.20567677542567253
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.3001008033752441,
"example_count": 125,... |
4f31a110-1183-47ae-b931-899db13e5697 | easy | chuk-uzowihe | 2026-08-12 19:35:19.597479+00:00 | succeeded | 529e7637b9ada56209afb7e8675f08e866f1c78466dcd3c0ba425f744b5f91db | 26,029 | null | """v10: dense fixed-loop recurrence with deep supervision.
Lineage: FPRM structure (alpha/beta-tied residual scaling, pre-rms_norm,
input injection every iteration) + the v8 2D digit lattice / RoPE task
encoding. Three deliberate departures from the v9 MoEUT line:
1. **Full BPTT.** The loop runs a fixed N_ITER iterat... | {
"id": "4f31a110-1183-47ae-b931-899db13e5697",
"created_at": "2026-08-12 19:35:19.597479+00:00",
"db_md5": "193f3353e7b5d52b8e8330e333137654",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "add65b7d-074b-4347-8ffc-562ba63089e2",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 14.432089805603027,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333334140479565
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 11.996133804321289,
"example_count": 1... |
4f343c74-9ab7-4e2f-be60-1037c8206817 | easy | Laz4rz | 2026-08-08 00:45:51.824520+00:00 | succeeded | 0264a1019c5981c58b05bb405982b9a5fc0c5f22e4f79192bba565104fe7bf50 | 27,691 | null | """PDiff DiT + masked GIDD, adapted to the One Layer Deeper API.
The model, constant-pi objective, and adaptive sampler follow the current
PDiff working tree's ``gidd_brevo1_paper_aligned_nplus2`` path. The benchmark
adapter keeps the prompt and completion canvas separate internally, then
converts sampled answer toke... | {
"id": "4f343c74-9ab7-4e2f-be60-1037c8206817",
"created_at": "2026-08-08 00:45:51.824520+00:00",
"db_md5": "de76380160fb8b62f20fcfe661a8b98a",
"submitter": "Laz4rz",
"github_login": "Laz4rz",
"run_id": "abb3d4a0-96f7-4481-99bf-ecb8b8a7f923",
"tier": "easy",
"dataset_id": "e4",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.8082885947753065,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.819422365007082,
"example_count": 120... |
4f344bde-f444-476d-974d-9390aa88cb88 | easy | khushidahi | 2026-08-21 20:17:52.511339+00:00 | succeeded | 8f95b25810393591541fc0e26ae9a388a7b851e49e1c4151b7edbcf9850b128e | 29,829 | null | """R354: two-layer-by-four-recurrence phase-capacity control.
This submission tests the strongest rules-safe generic baseline suggested by
the public competition discussion. It does not implement squaring, modular
reduction, or any other task solver. Numeric prompt fields are only rearranged
onto equal-length, right... | {
"id": "4f344bde-f444-476d-974d-9390aa88cb88",
"created_at": "2026-08-21 20:17:52.511339+00:00",
"db_md5": "a3bbfe20d46b4b4fb484343c8e6cbe4e",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "1f6bf68f-7b2d-431e-83b9-446f2de081d1",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.2763220700026174,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0025
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2765391632045864,
"example_count": 600,
"... |
4f36deb6-9377-4594-8adb-bd0ab53c7bec | easy | chuk-uzowihe | 2026-08-12 22:05:58.985372+00:00 | succeeded | d70f11e4bb72af5e4bd7ea67cde16a10ff9f20f7406c80cda023b0abbbd51083 | 27,102 | null | """v10: dense fixed-loop recurrence with deep supervision.
Lineage: FPRM structure (alpha/beta-tied residual scaling, pre-rms_norm,
input injection every iteration) + the v8 2D digit lattice / RoPE task
encoding. Three deliberate departures from the v9 MoEUT line:
1. **Full BPTT.** The loop runs a fixed N_ITER iterat... | {
"id": "4f36deb6-9377-4594-8adb-bd0ab53c7bec",
"created_at": "2026-08-12 22:05:58.985372+00:00",
"db_md5": "d7eab46e703816c31a19cb2954ec1c93",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "31ce78f4-c307-47c4-ad6d-227cc9d04c20",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 35.12791347503662,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 31.71270179748535,
"example_count": 100,
"exact... |
4f48c4bb-5c4f-4737-bcbf-9929b6d1788e | easy | viridale | 2026-08-28 13:32:58.183608+00:00 | succeeded | e8748267744a1e2dff48bfddd20aa7ff70117c88c4519fef5bc4645b5847c8d1 | 30,447 | null | """Wide-slab plus all-medium-cube MDL quadratic program learner.
hypothesis: v983 discovers x^2+117 without T1 and certifies four official identities,
but its union of three slabs allows only one coefficient outside +/-4. Add a separate
all-medium cube with a,b,c each in [-64,64]. This expands the plausible ti... | {
"id": "4f48c4bb-5c4f-4737-bcbf-9929b6d1788e",
"created_at": "2026-08-28 13:32:58.183608+00:00",
"db_md5": "96c30621f6eabaf2c390014f91682fcf",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "89175c21-fe96-49f1-9c2f-05f27de912db",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 600,
"exact_accuracy": 1.0,
"... |
4f49aed6-0fdd-4a4a-af05-81d20469e20a | easy | karanganesan | 2026-08-04 01:04:14.441709+00:00 | succeeded | 87dbb54164a858331d9fc36725268046b5062ed6cab6e33c2c8aefabce420636 | 11,992 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "4f49aed6-0fdd-4a4a-af05-81d20469e20a",
"created_at": "2026-08-04 01:04:14.441709+00:00",
"db_md5": "8d09f35a367cfeca03e1e2b6d6ff83de",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "7756cb42-0749-4ed7-bbe2-da28583da3b8",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.3829439878463745,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333334140479565
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.461333751678467,
"example_count": 10... |
4f5b0049-4af5-430a-9e9b-ace0a9a99f55 | easy | shreyash-chonkie | 2026-09-01 01:56:50.940227+00:00 | succeeded | 61682917202cb8a7eaabf75df478db8367cb5c1545705cd9e1db35ab25786442 | 16,165 | null | """Learned token scanner feeding a random recurrent path reservoir."""
from __future__ import annotations
import torch
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
NUM_OPERANDS = 4
NUM_OPERA... | {
"id": "4f5b0049-4af5-430a-9e9b-ace0a9a99f55",
"created_at": "2026-09-01 01:56:50.940227+00:00",
"db_md5": "5453b8b7ace53b39e086007531748ef5",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "abbadde0-8092-4332-86d1-6f444d61294f",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 0.41426442563533783,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.37333332747220993
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.6007943153381348,
"example_count": 10... |
4f6389e1-728a-4f6a-b6c6-d5fe94c9a75d | easy | DDanlov | 2026-08-11 23:54:11.111103+00:00 | failed | 050a0e75bc777a73f564fe6cedeadd11c590094f8a42a0f6f039196196c3269c | 21,886 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
def init_cutting_the_skip(module):
"""SVD-based orthogonal initialization to ensure initia... | {
"id": "4f6389e1-728a-4f6a-b6c6-d5fe94c9a75d",
"created_at": "2026-08-11 23:54:11.111103+00:00",
"db_md5": "48b24ebc33c6818c7c9462fe085b364f",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "8f4e3e7e-7d49-420f-bc2a-837366173231",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
4f67774a-4b0f-4ddf-acda-6d3d248a30ec | easy | erdavis0 | 2026-08-27 01:35:50.479956+00:00 | succeeded | f0f048421bab235b2c1c7f2958a7cff30fbcaa7cef983b31931d270fd5484080 | 31,301 | null | """Typed-prefix reversible transition B2: canonical sequence mixture.
Public delimiter syntax routes opaque N, X, and T fields into three strict
functional roles. X initializes only an evolving auxiliary tape, N initializes
only an invariant environment lane, and T reaches only a learned monotone
prefix controller. ... | {
"id": "4f67774a-4b0f-4ddf-acda-6d3d248a30ec",
"created_at": "2026-08-27 01:35:50.479956+00:00",
"db_md5": "84a6cfec2208ae11b2788eb24985c9f5",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "c9279bcb-ea15-4937-ab2f-388276d36f5a",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",... | {
"score": {
"mean_loss": 6.110305290189537,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011250000060535966
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.191762224485456,
"example_count": 800,... |
4f6c1eb6-b18b-4a4f-a2bd-bbaf2b13ea67 | easy | DDanlov | 2026-08-15 11:58:11.972662+00:00 | failed | 87bea72c763cd7a44ab486e43cdadcb496540b669199a85e1becc0ea0fa0d9eb | 16,676 | null | """
Official Submission for One Layer Deeper Challenge
Architecture: High-Resolution Dyadic Torus Embedding + Skipless DKS Isometry + Exact Per-Example T Unrolling + RohanShampoo
"""
from __future__ import annotations
import math
from typing import Optional, Tuple, Union
import torch
import torch.nn as nn
import torch... | {
"id": "4f6c1eb6-b18b-4a4f-a2bd-bbaf2b13ea67",
"created_at": "2026-08-15 11:58:11.972662+00:00",
"db_md5": "7968e1b2a274cc674d524414d513fe5d",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "4d58bb4f-7e57-4746-9d50-be86be170de5",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
4f6c8496-b10f-4b69-9f4c-6edd9c3f52d0 | easy | 0Chris5R | 2026-08-07 10:38:29.881014+00:00 | succeeded | 233fb04fedb9b1d31580bec2e8076acdec963a22a46fa30e0b58a7dfc156eb8b | 14,822 | null | """Learned convolutional-GRU over an aligned field-by-digit workspace.
The public field markers and decimal significance define only the layout of
the neural workspace. Every transition and output is learned; there is no
arithmetic routine, recurrence assumption, lookup table, or generated target.
"""
from __future__... | {
"id": "4f6c8496-b10f-4b69-9f4c-6edd9c3f52d0",
"created_at": "2026-08-07 10:38:29.881014+00:00",
"db_md5": "88f17177da0aed1a414cf2200a2ef5e5",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "8291b691-0b31-4e53-946f-9649f0a2f08f",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.551335147848715,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014583333358168603
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.787098225991287,
"example_count": 600,... |
4f7667a8-fd4e-499d-a13a-f7e1e58a2a63 | easy | kushasareen | 2026-08-10 21:24:43.813402+00:00 | succeeded | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "4f7667a8-fd4e-499d-a13a-f7e1e58a2a63",
"created_at": "2026-08-10 21:24:43.813402+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Kusha Sareen",
"github_login": "kushasareen",
"run_id": "61575266-f606-4265-abaf-7615d3c60793",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.869675917856358,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.94195332227801,
"example_count": 600,
... |
4f76a329-e899-4caa-b5f9-fbb4412268d9 | easy | velocizapkar | 2026-08-06 21:45:48.000093+00:00 | succeeded | 0439f0c080faab3c175acace5b1b4c58fd960d61da7b035aea0a2bba7c1b99fb | 52,962 | null | """Fast packed residual-free Attention-Residual transformer with Muon.
The model is a generic learned sequence model. It contains no parser,
arithmetic primitive, task constant, generated data, label feedback, or answer
construction. Every macro-state is produced by a serial pack of attention and
MLP transformations... | {
"id": "4f76a329-e899-4caa-b5f9-fbb4412268d9",
"created_at": "2026-08-06 21:45:48.000093+00:00",
"db_md5": "d926bb07b5c4509c473f973819163856",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "4582ff72-9110-4b98-ae8a-033281dda525",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 3.365175485610962,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01749999960884452
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.3824431896209717,
"example_count": 800,... |
4f850af3-5d8a-41e0-8f65-c19b2cb5a69f | easy | viridale | 2026-08-05 21:45:26.946462+00:00 | succeeded | 43a63ac96e9a2483c9124dbf28393f4205e3701b527c96384299b631162ccf05 | 8,369 | null | """Clean v177 recurrence with a learned-phase sinusoidal latent gate.
hypothesis: ordinary SiLU networks have a severe spectral/width bias on modular maps;
a sine gate in the shared recurrent computation may expose compact periodic latent
features while a linear carrier preserves nonperiodic prompt information... | {
"id": "4f850af3-5d8a-41e0-8f65-c19b2cb5a69f",
"created_at": "2026-08-05 21:45:26.946462+00:00",
"db_md5": "80a62a0f51b443f986f74c9dc1f65826",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "4938cd85-7206-450a-9f03-06f507fe9d09",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 2.772439479827881,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666716337204
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.762269973754883,
"example_count": 100,... |
4f92e2d7-ac57-4302-9186-dcd7083b7cc5 | easy | sapient-sapiens | 2026-08-18 10:19:54.989424+00:00 | succeeded | 512669003ca5929e750f194eb0039783cd44f276400db53c5dc1c0a507da0a3e | 24,076 | null | """v43: v40 lineage with the halting controller removed and a fixed unroll.
Diagnosis this arm responds to. Across every stored local run the learned
controller emitted a single constant depth for the whole dataset (1032 of 1037
depth-vs-T cells had exactly zero per-example variance), and forced-depth
sweeps showed lo... | {
"id": "4f92e2d7-ac57-4302-9186-dcd7083b7cc5",
"created_at": "2026-08-18 10:19:54.989424+00:00",
"db_md5": "04597942bd92287f233d24b24b6ce3c2",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "60ffb940-be6e-4780-828b-ce2e0d155f93",
"tier": "easy",
"dataset_id": "e3",
"status": "su... | {
"score": {
"mean_loss": 2.2490834548230554,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009375000000000001
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2413834689994494,
"example_count": 80... |
4fa743b6-2d15-44b4-9ca1-a1aa13d79664 | easy | velocizapkar | 2026-08-30 16:41:06.903623+00:00 | succeeded | 5c8ee47bbc801493bd7ba5eb52143381406519f108fd1d2986fa986067fa1ac7 | 7,486 | null | """HeightGrid: a (sequence x depth x height) transformer for One Layer Deeper.
The computation is a 3D grid. Each cell (t, d, h) holds a width-W vector and
receives three interactions:
* sequence: softmax attention over cells (., d, h) in the same plane;
* height: the running state ascending from (t, d, h-1);
... | {
"id": "4fa743b6-2d15-44b4-9ca1-a1aa13d79664",
"created_at": "2026-08-30 16:41:06.903623+00:00",
"db_md5": "60f12ad7eb595dacf72033c30dfd11dd",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "1721427c-1c88-4008-9465-f9742997466c",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 2.270150121027916,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018125000046566128
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2736069351672925,
"example_count": 800... |
4fb27633-76bc-477a-965e-9faae20eb3a1 | easy | Yalyenea | 2026-08-11 20:45:44.135596+00:00 | succeeded | e5bca2ead93086f06e8eee3bf8ee93748ddd43b8644e6c652467352798d3709f | 12,666 | null | """K2/R2 layer-loop candidate: B1, B1, B2, B2 with full-bandwidth feedback."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_stat... | {
"id": "4fb27633-76bc-477a-965e-9faae20eb3a1",
"created_at": "2026-08-11 20:45:44.135596+00:00",
"db_md5": "a18516e03b7d5107039249345a047576",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "395950ce-ecb0-4a11-923e-83ed795bb42f",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 6.76984431789265,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010833333805203437
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.729857444763184,
"example_count": 600,
... |
4fb3da20-eaac-4fb3-b621-e96339f101b6 | easy | khushidahi | 2026-08-21 21:38:06.237486+00:00 | succeeded | 0649c024af5ee1964e47628878a92c0c9a9023f41b6e6566c21f6afcc623497d | 40,006 | null | """R375: stable copy-gated ordered recurrent numerical program.
This retains R367's ordered, position-equivariant product/reduction core and
adds one content-dependent learned copy gate. At each public digit place the
gate forms a convex probability mixture between the computed next digit and
the current categorical ... | {
"id": "4fb3da20-eaac-4fb3-b621-e96339f101b6",
"created_at": "2026-08-21 21:38:06.237486+00:00",
"db_md5": "ba35ee39497aab800d8e6c5aac5f2e2a",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "2901c0df-6ca1-499a-89f8-6528edf43763",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 2.23033538304316,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000055879354
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.217695241321472,
"example_count": 800,
... |
4fb54ee9-e77a-46de-a027-947f9073f0f5 | easy | gauravmishra | 2026-08-10 08:34:58.938891+00:00 | succeeded | 814186802a4485b2baf01f9920fa0677a8b9b7e69393c53c68749de81a6488e8 | 9,452 | null | """E52: E25 TRM-lite with separate T5 relative bias in each unique cell."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_... | {
"id": "4fb54ee9-e77a-46de-a027-947f9073f0f5",
"created_at": "2026-08-10 08:34:58.938891+00:00",
"db_md5": "1d3080b958e4f72f5d3d6e767eea6f3b",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "72d2fae4-7986-434a-ae6d-2a651799edfd",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.266932569255304,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0104166666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.299385735806863,
"example_count": 600,
... |
4fb8bcb7-b837-45d9-ad3e-fdefe0571282 | easy | ehonig | 2026-08-14 00:43:39.396976+00:00 | succeeded | bed7b46c989217f150e8f32594973831db97cd435e186415766922012faf890c | 48,079 | null | """DigitLoop — a place-value recurrent model for repeated modular squaring.
The task is ``y = x^(2^T) mod N`` with the answer read back tail aligned, one
decimal place value per token position from the right. A stack of attention and
MLP layers has no primitive for the two operations this needs — a partial-product
co... | {
"id": "4fb8bcb7-b837-45d9-ad3e-fdefe0571282",
"created_at": "2026-08-14 00:43:39.396976+00:00",
"db_md5": "8a71b19028930edd1dd4facf7e650670",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "cbcf1f95-00b9-46a7-81a5-4429b60cef5f",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 5.791956067085266,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.14499999955296516
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.082489967346191,
"example_count": 100,
... |
4fb9dcc2-58b0-47b3-a5c1-9c2ea077857d | easy | achyutreddy24 | 2026-08-04 23:14:24.680939+00:00 | failed | 4883a94f26829ac63e2dbbcab6f1323dea8e06e76b67d6c79ebb14775b73c54a | 11,615 | null | """A looped transformer with learned, input-conditioned depth.
WHAT THIS IS
============
This submission contains no task-specific arithmetic: no modular structure, no
hand-set initialization, no token-map constants. It is a single generic
transformer block applied repeatedly to a latent state, with a learned halting... | {
"id": "4fb9dcc2-58b0-47b3-a5c1-9c2ea077857d",
"created_at": "2026-08-04 23:14:24.680939+00:00",
"db_md5": "3f5d758a65f74f85e85277418625e2ff",
"submitter": "Achyut Reddy",
"github_login": "achyutreddy24",
"run_id": "da69c33b-ebd5-4ca5-a217-c6a60e1077a5",
"tier": "easy",
"dataset_id": "e1",
"status": ... | null
|
4fbc266b-d8e7-4ec3-9acd-8575aeb4569c | easy | KaustubhKumar05 | 2026-08-25 15:53:52.289876+00:00 | succeeded | 66c0d17ac95330bfa44c7d3f372988009e3ed4e7cbcca72adeca514afd3de49f | 14,717 | null | """Neural-GPU-style conv GRU, v6: v2s + relaxed weight sharing.
Four cell copies cycled by iteration index (Kaiser & Sutskever's "crucial"
trick for multiplication). A pull-together penalty on copy divergence is
returned as auxiliary and added to the loss, annealing the copies toward one
shared cell while keeping earl... | {
"id": "4fbc266b-d8e7-4ec3-9acd-8575aeb4569c",
"created_at": "2026-08-25 15:53:52.289876+00:00",
"db_md5": "cebdbe1f9f283b2f115dc4641a0cdd29",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "384a6cfd-f5c3-4089-8787-fcaa1286439b",
"tier": "easy",
"dataset_id": "e3",
"status": "succee... | {
"score": {
"mean_loss": 2.2678898647923367,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.255710128078854,
"example_count": 800,
"ex... |
4fbe612a-3673-4dc9-9c1e-3e2c1548cbf4 | easy | gauravmishra | 2026-08-08 21:18:02.700354+00:00 | succeeded | b0b55fee7e56e6941e5a1bc53d69113790983b1be1b71bbe629ff7460b3da524 | 4,643 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
D_MODEL = ... | {
"id": "4fbe612a-3673-4dc9-9c1e-3e2c1548cbf4",
"created_at": "2026-08-08 21:18:02.700354+00:00",
"db_md5": "161f77f076521ee63b89c3e3b8b443d3",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "4a951fb9-0634-4670-bacd-8073613d5a38",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.221916959658599,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0037500000000000003
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2382747699327,
"example_count": 600,
... |
4fc57c52-35fc-4f65-9ca0-1a9b5f6824ad | easy | ehonig | 2026-08-20 08:49:40.194702+00:00 | succeeded | 52cd19652d4e681e568fdf901cb65712581a52dee2d238f71dcdb1a06575ab14 | 23,245 | null | """A plain Transformer for repeated modular squaring, and nothing else yet.
This is a deliberate restart. The previous submission accumulated a field
parser, a carry scan, a learned reciprocal, periodic features, a digit
bottleneck, a depth selector and two cell types, most of them tuned around a
memorisation table th... | {
"id": "4fc57c52-35fc-4f65-9ca0-1a9b5f6824ad",
"created_at": "2026-08-20 08:49:40.194702+00:00",
"db_md5": "b1508dcb986f47ef32ba38adcb9d207c",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "7203635d-f875-41ed-aa58-4eddfed93e9d",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 2.226195454597473,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006874999962747097
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2209815979003906,
"example_count": 800... |
4fc64fcd-592b-4c5a-a9d5-519a10e680b1 | easy | himalalps | 2026-08-06 16:07:13.381197+00:00 | succeeded | a86c1a6a6e5999dce31bba07e7ef02d1da36960999ba7be7737ddef7d28c462d | 62,352 | null | """Algorithmically constrained latent recurrence for One Layer Deeper.
One masked cross-attention pass creates two independent representations:
``state_0`` from X only and an immutable context from N only. T is excluded
from both representations and is used solely to choose how many times the
learned recurrent MLP tr... | {
"id": "4fc64fcd-592b-4c5a-a9d5-519a10e680b1",
"created_at": "2026-08-06 16:07:13.381197+00:00",
"db_md5": "1e5e504c6949fce9565c8ca96407eab4",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "a1acf6ba-dcde-425b-82c3-ca30a209a38a",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.213659193333205,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008749999900658926
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1967201232910156,
"example_count": 600... |
4fcd42ff-9b9b-47d3-a739-eff42dd1f945 | easy | DDanlov | 2026-08-08 09:47:46.222080+00:00 | succeeded | 78265216016704bb3a785025746d7f3f68f32c69f22c0c4f28abf6e770a9ac61 | 13,315 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class RohanShampoo(Optimizer):
"""Self-contained RohanShampoo optimizer with eigenvalue ma... | {
"id": "4fcd42ff-9b9b-47d3-a739-eff42dd1f945",
"created_at": "2026-08-08 09:47:46.222080+00:00",
"db_md5": "0b65b4c2256c2261a298ce468451477a",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "5a7e80df-11ac-4dd8-8485-1e2394294081",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 158.43460459721058,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.029999999552965166
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 158.7584686279297,
"example_count": 100... |
4fdb06f5-20ce-4b9f-9863-cdc74ae1b7b4 | easy | sapient-sapiens | 2026-08-09 14:50:50.964080+00:00 | succeeded | 1763607be036fe83d25121757c497c2d5b7d375786d04e914da63b7224eb45c8 | 11,416 | null | """Near-500M adaptive recurrence grown from the hosted-working baseline."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
D_MODEL = 6400
NUM_HE... | {
"id": "4fdb06f5-20ce-4b9f-9863-cdc74ae1b7b4",
"created_at": "2026-08-09 14:50:50.964080+00:00",
"db_md5": "7975d62f91a0f4a130ee0cdd08af97db",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "32a49d57-8fa2-4edd-96f8-e8ea4790b73f",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 2.298509778466339,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08333333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.35489224097622,
"example_count": 100,
... |
4ff26571-2e81-432f-9be6-9ae061a7f15c | easy | khushidahi | 2026-08-11 14:06:19.786442+00:00 | succeeded | 34ae3746989f6feda6c5a430d40339a8076f5f4de76b4099d0ec09c3a7559174 | 26,106 | null | """R51 learned bilinear diagonal-routing model.
The architecture uses a learned low-rank pair feature for decimal-place slots
i and j, then a parameter-free scatter-add to slot i+j. The pair features,
carry propagation, modulus conditioning, recurrent update, and output decoder
are all learned end-to-end.
No digit mu... | {
"id": "4ff26571-2e81-432f-9be6-9ae061a7f15c",
"created_at": "2026-08-11 14:06:19.786442+00:00",
"db_md5": "af9feb7025029682e9ae58fd208c934b",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "9feddcf8-83ed-4e52-8c88-f54d6079024b",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.160570863772355,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0075
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1623188091500456,
"example_count": 600,
"e... |
4ff6844f-744f-41ca-a775-a6cc7ffd9853 | easy | velocizapkar | 2026-08-03 22:59:48.577439+00:00 | succeeded | db121fc30d368b3c4cdc0684ef8269dd1dccd487b3b0a2ae5499439ffe9abe2f | 28,858 | null | """Orthogonal depth lattice with recurrent Block AttnRes.
The learned state has three explicit axes: sequence/work position, parallel
computational slot, and recurrent reasoning time. The working transition is
a composition of Householder mixers along the first two axes and invertible
radial twists along features. A... | {
"id": "4ff6844f-744f-41ca-a775-a6cc7ffd9853",
"created_at": "2026-08-03 22:59:48.577439+00:00",
"db_md5": "0e770dac07e1d7071cbeacd67880e0b6",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "fbc171b9-a40d-4b09-bbd3-45708c4bc0c6",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 2.156580746286268,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1571590966620664,
"example_count": 800,
"... |
4ffb7681-d9a0-43fd-b708-73407bedffc8 | easy | DDanlov | 2026-08-28 21:37:54.673556+00:00 | succeeded | 553ba637776f719d1ef30cbf80057c8d76ff885eeb125bf2b249073d45ce62e4 | 15,431 | null | """
Trial 08: Model A: D=384, H=6, Trec=256
"""
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, TokenLos... | {
"id": "4ffb7681-d9a0-43fd-b708-73407bedffc8",
"created_at": "2026-08-28 21:37:54.673556+00:00",
"db_md5": "2accf6b61b993e701e10a887aa1b9365",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "4195b602-2697-4bc0-866d-15c0cfdfa6fd",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.325386454095813,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.341868502810397,
"example_count": 600,... |
4ffda9dc-d447-4d84-bfab-10603db3949a | easy | oupadhyay | 2026-08-12 06:47:02.483366+00:00 | succeeded | 847707c5967b5218a81f0153f312f64b452cde20b1f0f55e79109c29847a6c79 | 8,254 | null | """Standalone RMS/axial crossing C21 for the Easy E5 tier."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
PAD, N_MARK, X_MARK, T_MARK, DIGIT_BASE =... | {
"id": "4ffda9dc-d447-4d84-bfab-10603db3949a",
"created_at": "2026-08-12 06:47:02.483366+00:00",
"db_md5": "2b7e39cd2eafc84dd31699ee85eaf203",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "536f8b9d-1892-44fa-a1a8-d00e01f47052",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.212588401339066,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000037252903
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1539957170529216,
"example_count": 600... |
4fff0c9e-be44-4f65-9dc5-d4d318eab3c0 | easy | blake-camp-surge | 2026-08-05 19:35:13.218347+00:00 | succeeded | 8a86a5db023f9b8c14f5958fe804f7fafc1bf829a26ecae9b881b03d7c3204d8 | 10,339 | null | """Dual pre/post NCAs with Hebbian attention-logit updates."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
TOKEN_WID... | {
"id": "4fff0c9e-be44-4f65-9dc5-d4d318eab3c0",
"created_at": "2026-08-05 19:35:13.218347+00:00",
"db_md5": "d2556d6ef412abe38a20dfb8beb84138",
"submitter": "Blake Camp",
"github_login": "blake-camp-surge",
"run_id": "fdee38f0-c9ef-4e97-859e-2a36f920465a",
"tier": "easy",
"dataset_id": "e3",
"status":... | {
"score": {
"mean_loss": 2.133340766519731,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011875
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1362852233771172,
"example_count": 800,
... |
50016996-ec63-4796-b60a-87133c6a0cf9 | easy | sapient-sapiens | 2026-08-14 12:22:13.729352+00:00 | succeeded | 3414902c6e5cc0c4e7c225ea8127776e2b43880fa8ff6c82b707d57d460ceb67 | 16,884 | null | """Fixed-T=2 tied Transformer for the E3/E4 operator-recovery campaign.
This file is the authoritative, standalone hosted submission template. The
variant generator changes only the constants in the marked configuration
section. The model never reads T: its four-block core is applied exactly twice.
"""
from __futur... | {
"id": "50016996-ec63-4796-b60a-87133c6a0cf9",
"created_at": "2026-08-14 12:22:13.729352+00:00",
"db_md5": "5d2efd0bdd8eef8a8db63d14c275e514",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "e707c33e-0910-41cf-8ed3-3069c9fa0c67",
"tier": "easy",
"dataset_id": "e4",
"status": "su... | {
"score": {
"mean_loss": 2.498629006036866,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004687500000000001
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5201393782231247,
"example_count": 120... |
5008b1a2-15f0-478c-9fbd-b72ccb00a274 | easy | AFeinfield | 2026-08-29 03:42:33.971411+00:00 | succeeded | f85d167c820d3782292152545fcf00c80dcf3cbf361b2d6cf10f6e6aa0bfa1be | 16,612 | null | """205: 204 + TABLE DETACH in the training forward. Local isolation
shows drills snap all 7 tables in 10-20 steps; in context the end-CE
gradient noise through hundreds of ST ops inflates Adam moments and
stalls them. Detaching tables in the forward (training only) leaves the
clean drill gradient as their sole teacher;... | {
"id": "5008b1a2-15f0-478c-9fbd-b72ccb00a274",
"created_at": "2026-08-29 03:42:33.971411+00:00",
"db_md5": "87f3fc3a23508c5f92e5740b330f9b1e",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "fd1bcaa2-1bca-4d3d-bf69-dd5223f17e5d",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.7904320738006945,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7870646919777142,
"example_count": 100,
"exa... |
500b6bc4-0f6f-4458-9b98-3d7dea451dcf | easy | chad-atexpedient | 2026-08-06 02:13:46.231820+00:00 | succeeded | b4723e4d659b238af40f8a09eb3f9491dcee13be2d562e8d527340ac9faacd0e | 7,157 | null | """A-Q5: scaled RoPE encoder with a distinct output-query decoder."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
WIDTH = 256
HEADS = 8
FFN_WIDTH = 512
EN... | {
"id": "500b6bc4-0f6f-4458-9b98-3d7dea451dcf",
"created_at": "2026-08-06 02:13:46.231820+00:00",
"db_md5": "24cd2858588d863a23d41a28394e3164",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "8c488148-adac-4bed-bd1d-3a8fb7f9b93b",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 6.428679643316312,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333277453978
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.455474853515625,
"example_count": 600,... |
500c1136-3aa3-45d1-8819-5555f441fc35 | easy | erdavis0 | 2026-08-24 09:19:00.428805+00:00 | succeeded | 9aa21cfd116990232dcd127ca3622994e9056aef336f9240b31d0736aea53bd0 | 8,287 | null | """Learned geometric prompt moments with multiplicative scale binding."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
WIDTH = 72
... | {
"id": "500c1136-3aa3-45d1-8819-5555f441fc35",
"created_at": "2026-08-24 09:19:00.428805+00:00",
"db_md5": "74f468822dce7e749383257b9e669998",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "d79b9d01-72ed-4fa4-bab7-a3103adf309d",
"tier": "easy",
"dataset_id": "e4",
"status": "succeeded",... | {
"score": {
"mean_loss": 7.259775876998901,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0032291666138917208
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.332302570343018,
"example_count": 120... |
5010964e-80d5-4ced-852c-817ef4a133e8 | easy | adwhit | 2026-08-15 07:21:47.429421+00:00 | succeeded | 54be2869571f373154ace5ddc9ea57ade14aa572e657ca2286f12f30c599b5ed | 19,604 | null | """GaussNet-demod3: independent 2^k phase heads, unwrap-ladder digit readout (parameter-free, unit-tested in tests/test_demod_math.py).
Structure (all parameters learned from random init):
- Learned parse: cross-attention queries extract LSB-first digit registers
for x, copy-supervised from the prompt's own toke... | {
"id": "5010964e-80d5-4ced-852c-817ef4a133e8",
"created_at": "2026-08-15 07:21:47.429421+00:00",
"db_md5": "02e227f235b64c870233e1d36c0cbd5b",
"submitter": "adwhit",
"github_login": "adwhit",
"run_id": "0fa92f9c-27b4-4494-b266-11fab5738141",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.7663707733154297,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7667908668518066,
"example_count": 300,
"exa... |
50122da7-d397-42cf-b4c4-a43d5b1eb025 | easy | erdavis0 | 2026-08-29 01:39:32.567666+00:00 | succeeded | eadd19360be6ecdb82c0d25f4d3447d162d2edd2acd789ba5e873fafaac8fc3b | 19,602 | null | """Four-code N-conditioned circulant operator over an X-initialized tape.
Public delimiters only route opaque categories into independent right-aligned
N, X, and T tapes. N makes a straight-through hard choice among four globally
learned cyclic operators; X alone initializes recurrent state; T is consulted
only after... | {
"id": "50122da7-d397-42cf-b4c4-a43d5b1eb025",
"created_at": "2026-08-29 01:39:32.567666+00:00",
"db_md5": "be2bf3989baac5e04978d28d9e6dbbb4",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "9088945e-6982-48f5-99ea-231dee3d81ff",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 4.947743892669678,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04833333194255829
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.535489559173584,
"example_count": 100,
... |
5013c4c4-0311-4fcc-a725-bea60505296b | easy | Yalyenea | 2026-08-11 21:05:26.330838+00:00 | succeeded | 135ae1ff1e13b0576b6b0881e9761382154209ce0f539db5d506f2178f315178 | 12,309 | null | """R4 full-bandwidth candidate with 0.1 residual dropout and cosine horizon 2000."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_mode... | {
"id": "5013c4c4-0311-4fcc-a725-bea60505296b",
"created_at": "2026-08-11 21:05:26.330838+00:00",
"db_md5": "98dcd95975a9d5c3f8659102fce241d3",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "5ff36282-f0c8-46a3-8883-d156251df985",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 6.0415686578363985,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012916667247191071
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.923677921295166,
"example_count": 600... |
50164b65-0ff8-47ad-84ce-39356047422e | easy | chad-atexpedient | 2026-08-27 03:08:09.683554+00:00 | succeeded | a5684f606d06f59b8ca4639cfef72988e9a37c07e5c6d217ac684ab0beb6f602 | 17,562 | null | """X134: low-weight horizon-agnostic successor alignment.
X133 activates corrected chains on fixed-T Medium but 0.05 overconstrains Easy.
X134 reduces only this denser chain loss to 0.01, seeking X131's ordinary safeguards
while preserving active Medium compositional supervision.
"""
from __future__ import annotation... | {
"id": "50164b65-0ff8-47ad-84ce-39356047422e",
"created_at": "2026-08-27 03:08:09.683554+00:00",
"db_md5": "34960291e5e5110a01beb38f4f57d640",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "702a71cd-61e0-44ac-8f12-0565656d0dda",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.2108352116070153,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.232450489506059,
"example_count": 600... |
5016e836-c586-447a-8808-d62373d6efb2 | easy | gauravmishra | 2026-08-07 21:01:26.923555+00:00 | succeeded | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "5016e836-c586-447a-8808-d62373d6efb2",
"created_at": "2026-08-07 21:01:26.923555+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "bb927da0-d5f5-4c8c-80fb-0e6239195629",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.828606568949912,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.9915636173812796,
"example_count": 600... |
50254790-78b2-4ef4-97f6-fede9f7ce462 | easy | karanganesan | 2026-08-05 11:01:43.999648+00:00 | succeeded | a4311084e5217cb8ae8acdf64c9f8c8fcdd1a34e9f432b5abc4dd2fa75fff1e9 | 22,082 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "50254790-78b2-4ef4-97f6-fede9f7ce462",
"created_at": "2026-08-05 11:01:43.999648+00:00",
"db_md5": "2a644a4e7d5021959fe0a80222e736b1",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "4ef20245-8cc1-4b1d-87eb-cf9dd095b2cf",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 14.477410793304443,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04333333298563957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 15.299894332885742,
"example_count": 100... |
502af0ec-1346-4642-8a0c-055b3820a458 | easy | DDanlov | 2026-08-15 14:15:24.090428+00:00 | succeeded | d1a86da0c79fc09427362c2a91eb83d1da98fae14207fc98e38608001e9f0376 | 18,985 | null | """
Official Submission for One Layer Deeper Challenge
Architecture: Pure No-Norm, No-Skip DKS Recurrent Core with Double Embedding (dim=512),
Input RMS Normalization only, 64-Channel Harmonic+Dyadic Torus, Power-of-10 Alignment, and RohanShampoo.
"""
from __future__ import annotations
import math
from typing import O... | {
"id": "502af0ec-1346-4642-8a0c-055b3820a458",
"created_at": "2026-08-15 14:15:24.090428+00:00",
"db_md5": "6802e795e913e453547a4cd42e8230c6",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "6e867323-ea20-4590-9b18-3681291f40f5",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 17.180274992698703,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.020000000298023225
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 21.248766272815306,
"example_count": 10... |
502d52f3-ffd3-4f3a-8773-aa97370f1554 | easy | sankalp1999 | 2026-08-04 12:11:04.140255+00:00 | succeeded | 1c5a8f36afd6aba8797a83dd14dbcbb51809ae0965bb857f5eba27979620f5e9 | 7,511 | null | """Latent recurrence: encode once, apply a shared squaring map T times, decode.
v4 put the loop inside a transformer and underfit badly (train acc 4% after
4,432 steps) because attention cost ~7ms/step at 13 tokens and bought nothing.
v2 fit the train set perfectly with a cheap flattened encoder but memorized,
because... | {
"id": "502d52f3-ffd3-4f3a-8773-aa97370f1554",
"created_at": "2026-08-04 12:11:04.140255+00:00",
"db_md5": "e0f6b68e236097e124d2bee64ffe8ecb",
"submitter": "Sankalp ",
"github_login": "sankalp1999",
"run_id": "6eee3b49-74cb-4d22-8720-32c59e82c777",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 8.138908386230469,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.10499999672174454
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 10.511959075927734,
"example_count": 100,... |
50308379-3552-48d4-a52b-1859111755d6 | easy | sapient-sapiens | 2026-08-16 19:50:31.889009+00:00 | succeeded | f2b8d7b3a5f780392b7a86763c751e69f8719be15a750e8e95936a97971c1dfd | 22,444 | null | """E6 SAM-family variant: frozen v55 with a one-variable SAM change.
Architecture, representation, HybridMuon split, rho=0.02, batch, wall-clock
schedule, and loss are unchanged from act64_e6_w1024_dh4_L6_v55. T is
diagnostic-only.
"""
from __future__ import annotations
import math
import time
import torch
import t... | {
"id": "50308379-3552-48d4-a52b-1859111755d6",
"created_at": "2026-08-16 19:50:31.889009+00:00",
"db_md5": "ea9b4daf95354e36aeff1ea73eadd96a",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "2b4e4af9-a996-43a5-869e-e3a621cac809",
"tier": "easy",
"dataset_id": "e6",
"status": "su... | {
"score": {
"mean_loss": 6.774322258283014,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.12184685529083819
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.842195987701416,
"example_count": 60,
... |
50343629-5f4d-429d-8c7f-c9fc52a72eb6 | easy | velocizapkar | 2026-08-29 15:43:02.252985+00:00 | succeeded | 50d5af08ff3a67a5e2e0d9f060a2b10945d2dde83052900c7a169ec17e7aed55 | 14,141 | null | """Population-of-polynomial-programs model for One Layer Deeper.
Hypothesis class: iterated quadratic maps over Z_N,
f_{a,b,c}(z) = (a * z^2 + b * z + c) mod N,
applied T times, with (a, b, c) drawn from a small integer grid. The model's
trainable state is a factorized categorical posterior over the coefficient... | {
"id": "50343629-5f4d-429d-8c7f-c9fc52a72eb6",
"created_at": "2026-08-29 15:43:02.252985+00:00",
"db_md5": "6ddbb61f8c00f256ec79e9012e5d6a72",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "ac03b201-4140-429d-9ff2-405faf733afd",
"tier": "easy",
"dataset_id": "e7",
"statu... | {
"score": {
"mean_loss": 3.013552998254454e-07,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.3553047273926495e-07,
"example_count": 85,
... |
503a3b8a-479f-4048-bf9b-c7d0a72872a1 | easy | AdityaVG13 | 2026-08-15 07:11:42.798048+00:00 | succeeded | 67963bacdc7792833d3fbf03151c865a43ffa15294d417745eda1bc1202e20bd | 109,392 | null | """AEGIS V10 aegis_v10_sym: Adaptive Epistemic Guarded Interval Synthesis.
One Layer Deeper submission candidate (parent: V6 aegis_primal384) plus a
sign-invariant cos-orbit context token over the unclamped parsed ratio x/N
and an in-batch partner-consistency loss targeting the negation root
symmetry (x ↔ N−x) of x² m... | {
"id": "503a3b8a-479f-4048-bf9b-c7d0a72872a1",
"created_at": "2026-08-15 07:11:42.798048+00:00",
"db_md5": "9f0cc78647d355b1de61851679711783",
"submitter": "AdityaG",
"github_login": "AdityaVG13",
"run_id": "475e3a46-20bb-4f52-ac93-b76991ecbf3d",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 2.774090588092804,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6233333349227905
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.4820085763931274,
"example_count": 100,
... |
503e15aa-47d3-4355-82b2-98b7892938b8 | easy | achyutreddy24 | 2026-08-04 23:03:25.731744+00:00 | failed | 3daeaca446c8e3068996027f4e5fbb62db9c5f2804925621c76d60a0516f7dec | 11,500 | null | """A looped transformer with learned, input-conditioned depth.
WHAT THIS IS
============
This submission contains no task-specific arithmetic: no modular structure, no
hand-set initialization, no token-map constants. It is a single generic
transformer block applied repeatedly to a latent state, with a learned halting... | {
"id": "503e15aa-47d3-4355-82b2-98b7892938b8",
"created_at": "2026-08-04 23:03:25.731744+00:00",
"db_md5": "57b7554a25d85d8e4ab5ebe50a5e80f6",
"submitter": "Achyut Reddy",
"github_login": "achyutreddy24",
"run_id": "6538e93d-0436-41c2-82c7-c25b9359e7c4",
"tier": "easy",
"dataset_id": "e1",
"status": ... | null
|
504f2b4a-fc27-407b-9def-0f78d28265ed | easy | ehonig | 2026-08-20 18:58:16.497149+00:00 | succeeded | 836b96fc84c65fae3cb8a2fde64f977e7d832e6b391feee225fac06fe5719c8d | 25,077 | null | """A plain Transformer for repeated modular squaring, and nothing else yet.
This is a deliberate restart. The previous submission accumulated a field
parser, a carry scan, a learned reciprocal, periodic features, a digit
bottleneck, a depth selector and two cell types, most of them tuned around a
memorisation table th... | {
"id": "504f2b4a-fc27-407b-9def-0f78d28265ed",
"created_at": "2026-08-20 18:58:16.497149+00:00",
"db_md5": "2a8cc00c911349e65d60bcb057920d6f",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "d920339d-38e7-499e-8293-3ef964032da1",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.1059519052505493,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07000000029802322
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0979931354522705,
"example_count": 100... |
504f8d59-f2d4-4ee4-89d7-1b6330529208 | easy | DDanlov | 2026-08-31 20:33:07.392650+00:00 | succeeded | 3d3f38ed516ebe3d9af4f18aa2cc102a2de70aeb328de1cdb8e9169d730dbbcc | 32,407 | null | """
Unified Continuous Hopfield Equilibrium Propagation Deliberation Engine
=======================================================================
Architecture:
- Deliberation: Continuous Hopfield Multi-Layer Energy System (L=4, D=768, d_ffn=2304, H=16)
- Integration: In-Place Strang-Split Conformal Symplectic Velocit... | {
"id": "504f8d59-f2d4-4ee4-89d7-1b6330529208",
"created_at": "2026-08-31 20:33:07.392650+00:00",
"db_md5": "9eed969d8003bebc391b8c07082dbe31",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "7c126e74-ecc5-4247-89b9-b06695418649",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.202297645578385,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.199663832952959,
"example_count": 800,
"e... |
50514d96-92fa-4053-8562-843ca6206b52 | easy | shan23chen | 2026-08-08 02:07:40.484238+00:00 | succeeded | beb9d7fc6c966d67493c983fa8a58dda84d570c246af729db1a49156ef1cb973 | 20,445 | null | """Generic recurrent Transformer with normalized numeric and CRF objectives.
The model is deliberately task-generic: it learns a transition over prompt and
workspace tokens instead of embedding the public modular-squaring algorithm.
A learned linear-chain CRF models dependencies among adjacent output digits;
its endpo... | {
"id": "50514d96-92fa-4053-8562-843ca6206b52",
"created_at": "2026-08-08 02:07:40.484238+00:00",
"db_md5": "d97dc6388ab9afcc7642e56913649419",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "359406fb-1222-40eb-816a-13ed0c58aad1",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.678447723388672,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01291666692122817
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6598410606384277,
"example_count": 600,... |
5054b405-d35a-455b-a990-47b55218c079 | easy | yashkant | 2026-08-15 14:15:03.414823+00:00 | succeeded | 2a920c54a3b75021f26183100b2c3c8babdcdc24886715ec4e1a11e6be85ad1c | 20,745 | null | """Round1956 Round493 block-two shared-head supervision.
The exact Round493 model, initialization, optimizer, and final prediction path
remain intact. During training both arms additionally read out the hidden state
after encoder block two with the already-existing norm, attention pool,
decoder, and answer head. The c... | {
"id": "5054b405-d35a-455b-a990-47b55218c079",
"created_at": "2026-08-15 14:15:03.414823+00:00",
"db_md5": "b3bf544ca21f3ce76ceb8742d748c33a",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "371190a5-3787-45d4-be65-94230378cff8",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 5.601728200912476,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.10333333723247051
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.721909999847412,
"example_count": 100,
... |
505b4d17-2476-4df1-8a01-ec6fb0ae7934 | easy | jordanrubin | 2026-08-07 01:36:08.785728+00:00 | succeeded | 0eb249f609a0101d70cb07bbf9b0f2aa4cc7c5f7465d3e5e29bc4c3e68abecc6 | 14,786 | null | """Autonomous digit-state recurrent Transformer for repeated modular squaring.
The model is deliberately organized around one learned transition:
p_0 = right_aligned_decimal_digits(x)
p_{k+1} = F_theta(p_k, decimal_digits(N))
The same two-block Transformer cell is applied exactly T times. T is used only
as ... | {
"id": "505b4d17-2476-4df1-8a01-ec6fb0ae7934",
"created_at": "2026-08-07 01:36:08.785728+00:00",
"db_md5": "497d08edf33fa52921753c6fa8942084",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "20730ec0-6dff-41ad-af01-e5fff5426d0a",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 4.03409126970103,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00958333311912914
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.113750434108914,
"example_count": 600,
... |
505e1f81-659c-4f17-9db4-f0a5486d7661 | easy | sapient-sapiens | 2026-08-31 20:42:10.640048+00:00 | succeeded | 1b07d4ed8ce05e18db86884456c068cce7977e8e16df196adfe75788f4b632ed | 13,822 | null | """E5 one-block N-conditioned weight-delta recurrence with two-call BPTT windows."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_mode... | {
"id": "505e1f81-659c-4f17-9db4-f0a5486d7661",
"created_at": "2026-08-31 20:42:10.640048+00:00",
"db_md5": "e26a888446daffdf88367e293f3eec43",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "703797d3-2517-47e7-8745-c6307dc6161c",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.157039170255891,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.16049900944992,
"example_count": 600,
"exac... |
505e24a2-007d-4122-ae95-fec6ace30dcf | easy | repst | 2026-08-25 20:20:20.009407+00:00 | succeeded | cb9558014131bedd1deffb2aae83a05978811f38bfcac65273e40f5f8770ac36 | 89,063 | null | # R704: label-free compositional program-key geometry.
#
# R703 localized semantic retrieval failure to anisotropic program embeddings.
# R704 adds an LM-like vocabulary-geometry objective to hosted-safe R699: a fixed
# target-agnostic sample of grammar programs must retrieve itself under cosine
# score plus the existi... | {
"id": "505e24a2-007d-4122-ae95-fec6ace30dcf",
"created_at": "2026-08-25 20:20:20.009407+00:00",
"db_md5": "3e33384578ca8eef380fa642927637fb",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "62427161-ebd5-4a3f-8ba4-876ef7a0228c",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 2.1573332016010234,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1198084354400635,
"example_count": 100,
"exa... |
506269ea-2390-4a70-94dd-801071c55c01 | easy | shirvani-jr | 2026-08-28 17:11:16.596770+00:00 | succeeded | 261880af279df95d325f3302079627f1c9fc6e0fa87aa36be5a9e9603c840a92 | 53,900 | null | """Adaptive learned discrete recurrent fabric for One Layer Deeper.
A gradient-trained, operation-free recurrent computational machine. The
model learns, from the competition's endpoint supervision: how to parse the
prompt into a bank of categorical digit registers, a reusable transition
circuit applied serially, a l... | {
"id": "506269ea-2390-4a70-94dd-801071c55c01",
"created_at": "2026-08-28 17:11:16.596770+00:00",
"db_md5": "d0342a8a403b32f57245d795c5a3f0cb",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "55da9b78-110b-430f-b855-b6b8ea6df1cb",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.8565909266471863,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06500000134110451
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8081023693084717,
"example_count": 100... |
5069bd72-eb13-4aec-9d50-444913455941 | easy | AFeinfield | 2026-08-28 00:45:48.667609+00:00 | succeeded | ac26defeaf4f0163a71bd057aae9291848ee382f3742358a0df94eab7f213c66 | 15,305 | null | """177: 174 + SCHEDULED INTEGER SNAP (soft -> hard at call 400 + eval).
The frontier's test accuracy equals per-op accuracy ^ ops-per-row: the
missing multiplier is per-op EXACTNESS. 170 proved snapping from step 1
removes early partial credit; scheduling it after the operators reach
~+-0.5 accuracy (and always at eva... | {
"id": "5069bd72-eb13-4aec-9d50-444913455941",
"created_at": "2026-08-28 00:45:48.667609+00:00",
"db_md5": "9eb28c6f655044b7dc2489eab3a591c9",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "0f9ee100-4a92-481d-b748-cea192da1c8d",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.281796962514159,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2823004776052294,
"example_count": 600,
... |
506d508a-3ac4-4c50-b5be-4742f4a5b1d9 | easy | Bananafly | 2026-08-11 18:51:32.933399+00:00 | succeeded | 94d832edca3d2bc0c0d39fdcdf0002d6ebc2729786b2873de1fb5b7758f22fbf | 10,217 | null | """Gated recurrent Transformer with latent scratch space."""
from __future__ import annotations
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
ass... | {
"id": "506d508a-3ac4-4c50-b5be-4742f4a5b1d9",
"created_at": "2026-08-11 18:51:32.933399+00:00",
"db_md5": "aee0d19c2285d65bb77413a5fa55a42a",
"submitter": "Andre Kreidemann",
"github_login": "Bananafly",
"run_id": "2bba74bb-e1a3-4751-9bd7-29bf9ae787a9",
"tier": "easy",
"dataset_id": "e3",
"status": ... | {
"score": {
"mean_loss": 2.2103694677352905,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0118749993853271
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.301089286804199,
"example_count": 800,
... |
507e1b54-91cc-4a8c-a250-b1f857f8edb4 | easy | viridale | 2026-08-13 05:49:02.620387+00:00 | succeeded | 5e49104e91a6044724cfa372e8fb11d194fe8497d80791c049a1acded2818d03 | 23,439 | null | """bf16-safe learned-Cauchy parallel multiple-shooting lattice.
hypothesis: v525 passed fp32 preflight and balanced useful gradients, but
evaluator-faithful bf16 found the Cauchy scatter destination inherited fp32 from
latent state while learned factors autocast to bf16. Allocate the accumulator in
the fac... | {
"id": "507e1b54-91cc-4a8c-a250-b1f857f8edb4",
"created_at": "2026-08-13 05:49:02.620387+00:00",
"db_md5": "c563d8699c4cb898def86d127fd6b130",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "d19a199f-71bf-49f3-84aa-afefd7ca39d7",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 16.01770858455477,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333532015483
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 14.77881908416748,
"example_count": 100,... |
50879982-f20b-4e70-b9f4-5ea2cc5cbee6 | easy | gauravmishra | 2026-08-09 18:52:57.961750+00:00 | succeeded | 7ea0c1b5551f67ae52f0e11f84d920ba261254eecfb9e8ddd8118b375dd9b9d1 | 6,078 | null | """Compound-scaled Transformer with distinct blocks and PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_st... | {
"id": "50879982-f20b-4e70-b9f4-5ea2cc5cbee6",
"created_at": "2026-08-09 18:52:57.961750+00:00",
"db_md5": "d0960335107ccaef81ffab81f3398a2d",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "2dd39fc3-b723-401f-907c-7a580a15479d",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.414255440212864,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000024835269
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4671679238032866,
"example_count": 600... |
508f0b22-dc56-46ff-ba96-c55e6eb4a1a3 | easy | DDanlov | 2026-08-13 22:42:47.907743+00:00 | failed | 7c4e37e8a1308f545df5c524c178f1f43226f4ad9b04a993013d99e545075e4d | 22,142 | null | import torch
import torch.nn as nn
import torch.nn.functional as F
import math
from dataclasses import dataclass
from typing import Optional
from torch.optim import Optimizer
try:
from client import Submission, assert_model_state, OptimizerBundle
except ImportError:
Submission = None
assert_model_state = N... | {
"id": "508f0b22-dc56-46ff-ba96-c55e6eb4a1a3",
"created_at": "2026-08-13 22:42:47.907743+00:00",
"db_md5": "7ceedf662db48e66980a3d1a9afc8632",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "a92aeecd-8372-4914-a06b-c7f99d2be2d4",
"tier": "easy",
"dataset_id": "e3",
"status": "failed",
... | null
|
5096e706-3596-418d-a6d9-a4abbc9dcd37 | easy | sapient-sapiens | 2026-08-09 14:17:10.231845+00:00 | succeeded | 3a30c06040fe41222a3639a00e24cb063d1d56cdd45233dce9983f38933c81ec | 7,161 | null | """Near-500M adaptive recurrence grown from the hosted-working baseline."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
D_MODEL = 6400
NUM_HE... | {
"id": "5096e706-3596-418d-a6d9-a4abbc9dcd37",
"created_at": "2026-08-09 14:17:10.231845+00:00",
"db_md5": "f12cb21f978516fc83e42ad3400fe41c",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "106abb63-040f-462d-8677-7fba7589b444",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 2.242600865764981,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.26838710103462,
"example_count": 100,
"exact_... |
5099a02f-097f-41f1-aadc-09e58ebf62d2 | easy | YangWang92 | 2026-08-17 05:38:21.306095+00:00 | succeeded | 44fc930d809c31a6480ef79b59b1620f4f36943377541193bab7b764615787de | 46,820 | null | """Limb (bignum) circuit submission.
The h1 metrics pinned the hard tier at ~60-64 bit moduli (the training-loss
floor of 100-250 is exactly the fp64 quantisation error of 18-19 digit
integers), so every value here is a little-endian limb vector, base 10^4.
No intermediate ever exceeds ~1.4e9, so a single float64 hold... | {
"id": "5099a02f-097f-41f1-aadc-09e58ebf62d2",
"created_at": "2026-08-17 05:38:21.306095+00:00",
"db_md5": "7c68cdea41972b37f3eb5b23a332c615",
"submitter": "Yang Wang ",
"github_login": "YangWang92",
"run_id": "82856706-d0d1-4e60-9804-87d316b234fd",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 800,
"exact_accuracy": 1.0,
"... |
509fbaa5-3417-4439-86e8-b5f0127a5436 | easy | yashkant | 2026-08-16 01:17:15.023747+00:00 | succeeded | af1b8e5136910504a0dc867d6ef5b6c4ac9cd3587750d4045eac81d291171960 | 34,395 | null | """Round1753 slotwise continued-fraction denominator recurrence.
The complete Round1141 D416 MemGLU carrier remains byte-identical. Its pooled
representation seeds a positive terminal denominator. Every shared call adds
one tied continued-fraction level, ``P(x) + softplus(beta)/(state + epsilon)``.
The matched control... | {
"id": "509fbaa5-3417-4439-86e8-b5f0127a5436",
"created_at": "2026-08-16 01:17:15.023747+00:00",
"db_md5": "940164b6590fe476770c4c940ff569f1",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "4598d968-21a6-4614-b8d3-eddc74b26995",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.1456532954482803,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.150198806027126,
"example_count": 600... |
50a5ed3c-cc5b-45d3-9f65-290ff0465fb3 | easy | karanganesan | 2026-08-07 07:34:42.698725+00:00 | succeeded | 2279c66c533e557b3e972c17b765a4a4c0496480f4f9bc526d2a3e02ae203a61 | 25,599 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "50a5ed3c-cc5b-45d3-9f65-290ff0465fb3",
"created_at": "2026-08-07 07:34:42.698725+00:00",
"db_md5": "3b62c89bb2a14de81aa653e6d5d89958",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "bd925b15-7c7d-4017-b6e6-295bda101f39",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.4783002138137817,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004999999888241291
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.480025291442871,
"example_count": 600... |
50a728c7-9a24-4aaf-8870-2ad1b0ea60df | easy | karanganesan | 2026-08-04 01:53:27.853209+00:00 | succeeded | c45f3fdbd3f2b919b60b1248a6543b7605186f42953d5f8a39397bced14985bc | 11,749 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "50a728c7-9a24-4aaf-8870-2ad1b0ea60df",
"created_at": "2026-08-04 01:53:27.853209+00:00",
"db_md5": "4848e0614a0ce13cc9f6abe91797fce7",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "613e3445-a42e-4de7-b6a8-df186ee4462d",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.5592193603515625,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.061666667461395264
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.694786310195923,
"example_count": 100... |
50af8e3a-0002-4a73-80f6-96f1cdae9fc0 | easy | pepinu | 2026-08-26 17:40:17.100778+00:00 | succeeded | 74307ba76d035d71e8330e51db839434ca24ae8d3efbab3f4b12dd2ec11dd273 | 46,270 | null | """Squaring mod N as a learned permutation on CRT components.
WHY THIS SHAPE. (Z_N, x) is a commutative monoid, so every finite-dimensional
linear representation splits into multiplicative characters and squaring is
diagonal there -- the component basis is the only one in which this operation is
shallow. Two consequen... | {
"id": "50af8e3a-0002-4a73-80f6-96f1cdae9fc0",
"created_at": "2026-08-26 17:40:17.100778+00:00",
"db_md5": "d7f34595d7f996ce3ab1db374501668a",
"submitter": "pepinu",
"github_login": "pepinu",
"run_id": "7cbd0d0f-d0a4-48fd-8cc4-7846da5d4e5e",
"tier": "easy",
"dataset_id": "e4",
"status": "succeeded",
... | {
"score": {
"mean_loss": 0.009740500626247175,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.009739833494149962,
"example_count": 1200,
... |
50b41661-3112-4b8c-934c-a80374c79f20 | easy | KaustubhKumar05 | 2026-08-26 00:46:35.490966+00:00 | succeeded | 0e3517abb95e58390c711f53b14275c82487d5f49a21786ba4eae136b3e50a97 | 13,654 | null | """Neural-GPU-style conv GRU, v2s: v2 + speedrun pack.
- LR: warmup, hold max to 85% of wall-clock, steep late cliff (2026 schedule theory)
- identity-at-init: cell output projection zero-initialized
- batch reuse x6 (multi-pass theory)
Base (v2):
Changes over v1 (each backed by measured results in the literature):
... | {
"id": "50b41661-3112-4b8c-934c-a80374c79f20",
"created_at": "2026-08-26 00:46:35.490966+00:00",
"db_md5": "b212c606fed507ca19be2e2e15d31e75",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "6fd3c8c0-f5ea-4fd2-a4ba-0a4395fba711",
"tier": "easy",
"dataset_id": "e6",
"status": "succee... | {
"score": {
"mean_loss": 4.708047151565552,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.37882882356643677
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.623785018920898,
"example_count": 60,
... |
50bf731e-691e-4cb3-90ed-b0d2a3707b53 | easy | sapient-sapiens | 2026-08-23 07:36:36.891072+00:00 | failed | b62ef50b88b1b5d04ac8e995c9ff4f7e3cce2ffe7640b73f0e7894fcd6cce128 | 16,183 | null | """Approximately 10x-scale adaptive child of the D6 pair/quotient architecture.
The architecture uses the required learned additive token/group/left-relative
representation. Marker-derived field lengths additionally route categorical
digit embeddings into right-aligned lanes; this alternative place alignment is
an ex... | {
"id": "50bf731e-691e-4cb3-90ed-b0d2a3707b53",
"created_at": "2026-08-23 07:36:36.891072+00:00",
"db_md5": "fbfd2860a6cd426f28ad4814e943e34c",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "b75fea3d-e627-4728-81d8-9db37d7eccae",
"tier": "easy",
"dataset_id": "e5",
"status": "fa... | null
|
50bf936d-485e-46fa-9f06-c60190a7d1bf | easy | sapient-sapiens | 2026-08-20 12:12:22.508514+00:00 | succeeded | 58976c53968b056b487e8fdaa6bc6d049e5b22392b43bee5663c4c3c14442541 | 33,891 | null | """Depth-quantized recurrence with a mixture-over-depth objective.
The recurrent state is a bank of right-aligned digit slots. Every operator
application is followed by a soft quantization back onto the token simplex, and
the digit logits produced by that quantization are the answer logits at that
depth. There is no s... | {
"id": "50bf936d-485e-46fa-9f06-c60190a7d1bf",
"created_at": "2026-08-20 12:12:22.508514+00:00",
"db_md5": "44e984c5ac7ee439e6456fc1f06478ae",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "5e26059d-3132-43ed-8a41-41bfeac47ab6",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.180396298626299,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.18306185672636,
"example_count": 600,
"exac... |
50c68e96-6ec1-4537-a43a-71c858766a9f | easy | MarcoF1 | 2026-08-13 08:24:07.129757+00:00 | succeeded | fbe213b38fad5332ced983962119ff0670c520f28a36e11de082b55e17c1ea24 | 14,397 | null | """Residue-clock model for fixed-N generalization.
Representation: parsed integer values of x and N feed (a) exact residue-phase
features over a bank of small integer periods, (b) learned double-hashed value
embeddings (per-residue identity), and T feeds cyclic phase features
(T mod 2..48) plus saturating channels — p... | {
"id": "50c68e96-6ec1-4537-a43a-71c858766a9f",
"created_at": "2026-08-13 08:24:07.129757+00:00",
"db_md5": "498f495d8a0b910efb7ca633ca25eee2",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "b346e253-1313-4ee4-b995-cc0cf9695fc5",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.5107572799336424,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4808995438508177,
"example_count": 300,
"exa... |
50d5c08a-1b1c-4df8-9f56-db9ad99d3385 | easy | 0xDEADD06 | 2026-08-12 14:29:50.650036+00:00 | succeeded | c7918823fc7f63551630b7ead01af93d2cf7e7f34e44c924b5f4852cfdfde453 | 5,741 | null | """ParaLSTM: recurrent LSTM with weight-tied passes for algorithmic tasks."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn, Tensor
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL... | {
"id": "50d5c08a-1b1c-4df8-9f56-db9ad99d3385",
"created_at": "2026-08-12 14:29:50.650036+00:00",
"db_md5": "4bf8408490382b83894639b925c716ab",
"submitter": "0xDEADD06",
"github_login": "0xDEADD06",
"run_id": "e8606f4c-aa07-415b-be01-a51874bc5ded",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 2.5661836205634962,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666865348816
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7650183649073803,
"example_count": 60... |
50d94553-5e6e-4e4c-be78-e979d85ccb47 | easy | oupadhyay | 2026-08-14 16:13:42.842344+00:00 | succeeded | 138e203384b301cb857545deade465c59a0b613c78ed68f9cecf4ac8d8cde977 | 4,030 | null | """Low-capacity dual-stream transducer with one shared digit scanner."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,assert_model_state
W,D,H,A=4,64,32,16
class C:
def __init__(s,vocab_size... | {
"id": "50d94553-5e6e-4e4c-be78-e979d85ccb47",
"created_at": "2026-08-14 16:13:42.842344+00:00",
"db_md5": "5a1f3cfeadace2bfac64fa7cb81922cd",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "581c0ee1-911f-46cc-8ddb-c0e0fca0629f",
"tier": "easy",
"dataset_id": "e3",
"status": "s... | {
"score": {
"mean_loss": 2.310420732994542,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000009313226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2991666310366132,
"example_count": 800... |
50df29fc-e404-4f4e-b35b-6e24fef35c22 | easy | oupadhyay | 2026-08-07 12:20:06.580817+00:00 | succeeded | 848d76b57d6939123e18788bbe17ca11a52a1909e179b9f31195cff39bc44267 | 8,335 | null | """Factorized ordered digit-pair context with a tied learned reducer."""
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle, Submission, assert_model_state
PAD, N, X, T, ANS, DIGIT = 0, 2, 3, 4, 5, 7
D, PLACES, MICROPHASES, MAX_STEPS = 48, 4, 1, 64
STATE_ELEMENTS =... | {
"id": "50df29fc-e404-4f4e-b35b-6e24fef35c22",
"created_at": "2026-08-07 12:20:06.580817+00:00",
"db_md5": "eafe5a093f4758be8508b8a9c1573d12",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "6a6fecaa-b615-4a15-8147-1b38e93953fb",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.19550039287027,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2006520476576457,
"example_count": 600,... |
50e4e057-3721-4744-a07d-d3f88d9fea8c | easy | 0Chris5R | 2026-08-07 12:47:14.594852+00:00 | succeeded | 75e07d0798a15b5a529d20dabbaf3a084a40090bace134cce1f06cfc97824c89 | 20,838 | null | """Shared-digit Neural GPU with a compute-matched operator curriculum.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the ... | {
"id": "50e4e057-3721-4744-a07d-d3f88d9fea8c",
"created_at": "2026-08-07 12:47:14.594852+00:00",
"db_md5": "6f108959f530672f53b0bb40916f21b1",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "320ee139-9a3e-424b-9ec1-6f6f10e1bcc4",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.7707890944805755,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02875000002483527
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.731927360654412,
"example_count": 600,... |
50ea7756-0a84-4b9d-bfda-4357459a155c | easy | KaustubhKumar05 | 2026-08-25 16:20:49.005793+00:00 | succeeded | cede8f877bf90dc9d38af3879dc3680a73b6693631c3476085830cfc30a955d0 | 16,027 | null | """final_v1: culmination candidate. v8 (N-FiLM) + Muon momentum 0.5 + tail-EMA.
A small MLP embeds the modulus digits (mean-pooled over the N span) and
emits per-channel scale/shift applied to the cell input and a bias on the
update gate, every iteration. Hypernetwork-lite: the executor is modulated
by N rather than h... | {
"id": "50ea7756-0a84-4b9d-bfda-4357459a155c",
"created_at": "2026-08-25 16:20:49.005793+00:00",
"db_md5": "4c081115900db9bcea126ae905aeee44",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "afb615e2-065c-41f1-8f54-6dbd0a281d98",
"tier": "easy",
"dataset_id": "e8",
"status": "succee... | {
"score": {
"mean_loss": 6.57348108291626,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.14215686544775963
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.950469493865967,
"example_count": 85,
... |
50f1d468-d0a4-4339-88c8-56c08278f2dd | easy | repst | 2026-08-24 21:45:59.998615+00:00 | succeeded | 465eddd238a8f89c280b3cec7d88b0b2ddfb2c11d91ef54fecd5e8916c0bea45 | 30,476 | null | # R638: compute-scaled persistent-register neural recurrence machine.
#
# R638 COMPUTE-SCALING DELTA:
# The complete general quotient receives 16 rather than 8 uniform proposals per
# update, and training may continue for 4,096 updates. Time-limited tiers stop
# at their evaluator deadline; Hard can spend its larger a... | {
"id": "50f1d468-d0a4-4339-88c8-56c08278f2dd",
"created_at": "2026-08-24 21:45:59.998615+00:00",
"db_md5": "26019cd01b2e4f1ada156e7a67293fcb",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "1b76e326-4846-4e10-a70e-f63c409ae6dc",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 0.09090183886946535,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.09133659297759089,
"example_count": 600,
"e... |
50f4383d-03c8-4060-8db9-15dd52fe2cff | easy | DDanlov | 2026-08-13 22:38:56.232129+00:00 | failed | 654bcba4cea296daf285211d1b0d76aeccbd2a9385f9b8f0239cc1c4836874b8 | 22,145 | null | import torch
import torch.nn as nn
import torch.nn.functional as F
import math
from dataclasses import dataclass
from typing import Optional
from torch.optim import Optimizer
try:
from client import Submission, assert_model_state, OptimizerBundle
except ImportError:
Submission = None
assert_model_state = N... | {
"id": "50f4383d-03c8-4060-8db9-15dd52fe2cff",
"created_at": "2026-08-13 22:38:56.232129+00:00",
"db_md5": "fec3750d33d011c885790a1ecc498900",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "5fc87a11-9d06-4c2e-b8fd-8e8f509bffb2",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
5109e479-ce19-4a08-9a35-ed9a84f1f021 | easy | jordanrubin | 2026-08-14 01:28:07.998382+00:00 | succeeded | 69ece06bd94846b5a380dc94d4c00c6798b85f8c1e9f0ba91ba31bcdb9376a77 | 29,580 | null | """Weight-tied MLP-cell loop over soft digit states (mlploop).
Cell = MLP over [expected digits, pairwise digit products, RNS residue
simplices (fixed differentiable mixing over Z_p), N digits]. The cell that
learns one-step modular squaring from direct pairs (52% unseen at 14k rows,
day-20 screen) inside the exact-T ... | {
"id": "5109e479-ce19-4a08-9a35-ed9a84f1f021",
"created_at": "2026-08-14 01:28:07.998382+00:00",
"db_md5": "0cdede812da1d9957f86c9ea89ed1f0f",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "50df8fc6-9c99-42d5-bbe1-e01037f04b25",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.4945001242904246,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08791666524484754
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.135016612541515,
"example_count": 600,... |
510b8383-9bb7-4908-95ad-c59b80cf83f7 | easy | MarcoF1 | 2026-08-13 19:14:34.986734+00:00 | succeeded | e52768dc62eae6be73187a88fca093f1e82a7ec55d4b621b337a92d0249b1804 | 13,979 | null | """Residue-clock model for fixed-N generalization.
Representation: parsed integer values of x and N feed (a) exact residue-phase
features over a bank of small integer periods, (b) learned double-hashed value
embeddings (per-residue identity), and T feeds cyclic phase features
(T mod 2..48) plus saturating channels — p... | {
"id": "510b8383-9bb7-4908-95ad-c59b80cf83f7",
"created_at": "2026-08-13 19:14:34.986734+00:00",
"db_md5": "5e1f0026fa1c014dc8b5d95c25314086",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "e2687875-76cb-470c-b675-267b61190ad1",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.4861901029387017,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.053333332439263664
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.503406047821045,
"example_count": 100... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.