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 |
|---|---|---|---|---|---|---|---|---|---|---|
4c8d9bb6-0129-473a-b3e1-f99151fd9cd2 | easy | benjaminW2025 | 2026-08-19 06:06:12.328625+00:00 | succeeded | 100ef7f9a44b81ba8fb64985bc5eb2c6a56acf18fed285a66c2bd4157587822c | 13,369 | null | """Phase 0 S2x4: two scratchpad blocks shared across four repeats."""
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 = 256
N... | {
"id": "4c8d9bb6-0129-473a-b3e1-f99151fd9cd2",
"created_at": "2026-08-19 06:06:12.328625+00:00",
"db_md5": "e4d36cc144d887516bdf92c3f30d5e9d",
"submitter": "benjawesome",
"github_login": "benjaminW2025",
"run_id": "1107f8a1-d4e2-4f24-b797-2a1827da6bde",
"tier": "easy",
"dataset_id": "e5",
"status": "... | {
"score": {
"mean_loss": 2.281430209791105,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2936530498111196,
"example_count": 600,
... |
4c8fd49c-c62d-47ed-a188-f06f56c1b555 | easy | DDanlov | 2026-08-08 09:55:19.133701+00:00 | succeeded | ce2a14ca6aecd6db3e18f83e588b4973a8d80b6f4c8527477a550210c61db665 | 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": "4c8fd49c-c62d-47ed-a188-f06f56c1b555",
"created_at": "2026-08-08 09:55:19.133701+00:00",
"db_md5": "d79674621788e71437c9495207700ad2",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "0aee913d-f21b-4808-8460-bfa375493762",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 89.01550838061402,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.028333331843217215
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 98.76985168457031,
"example_count": 100,... |
4c918873-c598-4dc6-8ad6-0a24767986f5 | easy | DDanlov | 2026-08-21 01:55:06.703640+00:00 | succeeded | e7f2b7b095f9a044416860aea1425e67858ef2c05406ca10a37e068e27dd34f2 | 25,928 | null | from __future__ import annotations
import math
import time
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
except Imp... | {
"id": "4c918873-c598-4dc6-8ad6-0a24767986f5",
"created_at": "2026-08-21 01:55:06.703640+00:00",
"db_md5": "cf5ce82836af974ed3a9c5b7d7988d4a",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "f2bde973-b397-4ded-ae80-2ab667a773cb",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.6415552394977437,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01041666670391957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.64201050582488,
"example_count": 600,
... |
4c931dfb-c797-455f-97ab-4d946518ee79 | easy | TheYkk | 2026-08-04 12:41:29.237383+00:00 | succeeded | 7316f9de96dda6b6d47a9bd85ae32087013f84a669512b472060a8ca4c1ad21f | 12,192 | null | """Encoder / iterated-operator / positional-decoder composition model.
Why this shape
--------------
The label is a composition ``S^T(x)``: one unknown step map applied a
prompt-dependent number of times. A transformer stack -- looped or not --
spreads that map across per-layer parameters and per-position activations... | {
"id": "4c931dfb-c797-455f-97ab-4d946518ee79",
"created_at": "2026-08-04 12:41:29.237383+00:00",
"db_md5": "88cc0493ed614d93f98e112f2e56e829",
"submitter": "Kaan Karakaya",
"github_login": "TheYkk",
"run_id": "c9c0fc81-de0e-4eb6-9509-6cfaca52f80c",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.8425952792167664,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07999999821186066
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.7888495922088623,
"example_count": 100... |
4c932dde-b6b8-4091-a830-e19ca078da1b | easy | sapient-sapiens | 2026-08-30 21:04:56.972579+00:00 | succeeded | b32dda5ae8086e1cdbf0db9d5703f7babef14a933628dc6c7613489839a57b8f | 20,395 | null | """Standalone nested inner/outer recurrence candidate.
Representation is categorical only: ``E_token + E_group + E_leftrel``. No
field is decoded into a number, no field value sets depth, halting, or any
mixing weight, and the only supervised target is the final answer.
Structure
---------
One prompt encoder produce... | {
"id": "4c932dde-b6b8-4091-a830-e19ca078da1b",
"created_at": "2026-08-30 21:04:56.972579+00:00",
"db_md5": "2f429ac1e30fc2ebf70ab2bf057a64ff",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "5bf8e32e-cac0-4eae-a57b-e3a260eee7ab",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1412591689969056,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01208333338300387
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1418428976974146,
"example_count": 600... |
4c9f33fb-066e-491d-8208-53d17020226c | easy | velocizapkar | 2026-08-06 21:58:36.610939+00:00 | succeeded | 083a18c7e370dfd6494d31ce5f1ceeca370b2e08acc9b65b4c957205831dfd75 | 53,087 | 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": "4c9f33fb-066e-491d-8208-53d17020226c",
"created_at": "2026-08-06 21:58:36.610939+00:00",
"db_md5": "a1e7a5600a8387dbec2aac2b08d1cdd6",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "e983819b-73e2-471d-aa68-31c532bc0518",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 2.511188507080078,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010624999646097422
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.60249662399292,
"example_count": 800,
... |
4cae4f9d-03f6-4fc5-a88f-32303e947b60 | easy | KaustubhKumar05 | 2026-08-27 17:03:50.567573+00:00 | succeeded | 28f00ad9ad52cc411a7ea62c8cff713373e3778f7b1768a7a1e10198ea2b0fac | 20,492 | null | """cell_best -- the three arms that MEASURED positive, combined. No optimizer bundle.
Selected on evidence, not on enthusiasm:
cell_fix (architecture) local REG-2 72.3% train exact vs 54.3% base, 61% faster
cell_aux (carry aux) easy/e3 testX 2.50% / oodX 2.00% vs base 1.50% / 1.30%,
... | {
"id": "4cae4f9d-03f6-4fc5-a88f-32303e947b60",
"created_at": "2026-08-27 17:03:50.567573+00:00",
"db_md5": "186d0c0910371524962d40d18f5c562d",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "ed2c4508-0f91-4c6d-aa43-02378107916a",
"tier": "easy",
"dataset_id": "e3",
"status": "succee... | {
"score": {
"mean_loss": 6.066466367461455,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018750000000000003
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.522867577287596,
"example_count": 800,... |
4cafb560-2bb4-45c3-8956-4d9cb5d603ed | easy | jordanrubin | 2026-08-08 10:49:33.704571+00:00 | succeeded | 19e434c7e9fc9d9d601292a0f5c0a4083813bc7790caf87c0582fc1bbcc2cad8 | 16,909 | 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": "4cafb560-2bb4-45c3-8956-4d9cb5d603ed",
"created_at": "2026-08-08 10:49:33.704571+00:00",
"db_md5": "671c15c2db39c821999bae96a7defa51",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "4bae7e75-8d97-4e9f-aa32-247d12e5f2bc",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 5.146107461998926,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012499999565382797
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.1789701622430515,
"example_count": 600... |
4cb155e0-4506-475d-a2f6-5eb72de98302 | easy | yashkant | 2026-08-09 02:56:19.994093+00:00 | succeeded | d5fde88fe21de21e443530ce520067a660084202845f6e63cc3e645dcbef5720 | 22,836 | null | """Round 211 common TRM-style x/y/z shared-transition template."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
from torch import... | {
"id": "4cb155e0-4506-475d-a2f6-5eb72de98302",
"created_at": "2026-08-09 02:56:19.994093+00:00",
"db_md5": "c941879f160765469a08cd6840b8b992",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "75224417-12ce-47d3-a8f2-ef7210b27d74",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 2.1164562427966915,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013750000027939677
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1143677541574712,
"example_count": 80... |
4cb8733c-5a0c-4555-b499-f8fcd98b7bd5 | easy | SunstarKK | 2026-08-18 13:32:02.924549+00:00 | succeeded | 31557d6d02110d96040ececc14b7c28cdd0798b9cb3a734a44b425feb1239abb | 14,929 | null | """Heterogeneous two-dimensional neural workspace for One Layer Deeper."""
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_s... | {
"id": "4cb8733c-5a0c-4555-b499-f8fcd98b7bd5",
"created_at": "2026-08-18 13:32:02.924549+00:00",
"db_md5": "aeabfccb834a656f75086fa7ec5f3fdb",
"submitter": "SunstarK",
"github_login": "SunstarKK",
"run_id": "c226552d-1fdb-4c1f-82ee-c4936919068c",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.2301700683394534,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00499999995653828
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2602527141571045,
"example_count": 600... |
4cbd8f11-5481-4554-993f-480bd7c14b4f | easy | jordanrubin | 2026-08-03 21:26:30.230347+00:00 | succeeded | 09586f90ffa296e3bdbb25e6a0b059c7579344900d2f7d9eb12b0e902c4848f6 | 15,390 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "4cbd8f11-5481-4554-993f-480bd7c14b4f",
"created_at": "2026-08-03 21:26:30.230347+00:00",
"db_md5": "ef7d67d872a65aa96e81ca53991ab27d",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "8815bd0b-5bcd-4bfc-b09a-ecd4e653ff2a",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 3.4377510698294524,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.28312499277914566
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.80864444814631,
"example_count": 300,
... |
4cc4d2aa-fb38-4367-a0ac-1161d76be2c6 | easy | khushidahi | 2026-08-07 01:13:33.301022+00:00 | succeeded | 27f229fb50cb260fdf201062e832f834fa6f75302c79302281008d5c2c3563e1 | 14,691 | null | """R6 structured decimal-workspace model for One Layer Deeper.
The model does not implement multiplication, modular reduction, or the public
recurrence. It uses the public tokenizer structure to build learned, right-
aligned decimal tapes for N, X, and T, then applies a shared neural transition
to a mutable work tape.... | {
"id": "4cc4d2aa-fb38-4367-a0ac-1161d76be2c6",
"created_at": "2026-08-07 01:13:33.301022+00:00",
"db_md5": "1e853bfe193dffed358f1d7f481ad6c7",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "687bd26c-1dc7-441a-96bf-064a6fb77538",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.0778331756591797,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005104166688397527
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.076585531234741,
"example_count": 120... |
4cc86e5d-85e9-4197-83ee-8789b7968a28 | easy | shirvani-jr | 2026-08-29 00:42:21.491493+00:00 | succeeded | d068b434ab864217b994258f5faffd084a089175e7dee45a0f24d694a72fe335 | 49,680 | 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": "4cc86e5d-85e9-4197-83ee-8789b7968a28",
"created_at": "2026-08-29 00:42:21.491493+00:00",
"db_md5": "2ca3459d14de74e97b99760b922a8174",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "72637e15-ab7c-42d5-9a26-7850a20efb66",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.607854723930359,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4033333510160446
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.085686206817627,
"example_count": 100,
... |
4cc903e0-bdbb-43c3-974c-b6e89e6643ce | easy | rteehas | 2026-08-19 08:28:15.612128+00:00 | succeeded | 7c89da12c3e257a53f9f75a86eaf44cab870da3af4f027de65b2b1a6259721d3 | 41,409 | null | """Anonymous marginal-program submission.
Mechanism
---------
The model is a differentiable probabilistic program. Its forward pass
computes, exactly, the answer-token predictions of a small structured latent
hypothesis space:
(prompt segmentation = which token types head the three fields)
x (field-role permutat... | {
"id": "4cc903e0-bdbb-43c3-974c-b6e89e6643ce",
"created_at": "2026-08-19 08:28:15.612128+00:00",
"db_md5": "09ac2233c31b4fc6e188df5974579674",
"submitter": "Ryan Teehan",
"github_login": "rteehas",
"run_id": "f5baafe4-77e1-4601-98a5-e7154fc8925e",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 7.2154341720666775,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0050000000248352685
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.229568301829522,
"example_count": 60... |
4ccf0693-3ac0-4ce0-8344-0ab7875c3fea | easy | pculaf | 2026-08-20 13:57:46.635535+00:00 | succeeded | ff4d191eac9cde946fea4361b32f1f60ed496120eb51a75925db00f59889a050 | 13,575 | null | """Muon recurrent Transformer with arithmetic-aware positional embeddings."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submissio... | {
"id": "4ccf0693-3ac0-4ce0-8344-0ab7875c3fea",
"created_at": "2026-08-20 13:57:46.635535+00:00",
"db_md5": "3522b2b3145642b7aa56065ffd33991a",
"submitter": "Pavle Culafic",
"github_login": "pculaf",
"run_id": "720be111-3c08-4bd2-a68b-9a8476057a86",
"tier": "easy",
"dataset_id": "e6",
"status": "succe... | {
"score": {
"mean_loss": 2.195988714694977,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.031981983222067356
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.464113473892212,
"example_count": 60,
... |
4cd1634d-febd-4043-8758-50379640ce31 | easy | shan23chen | 2026-08-08 02:45:16.232810+00:00 | succeeded | 78ca8e5d6066f76f50aec1313b2c8f2efbec8fe58ed743244822b66dc3237607 | 24,961 | null | """Boundary-reset recurrent register Transformer.
One learned transition is applied once per requested outer step. The
transition receives an immutable decimal N register and the current learned
boundary state, but never receives T or a separate copy of the original X.
Within each outer application, scratch persists ... | {
"id": "4cd1634d-febd-4043-8758-50379640ce31",
"created_at": "2026-08-08 02:45:16.232810+00:00",
"db_md5": "57c6213eda41c31c213c0ac18948a5dd",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "ac24fabc-99f7-483f-9aec-d72371894b55",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.7898993492126465,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166667005047202
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.711155414581299,
"example_count": 600... |
4cd461d0-56f4-4ec3-8e0e-38be06776ad7 | easy | lpbb | 2026-08-13 12:11:36.230692+00:00 | succeeded | e41375cc4836decb2da4ecae72cf7ff8a1c2a44235cde28c6f18a4f4e3e74b1e | 5,813 | null | """S2 — S1 plus deep supervision: every loop iteration is graded on the final answer."""
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,
... | {
"id": "4cd461d0-56f4-4ec3-8e0e-38be06776ad7",
"created_at": "2026-08-13 12:11:36.230692+00:00",
"db_md5": "639fb41d6c84fe60db8ae665a748292c",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "08b70c4f-b1d9-4172-9125-d6f847804aa5",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 4.30188245183812,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000004656613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.409739272053822,
"example_count": 800,
... |
4cd9e312-520e-452a-8341-7bdfa4a40020 | easy | richardcepka | 2026-08-13 20:02:56.035665+00:00 | failed | 561628654b944b90133547fb591b400763f6de2d6d729263430135f1505ec043 | 23,100 | null | """Recurrent digit-register model for One Layer Deeper.
A looped transformer whose recurrent state is a decimal digit register that is
re-quantised through a ten-entry codebook on every iteration. Nothing inside
the loop depends on the iteration index and no parameter is indexed by an
absolute slot position, so the e... | {
"id": "4cd9e312-520e-452a-8341-7bdfa4a40020",
"created_at": "2026-08-13 20:02:56.035665+00:00",
"db_md5": "4ff4f050499b10e4218c189dfad7cc95",
"submitter": "Richard Cepka",
"github_login": "richardcepka",
"run_id": "088a88ac-96a4-497b-83ed-4e71d3185ba3",
"tier": "easy",
"dataset_id": "e5",
"status": ... | null
|
4cf5a210-8f5e-4e1e-97b8-0c53f17453af | easy | sapient-sapiens | 2026-08-20 17:21:59.709673+00:00 | succeeded | c8ac0e2bc843654b0c5a032f814997e31db4a05dd376c9f080511d36d2d02940 | 44,085 | null | """Persistent pair-indexed relational state around a categorical operator.
The recurrent answer state remains a bank of right-aligned categorical digit
slots. A separate continuous S-by-S grid retains every ordered digit-pair
cell across applications. Learned bilinear bindings and categorical modulus
context update ... | {
"id": "4cf5a210-8f5e-4e1e-97b8-0c53f17453af",
"created_at": "2026-08-20 17:21:59.709673+00:00",
"db_md5": "081d657b94072d9174107b50bbc10a1c",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "8930e616-89f7-4664-9d61-8f2ca16ea3c8",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.7054020884729626,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0020833333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7065744047207683,
"example_count": 6... |
4d0432f6-fc1c-455f-9b93-875bbe370b9b | easy | pculaf | 2026-08-15 17:54:57.643249+00:00 | succeeded | 031ae0f8af8ffe9fdd5217b71dae05ce39f293ae6c2a0c3b938be669a66e9c3c | 8,356 | null | """Muon-optimized recurrent Transformer for the E6 activation/depth sweep."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submissio... | {
"id": "4d0432f6-fc1c-455f-9b93-875bbe370b9b",
"created_at": "2026-08-15 17:54:57.643249+00:00",
"db_md5": "251795e8978a684f14c8f4590a0b864b",
"submitter": "Pavle Culafic",
"github_login": "pculaf",
"run_id": "d7f5bace-076b-4c0d-b5ad-3e3fedd841f1",
"tier": "easy",
"dataset_id": "e6",
"status": "succe... | {
"score": {
"mean_loss": 2.815073609352112,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.021846847608685493
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.6477198600769043,
"example_count": 60,... |
4d072956-05f8-4a99-9158-8a356003d6a8 | easy | DESU-CLUB | 2026-08-24 10:40:39.805681+00:00 | succeeded | a9739e0c701d51475803ce2ea8ec103730b317792a27a070c783defc3cb6f11e | 10,579 | null | """Huginn-style looped Transformer whose iteration count is read from the T token.
prelude -> [core applied T times, input re-injected each step] -> coda
Forked from experiments/_baselines/adamw_transformer.py; the Block, RMSNorm and
optimizer are unchanged so that the only varied axis is the depth mechanism.
"""
fr... | {
"id": "4d072956-05f8-4a99-9158-8a356003d6a8",
"created_at": "2026-08-24 10:40:39.805681+00:00",
"db_md5": "6a406b2f02963938a85eb069acfdc56b",
"submitter": "Low Keng Hoong, Warren",
"github_login": "DESU-CLUB",
"run_id": "b173125f-dead-45ba-a37c-6404f81ade54",
"tier": "easy",
"dataset_id": "e6",
"sta... | {
"score": {
"mean_loss": 11.044672515294323,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07342342690036104
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.152322769165039,
"example_count": 60,
... |
4d1d5cd5-a080-4c62-bdc6-1609b08fb165 | easy | himalalps | 2026-08-09 01:59:42.406888+00:00 | succeeded | 2c1687908cb5c318ec1185997c5ffe3fc8fe0d815596106cf82abe19caf7ebfa | 70,605 | 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": "4d1d5cd5-a080-4c62-bdc6-1609b08fb165",
"created_at": "2026-08-09 01:59:42.406888+00:00",
"db_md5": "f188d6b1cb9440880dce3fdfd8caa5a7",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "af64f4ce-42c1-4652-986c-3150ebd891cc",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.180005490846999,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011250000161429246
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1749837398529053,
"example_count": 600... |
4d1e7d77-7f85-49c7-8601-17b883703f98 | easy | CodeReclaimers | 2026-08-19 17:31:34.115911+00:00 | succeeded | d5ecd4eb3b40aa10e57be922e5c7e91a5857ff0d0e4270db3f5cb2c9a2ce2797 | 36,181 | null | """v39_keyed: keyed-modulus CRT learner — Run A of T1_CERT_PLAN_20260819.
HYPOTHESIS A (alias modulus): h1's displayed N is a stable per-identity
alias phi(M) of a hidden modulus M <= 2^20 from enumerated factor pairs
(<=10-bit primes), while x and y are displayed plainly and the step closes
per prime channel (squarin... | {
"id": "4d1e7d77-7f85-49c7-8601-17b883703f98",
"created_at": "2026-08-19 17:31:34.115911+00:00",
"db_md5": "87cd360f53ff10d7d3208c12bea17653",
"submitter": "CodeReclaimers",
"github_login": "CodeReclaimers",
"run_id": "296b51c3-f81f-45dd-8883-f116adf1cb99",
"tier": "easy",
"dataset_id": "e2",
"status... | {
"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": 300,
"exact_accuracy": 1.0,
"... |
4d20e1ea-fc3f-4986-819e-66087bfa3e95 | easy | ehonig | 2026-08-21 01:05:57.005635+00:00 | succeeded | a9ad0d4126ab74869f8e7f4109841541ffaa4c10fe840580936c9852aab9df05 | 30,222 | 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": "4d20e1ea-fc3f-4986-819e-66087bfa3e95",
"created_at": "2026-08-21 01:05:57.005635+00:00",
"db_md5": "e6d7ec5bd1fddf2d609c6fc1ca4d18f7",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "3cada688-c05e-49c4-8fd2-4c27191f1c5c",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.1786273999251167,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0025000000124176342
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1833307965216378,
"example_count": 1... |
4d20e25f-d392-433d-868f-b1f738e3ff3a | easy | oupadhyay | 2026-08-27 08:21:16.512186+00:00 | succeeded | d0ed0e69a16c0b6cf6a9656354534b9fca339da7a8509b54675c4eb75ae30cae | 3,719 | null | """Persistent globally mixed polynomial recurrence over a whole-number latent state."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle,Submission,TokenLossBatch,assert_model_state
W,D,R=4,32,64
class C:
def __init__(s,vocab_siz... | {
"id": "4d20e25f-d392-433d-868f-b1f738e3ff3a",
"created_at": "2026-08-27 08:21:16.512186+00:00",
"db_md5": "07d8b96070916400c9d0d193e2efd099",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "c14f253f-5fd1-453e-8723-407ee06db410",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 3.0954388164956645,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0054166666790843015
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1827117487988663,
"example_count": 6... |
4d265d19-b0a3-4228-9a37-99475ee0d10e | easy | yashkant | 2026-08-13 00:03:29.596695+00:00 | succeeded | 03b304d52edd6ab0187de662ff3eecd25efca4637bc588d916c6022f98bb2af0 | 33,286 | null | """Round1786 evolutionary crossover of two controlled survivors.
The Round1780 W8/D2 carrier keeps its winning contrastive, exactly
antisymmetric learned set codes. The candidate additionally removes each
initial and recurrent Dykstra correction residual's coordinate mean before
the unchanged tanh bound, combining Ro... | {
"id": "4d265d19-b0a3-4228-9a37-99475ee0d10e",
"created_at": "2026-08-13 00:03:29.596695+00:00",
"db_md5": "267de64e1cd8a82c539c970185491b70",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "280fa17e-deb4-4353-b3bf-e4d8fc3165f6",
"tier": "easy",
"dataset_id": "e4",
"status": "succeed... | {
"score": {
"mean_loss": 2.073971961243717,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0767836903955548,
"example_count": 120... |
4d2f29a8-6e0f-42d6-b4f3-4b483eb3956c | easy | karanganesan | 2026-08-04 02:12:03.662468+00:00 | succeeded | bcc540a27eaa99769ad1480decaa17cf11ee45810affcd67d4bb69ddc6e0d663 | 11,980 | 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": "4d2f29a8-6e0f-42d6-b4f3-4b483eb3956c",
"created_at": "2026-08-04 02:12:03.662468+00:00",
"db_md5": "36b972d532dc5e5e79b296b225cb000b",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "f9eae609-f143-4883-a497-4d7f1d2012a7",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.6853920221328735,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333563059568
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6263110637664795,
"example_count": 60... |
4d300f63-5000-4995-af30-c9735030d357 | easy | khushidahi | 2026-08-02 22:42:41.223599+00:00 | succeeded | df9cdd60c8a849a4e458b59d07dd78f6f4110fea4d3fb8ac3c8678d5eb75baf6 | 8,854 | null | """R4 exactness-aligned-loss Easy E5 probe.
The model, optimizer, stochastic recurrent-depth curriculum, evaluation depth,
batches, and 200-step checkpoint match the R3 Easy control. The only change is
a boundary-preserving TokenLossBatch objective that blends per-example mean
cross-entropy with a smooth maximum over ... | {
"id": "4d300f63-5000-4995-af30-c9735030d357",
"created_at": "2026-08-02 22:42:41.223599+00:00",
"db_md5": "bc7042baf6e26265325fd0cf796d3b7c",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "46d6ce9f-a544-40af-a2a0-26138c0a803a",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.2286049127578735,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333656191826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2281880378723145,
"example_count": 30... |
4d31cb1d-da9b-4839-a7f5-f8bf53642a5c | easy | Dandandan | 2026-08-05 16:53:40.804100+00:00 | failed | 4c41602e544c99437598bb855910fdba6c820ff925f5e1c9e952f71283183637 | 32,995 | null | """End-to-end recursive function learner for One Layer Deeper.
A population of randomly initialized polynomial register networks is reused
exactly once per requested outer step. The evaluator trains every coefficient
and the evidence selector through its normal forward/backward/optimizer loop.
There is no recurrence-... | {
"id": "4d31cb1d-da9b-4839-a7f5-f8bf53642a5c",
"created_at": "2026-08-05 16:53:40.804100+00:00",
"db_md5": "03cf51f406709145b345c05d167ba3ef",
"submitter": "Daniël Heres",
"github_login": "Dandandan",
"run_id": "d2db425f-491a-4075-918b-4d54927f2639",
"tier": "easy",
"dataset_id": "e5",
"status": "fai... | null
|
4d377bfa-77ed-4d2f-bc27-d080b3ed4ec5 | easy | khushidahi | 2026-08-20 04:28:36.703326+00:00 | succeeded | 48748e021868249bec029233a97f8df1a93505c5ce365ee6d83fde36672c38b6 | 37,808 | null | """R310: hard-state anti-collapse endpoint-product functional root.
This candidate is deliberately diagnostic. It trains on evaluator-owned final
labels at several depths so that one shared transition must compose. The model
never converts N or X to native numbers and never implements multiplication or
modular reduc... | {
"id": "4d377bfa-77ed-4d2f-bc27-d080b3ed4ec5",
"created_at": "2026-08-20 04:28:36.703326+00:00",
"db_md5": "6364a1e99738219744477b244dd32cc5",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "1dca2076-4414-4808-9147-adcfa8d70bea",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 2.2212698692553596,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005625000013969839
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.208535491392098,
"example_count": 800... |
4d48ca70-d329-41ca-ba68-cc2d30d2a8b5 | easy | briheet | 2026-08-07 08:46:51.629538+00:00 | succeeded | 4c0538b8e75354d4bacdf4dc9c69f58f59dcdb72e3b93ff3fec976f9d857ffa6 | 8,232 | null | """Generic looped Transformer with AdamW optimization.
The model learns one shared transition and applies it as many times as requested
by T. The embedded prompt is reinjected at every loop, while decimal-place
embeddings provide numerical position information without implementing the task.
"""
from __future__ import... | {
"id": "4d48ca70-d329-41ca-ba68-cc2d30d2a8b5",
"created_at": "2026-08-07 08:46:51.629538+00:00",
"db_md5": "00e37622a7dbc32d8f4b9b9d6ee6435c",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "e1967b1e-b1a8-4f4d-a49a-162f8495ac6b",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.2854292645634473,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.36687568484935,
"example_count": 600,... |
4d49dbb7-f60b-4d54-8022-a98fe10490b4 | easy | khushidahi | 2026-08-07 02:25:04.502577+00:00 | succeeded | 3730b7e97c72ec22b823cc4d7ea0be2fa61bb57b2291977fb24c4218c61d5ebf | 18,882 | null | """R6 structured decimal-workspace model for One Layer Deeper.
The model does not implement multiplication, modular reduction, or the public
recurrence. It uses the public tokenizer structure to build learned, right-
aligned decimal tapes for N, X, and T, then applies a shared neural transition
to a mutable work tape.... | {
"id": "4d49dbb7-f60b-4d54-8022-a98fe10490b4",
"created_at": "2026-08-07 02:25:04.502577+00:00",
"db_md5": "d3a91250e57d79de985dc91c9e498215",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "d3f90ab8-158a-4e23-bfbb-0664cd28fd8e",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.077609062194824,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008124999701976776
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.075892686843872,
"example_count": 1200... |
4d555d61-021c-48c7-bb2b-76baac80137a | easy | DDanlov | 2026-08-28 21:51:46.967408+00:00 | succeeded | cd13d6c2e886f1de8aba9d6a29a0b011fa27143e8469e63bc03bc8850b645183 | 14,934 | null | """
Trial 20: Model B: D=512, H=8, M=4, u=2 (T=8)
"""
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, To... | {
"id": "4d555d61-021c-48c7-bb2b-76baac80137a",
"created_at": "2026-08-28 21:51:46.967408+00:00",
"db_md5": "26ce730a1c1872dc698944023be59ebc",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "a8c1608f-78c0-4210-966c-46d8fab7225f",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.8332028262303934,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012916666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.833203054062454,
"example_count": 600... |
4d5ae65c-f679-4f43-9024-0e99e1c3ce54 | easy | khushidahi | 2026-08-03 19:38:29.030362+00:00 | failed | a45cdada2873dbe24bcbc18d857702e22be5ce50d333828c16cbd939f48d3b6e | 14,360 | null | """R6 structured decimal-workspace model for One Layer Deeper.
The model does not implement multiplication, modular reduction, or the public
recurrence. It uses the public tokenizer structure to build learned, right-
aligned decimal tapes for N, X, and T, then applies a shared neural transition
to a mutable work tape.... | {
"id": "4d5ae65c-f679-4f43-9024-0e99e1c3ce54",
"created_at": "2026-08-03 19:38:29.030362+00:00",
"db_md5": "6dd2fd9ac880c1d888b3edd4551df7db",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "5a4669f4-9054-4ec2-8708-ea460c6c58e4",
"tier": "easy",
"dataset_id": "e5",
"status": "fail... | null
|
4d5c87be-a36a-435c-b5f9-23a4629feade | easy | chad-atexpedient | 2026-08-10 02:09:24.405976+00:00 | succeeded | 5409f0c89e308948d08a624fcd831c244f8a04b41ab56de89761eb7123f58450 | 10,938 | null | """BL8D: width-64 trajectory readout."""
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
IMMUTABLE_LANES ... | {
"id": "4d5c87be-a36a-435c-b5f9-23a4629feade",
"created_at": "2026-08-10 02:09:24.405976+00:00",
"db_md5": "45a1cf0456a33b75ca2fe955bb0b2514",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "728f73b0-8fea-49e9-96b2-aba16823c908",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 6.061675279220983,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009583333383003871
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.126241611258331,
"example_count": 600,... |
4d5e3f65-0291-4a69-b86c-7d513dd7450c | easy | repst | 2026-08-09 21:33:01.621948+00:00 | succeeded | b75739852224db0ac11e6a820d3d96eb38a52c4f483c0d9680860ce31f3b1f6e | 23,176 | null | """Strict forced-T operator-learning diagnostic for the public prompt contract."""
from __future__ import annotations
from typing import NamedTuple
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,... | {
"id": "4d5e3f65-0291-4a69-b86c-7d513dd7450c",
"created_at": "2026-08-09 21:33:01.621948+00:00",
"db_md5": "f9a3a87882277ce8d7e915d20ec36681",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "03b8d3a3-a94a-4e82-8a50-96ece3b238fa",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 7.918519020080566,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.904993057250977,
"example_count": 100,
... |
4d5ec8fa-3077-4714-850f-439646628c92 | easy | chuk-uzowihe | 2026-08-07 14:39:46.584868+00:00 | succeeded | ece4cc05d3fc151834dde92e4ea8716431dce71e9a58aa62a680de2af4a0b331 | 27,181 | null | """One Layer Deeper submission: FPRM over a compressed (N, x, T) lattice.
v6 = v2's architecture (the only version that has generalized on m1: 15.18%
test, 11.6% at unseen T=32) rebuilt on the v5 chassis, with its one fatal
flaw fixed: input numbers are no longer parsed into int64 (silent garbage
past 19 digits) — ins... | {
"id": "4d5ec8fa-3077-4714-850f-439646628c92",
"created_at": "2026-08-07 14:39:46.584868+00:00",
"db_md5": "f7ac740701518dd9e1cbcd3cb8761a1e",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "17d7ee4e-1ba1-4f9f-93ae-6af4b5de38ea",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 41.98383712768555,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.053333332762122154
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 42.44749450683594,
"example_count": 100,... |
4d629974-3c55-4f9b-ac25-7ec5719089af | easy | DDanlov | 2026-08-31 21:21:32.498752+00:00 | succeeded | 0b79870cec9e7f76c2fa35e565e4bfac2b4267887017ccf955f7b4baf55808ad | 32,408 | 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": "4d629974-3c55-4f9b-ac25-7ec5719089af",
"created_at": "2026-08-31 21:21:32.498752+00:00",
"db_md5": "5e80ca5b45e1feda5737c1aa115455a9",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "b76759c4-1b65-4558-9797-9f098c63b723",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.1780184230002804,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.163820342050664,
"example_count": 800,
"... |
4d633fb7-0672-43df-a7c4-d3b45270e4c9 | easy | andreamiele | 2026-08-06 09:11:03.692064+00:00 | succeeded | 2bc3e921e98926068cc401f53c9f943667e4ecd0b6a469391bb7ec6caa32a172 | 22,805 | null | """Loop model with recall, composition consistency, and deep supervision.
Motivated by what the easy-to-hard / latent-reasoning literature converges on
and which earlier attempts here lacked:
* **Recall (input injection).** Deep Thinking (Schwarzschild et al. 2021;
Bansal et al. 2022) found iterating past the trai... | {
"id": "4d633fb7-0672-43df-a7c4-d3b45270e4c9",
"created_at": "2026-08-06 09:11:03.692064+00:00",
"db_md5": "db746210002bc7a2289cf13b9312e6bf",
"submitter": "Andrea Miele",
"github_login": "andreamiele",
"run_id": "f1427a8d-5b1f-4118-b95f-f7d24cc14f26",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 1.5556217810521682,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.7566666853427887
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.5951935479890055,
"example_count": 100,... |
4d6f1d3a-60cd-47f3-bacf-1b5803e80366 | easy | pculaf | 2026-08-24 14:14:04.235002+00:00 | succeeded | 44d66f5d252756706b6e21c572ce84f94df1ab6ed925254a676956dd35236b82 | 7,772 | null | """Small single-pass post-RMSNorm Transformer for One Layer Deeper."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
... | {
"id": "4d6f1d3a-60cd-47f3-bacf-1b5803e80366",
"created_at": "2026-08-24 14:14:04.235002+00:00",
"db_md5": "5c8e2fd6824a01aceb46922e13736bed",
"submitter": "Pavle Culafic",
"github_login": "pculaf",
"run_id": "85a350c2-6ef0-4465-890e-e3d8de7f0cce",
"tier": "easy",
"dataset_id": "e7",
"status": "succe... | {
"score": {
"mean_loss": 2.0070090889930725,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03827985841780901
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9971429109573364,
"example_count": 85,... |
4d727872-389c-4a33-8f53-12cf7e6e1c26 | easy | erdavis0 | 2026-08-22 08:02:02.051666+00:00 | succeeded | a99223541e18c93fde96cfe78ef10d8221dc373367857a0a6be839c20f16f1a1 | 4,791 | null | """Compact local cellular carry transducer."""
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 = 48
HIDDEN = 96
ROUNDS = 6
cl... | {
"id": "4d727872-389c-4a33-8f53-12cf7e6e1c26",
"created_at": "2026-08-22 08:02:02.051666+00:00",
"db_md5": "71a6dd89ba0639e5dc2d991c8e8d02f8",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "f654ebda-ea47-4c65-aa81-1a9c2c21d383",
"tier": "easy",
"dataset_id": "e8",
"status": "succeeded",... | {
"score": {
"mean_loss": 4.425987005233765,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05672905594110489
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.9566383361816406,
"example_count": 85,
... |
4d78bceb-a38f-4fb8-a38b-bdd4cda96444 | easy | epicosy | 2026-08-26 21:20:29.668336+00:00 | succeeded | a68f89f1f39252e791e1300ce6b10079e11f2a6262e71ec80dfe9f936d11d85e | 7,436 | null | """Latent-state Universal Transformer with position-aware output decoding.
Fixes the v1 bottleneck: instead of broadcasting one pooled latent vector to
every output position, each input position now cross-attends to the final
latent state to produce its own logits.
"""
from __future__ import annotations
import torch... | {
"id": "4d78bceb-a38f-4fb8-a38b-bdd4cda96444",
"created_at": "2026-08-26 21:20:29.668336+00:00",
"db_md5": "e555de595e223b7b414696cbcc4256e1",
"submitter": "Eduard Pinconschi",
"github_login": "epicosy",
"run_id": "caded317-5d3e-48b4-9dad-37f1a7a55a23",
"tier": "easy",
"dataset_id": "e6",
"status": "... | {
"score": {
"mean_loss": 5.575921297073364,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07680180575698614
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.7683868408203125,
"example_count": 60,
... |
4d7965b9-bd0d-4038-8152-3e90aa09d3ff | easy | khushidahi | 2026-08-07 04:33:01.032734+00:00 | succeeded | b658c2fe1da81bbc5780fc0f1f2e94f2f5960f3e6b3ee886a6c66829e309d048 | 16,966 | null | """R6 structured decimal-workspace model for One Layer Deeper.
The model does not implement multiplication, modular reduction, or the public
recurrence. It uses the public tokenizer structure to build learned, right-
aligned decimal tapes for N, X, and T, then applies a shared neural transition
to a mutable work tape.... | {
"id": "4d7965b9-bd0d-4038-8152-3e90aa09d3ff",
"created_at": "2026-08-07 04:33:01.032734+00:00",
"db_md5": "c10a51c40c23254a16eaf3c84980733b",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "ab61e1ce-4f24-4c6c-b50c-02b6b0af6490",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.19686496257782,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666772216558
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1961379051208496,
"example_count": 600,... |
4d7ff80c-510c-4551-b068-1c1f99bab330 | easy | khushidahi | 2026-08-06 02:51:13.025325+00:00 | succeeded | 45f4f1218d24e36293de9570a60f9aac5d21b3688ab87464a87b284a654519f2 | 14,389 | null | """R6 structured decimal-workspace model for One Layer Deeper.
The model does not implement multiplication, modular reduction, or the public
recurrence. It uses the public tokenizer structure to build learned, right-
aligned decimal tapes for N, X, and T, then applies a shared neural transition
to a mutable work tape.... | {
"id": "4d7ff80c-510c-4551-b068-1c1f99bab330",
"created_at": "2026-08-06 02:51:13.025325+00:00",
"db_md5": "74ee1054f8129ce85ca9d8919eb3c5a0",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "5d87d03e-db21-4837-a928-3e6ddc6a8312",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.0859004259109497,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0071875001303851604
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.085207462310791,
"example_count": 12... |
4d836958-754b-4728-abc1-b2ec50a7e29d | easy | sapient-sapiens | 2026-08-30 21:06:56.173903+00:00 | succeeded | 25ea405b81e6a0d929b387313f8e7a6f0c693945d9bfafae3465bb5db86f6ab6 | 20,535 | null | """Clean eligible Loop3 baseline: immutable N, square-local X, answer A, scratch S.
There is one literally shared pre-norm Transformer block. Within every
Horner position the scratch workspace is reset, then S is updated before A.
Only A is decoded. S is latent and has no decoder, target, or auxiliary loss.
The eli... | {
"id": "4d836958-754b-4728-abc1-b2ec50a7e29d",
"created_at": "2026-08-30 21:06:56.173903+00:00",
"db_md5": "8de6e8e4f95f1b707a5cef3d65e934f1",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "773e908b-b60f-4876-93fa-b34edb996b41",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.4738267529029674,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.462497005815463,
"example_count": 600,
"ex... |
4d8b80a7-09d5-4988-b480-1e233b5081a7 | easy | shan23chen | 2026-08-08 01:58:01.700716+00:00 | succeeded | 2ea64ab2c28284f7978881fc115c51dee9f41c0ac847339181b77b2f4415ba6f | 20,441 | 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": "4d8b80a7-09d5-4988-b480-1e233b5081a7",
"created_at": "2026-08-08 01:58:01.700716+00:00",
"db_md5": "f54b3d399e89d39b6066b7682efae526",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "9354266b-a15c-4127-8ef0-1f9f8f431cad",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.9753390550613403,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009999999776482582
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.9240360260009766,
"example_count": 60... |
4d8dcab9-037f-4e26-899c-6da57a23b16a | easy | khushidahi | 2026-08-23 20:23:36.277976+00:00 | failed | dc30f96367eeccb5fe562f1ff3bdff32abfe024a824930c0ad6f787d71aa41f0 | 11,734 | null | """ART-DT recall candidate.
A learned recurrent sequence model with right-to-left Abacus positions and
global-sequence rotary attention. This variant changes only optimizer LR and
warmup schedule from its architectural parent. This branch ties the two
within-visit layer calls to one TransformerLayer object. One imm... | {
"id": "4d8dcab9-037f-4e26-899c-6da57a23b16a",
"created_at": "2026-08-23 20:23:36.277976+00:00",
"db_md5": "25f8e724d7ec7f02d7c56a4ef3401964",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "e630dc58-89b2-43cd-8cb1-21e6fea165dc",
"tier": "easy",
"dataset_id": "e9",
"status": "fail... | null
|
4d965f05-0f48-4a3f-b1a4-94c31f44c5f3 | easy | erdavis0 | 2026-08-25 02:40:23.703382+00:00 | succeeded | f926cce1aea0a9d136760e9e88480e0e6166b89da0cffdae4dde2b93e05d4730 | 26,413 | null | """Reversible augmented-state transducer over aligned learned digit tapes.
Public delimiter syntax only routes opaque token categories into distinct,
right-aligned N, X, and T tapes. Learned local cells compile separate
interaction, carry, and modulus-conditioned workspaces. N is isolated from
interaction and carry ... | {
"id": "4d965f05-0f48-4a3f-b1a4-94c31f44c5f3",
"created_at": "2026-08-25 02:40:23.703382+00:00",
"db_md5": "95075821638c48bd5050ddf42eabf8c2",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "add1bb9d-6cbe-46ad-87d1-34c636a2f770",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 3.34941029548645,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.1353603694587946
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.0561306476593018,
"example_count": 60,
... |
4d9d94e4-0702-4389-b51d-0ee5e8607e4b | easy | MarcoF1 | 2026-08-13 18:43:10.590549+00:00 | succeeded | d0e4147b03e0296899b8f35cea06d41a127c785d9ab0ec381427944ba8a552e2 | 14,193 | 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": "4d9d94e4-0702-4389-b51d-0ee5e8607e4b",
"created_at": "2026-08-13 18:43:10.590549+00:00",
"db_md5": "0d1955da8d6e70a51ef60bbf2fbfa8e6",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "976455ed-a486-4f87-8012-9a1cebcc6457",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.743892479418519,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0849999988079071
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.350461483001709,
"example_count": 100,
... |
4d9eb008-9474-420b-9643-47ad1d3a08b1 | easy | chuk-uzowihe | 2026-08-06 22:19:36.225739+00:00 | succeeded | 2709b04f4f246ca38a87455672eb1ad19f182d3f5a19c03fcb80533e440f3857 | 27,646 | null | """One Layer Deeper submission: FPRM over a digit lattice with a Fourier T latent.
Structure (v3)
--------------
1. N and x stay as *digit slots*: each prompt token gets its token embedding
plus a learned place-value embedding (offset of the digit from its
number's end), so "different digits live on different sc... | {
"id": "4d9eb008-9474-420b-9643-47ad1d3a08b1",
"created_at": "2026-08-06 22:19:36.225739+00:00",
"db_md5": "85fdc833e822fa7a97ad82410385e81c",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "889caf7d-83be-4caa-aebe-92b0dd5ffb2c",
"tier": "easy",
"dataset_id": "e4",
"status": "... | {
"score": {
"mean_loss": 13.905707452687384,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004375000024835269
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 14.4253778322217,
"example_count": 1200... |
4da6e94a-0ccd-411c-a163-da4a5d91a8f9 | easy | blake-camp-surge | 2026-08-05 18:40:12.984346+00:00 | succeeded | bf70b7b9140201f2e1ad69b90e56e47535bf89b9383b889f76a49c5d429b8e1a | 10,387 | null | """Vector-state neural cellular automaton over attention edges."""
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... | {
"id": "4da6e94a-0ccd-411c-a163-da4a5d91a8f9",
"created_at": "2026-08-05 18:40:12.984346+00:00",
"db_md5": "29e20a618054caf5db81ab15705cfadf",
"submitter": "Blake Camp",
"github_login": "blake-camp-surge",
"run_id": "6d532ba9-7a96-4139-979e-1c96487092b3",
"tier": "easy",
"dataset_id": "e3",
"status":... | {
"score": {
"mean_loss": 3.7163206647755325,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00625
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.7563383636426466,
"example_count": 800,
... |
4da96562-e1d0-4c46-b2d5-15a9312941c6 | easy | DDanlov | 2026-08-08 09:16:46.580664+00:00 | failed | 9486103015fe92d712f383650f27474e89c54c2c0ef3eacb1365af99598f251b | 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": "4da96562-e1d0-4c46-b2d5-15a9312941c6",
"created_at": "2026-08-08 09:16:46.580664+00:00",
"db_md5": "118410231b5c7b5fad1707e2787e19ef",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "9e12e826-caaa-4a5f-8a37-8665275d69d1",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
4db0726a-2501-4fc2-92a7-ba743f95db5a | easy | sapient-sapiens | 2026-08-23 20:39:03.010298+00:00 | failed | 50281034562e83fd1f92b97c13250caf44700b72f462b244927f46841db34f77 | 18,319 | null | """42M Q-gated SwiGLU recurrent Transformer for the E5 campaign.
The arithmetic stream uses categorical token embeddings plus marker-derived
field and left-relative embeddings. N, X, and T are never decoded as numbers.
T-field values are excluded entirely from the arithmetic and halting streams.
All recurrent readout... | {
"id": "4db0726a-2501-4fc2-92a7-ba743f95db5a",
"created_at": "2026-08-23 20:39:03.010298+00:00",
"db_md5": "2e7b9283992424a604293b4b34d67b61",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "10bb9176-5d74-4c47-adf5-95c1d41b596f",
"tier": "easy",
"dataset_id": "e5",
"status": "fa... | null
|
4db0f75f-4ee4-479b-91f9-a8e04d8d12d8 | easy | AFeinfield | 2026-08-28 02:22:01.332371+00:00 | succeeded | cae587efadbcb666d1204659aafaaa07ad7abfab3f36ae118af1bf26d0609bbd | 7,797 | null | """179: GROKKED RESIDUE MAP — the literature bet for fixed-N datasets.
Grokking modular arithmetic to 100% held-out accuracy is established
(Power et al. 2022; Gromov 2023 for MLPs; GrokFast for 50x acceleration).
Squaring mod a FIXED N is in that task class, so for fixed-N datasets we
learn the entire one-step map x ... | {
"id": "4db0f75f-4ee4-479b-91f9-a8e04d8d12d8",
"created_at": "2026-08-28 02:22:01.332371+00:00",
"db_md5": "a4039c23ea39a76d70bb8d4a19a1a2c7",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "da4202d1-0d61-407e-8346-c2867e3a7c5d",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.2782405614852905,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333656191826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2560207843780518,
"example_count": 30... |
4db5aca6-bce2-4e4c-83ca-acaf1fe1c2ea | easy | viridale | 2026-08-28 11:32:41.485693+00:00 | succeeded | 02734c0cfe811b709c8c914eda79f9d432efdb188b4946d5332bada2dbc75d7a | 32,233 | null | """Dynamic-horizon shallow-evidence absolute learned coefficient slab.
hypothesis: v975's selected learned program executes all 64 transitions for every
train/eval batch, even though tensor masking makes every transition above max(T) an
exact no-op. Its complete candidate path already elides the same masked su... | {
"id": "4db5aca6-bce2-4e4c-83ca-acaf1fe1c2ea",
"created_at": "2026-08-28 11:32:41.485693+00:00",
"db_md5": "5979b4e4e41f59e41dbdd5d87806a165",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "7f668fd9-da70-4a01-b741-73f7f996378c",
"tier": "easy",
"dataset_id": "e2",
"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": 300,
"exact_accuracy": 1.0,
"... |
4dbc5ddf-0c09-40fa-b697-eabe27a86e66 | easy | newjordan | 2026-08-15 23:30:46.704725+00:00 | succeeded | 6da6e937508b55c0269fa90d648775dd5beecc4c8d7f4934fa089f933655e8a8 | 178,654 | null | """Helix-Crawler — a recursive-weight-shared "crawler" ALU for modular squaring.
The task is r_{i+1} = r_i^2 mod n, chained T times: a *recurrence* over a single
operator. This build embodies the crawler / Frugendorff lineage (spark:~/sota_crawler)
mapped onto that structure. n never enters as a table row (unseen modu... | {
"id": "4dbc5ddf-0c09-40fa-b697-eabe27a86e66",
"created_at": "2026-08-15 23:30:46.704725+00:00",
"db_md5": "e6c32df21410d4316bc488cf172a8410",
"submitter": "Frosty40",
"github_login": "newjordan",
"run_id": "87b39e3f-3a2c-4a93-8a08-bda14f5a278f",
"tier": "easy",
"dataset_id": "e4",
"status": "succeed... | {
"score": {
"mean_loss": 6.78293052085217e-06,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.774206531213088e-06,
"example_count": 1200,
... |
4dbd1f88-61c8-4364-9830-793447086177 | easy | sapient-sapiens | 2026-08-30 17:03:07.476083+00:00 | succeeded | fd3025a00e090483c3fda42647e308d4ef052391e527011210f5a1a8ef5ca3fe | 14,827 | null | """Eligible direct outer-square Transformer with tied S then A calls per square.
The model makes exactly sixteen input-independent calls: one latent S write and
one A write at each of eight outer squares. N and square-local X are immutable
within a square, A is the only decoded workspace, and T only selects among the... | {
"id": "4dbd1f88-61c8-4364-9830-793447086177",
"created_at": "2026-08-30 17:03:07.476083+00:00",
"db_md5": "677616e21205be0c4f95ca1b1ad92b7b",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "933f5bcb-855d-49ac-bede-11e731bbbb9d",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.855387209222068,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0066666666915019355
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8725424764402243,
"example_count": 60... |
4dbeee25-6d83-4cbb-be8e-53a39a6bb7c6 | easy | ehonig | 2026-08-13 20:14:54.902668+00:00 | succeeded | bb35d0f44d9d62ed79953fbb3b15faba8aa849bc3076f84af4709b67dd138e81 | 38,938 | 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": "4dbeee25-6d83-4cbb-be8e-53a39a6bb7c6",
"created_at": "2026-08-13 20:14:54.902668+00:00",
"db_md5": "fff875c0d38952c310ee4c7148b83328",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "bf93fab8-134d-429f-8cc9-7da536c46c87",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 5.779201030731201,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06999999657273293
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.260680675506592,
"example_count": 100,
... |
4dc07c81-7590-4697-90f6-04715c1c0f37 | easy | DDanlov | 2026-08-27 21:32:22.989434+00:00 | succeeded | 0c5ae7f976c37370a023ed259174730bbf7e55f3f692eca2843435b4881ccef0 | 38,196 | null | import math
import time
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from onelayerdeeper.api import Submission, OptimizerBundle, ModelSpec, OptimizerSpec
except ... | {
"id": "4dc07c81-7590-4697-90f6-04715c1c0f37",
"created_at": "2026-08-27 21:32:22.989434+00:00",
"db_md5": "6679c6629c4d0246b766eaf1449f00e7",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "b5b2a143-bb39-4d37-8830-18881403ec5d",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.543638152749957,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002083333373690645
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5518319606781006,
"example_count": 600... |
4dd07bf2-c2f0-434d-8339-69cb83f60593 | easy | shreyash-chonkie | 2026-08-23 19:10:19.419996+00:00 | failed | c3695b8bfc9cf1913b213b4c8e46e1f19e81aa64d832d4615d0df0679c365f3a | 8,233 | null | """Four-stage recurrent register Transformer for One Layer Deeper."""
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 = 1024
... | {
"id": "4dd07bf2-c2f0-434d-8339-69cb83f60593",
"created_at": "2026-08-23 19:10:19.419996+00:00",
"db_md5": "089c7bc3558fdaef1af6b37e19105e43",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "343bfe1d-a7da-4c92-8db5-a5fab5d13089",
"tier": "easy",
"dataset_id": "e6",
"status": "... | null
|
4dd5aa1b-c8b9-48ce-af60-6ca1940e6e92 | easy | shan23chen | 2026-08-08 01:59:15.693683+00:00 | succeeded | b0622ead959f5bd9917eb82b303c200cfd9cb274031c016e961554892eb1518f | 20,441 | 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": "4dd5aa1b-c8b9-48ce-af60-6ca1940e6e92",
"created_at": "2026-08-08 01:59:15.693683+00:00",
"db_md5": "2b7b75290a960232c12bfd5171cb7ab3",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "7dfc65e0-a3ef-4135-85d0-c6b4ccc183d5",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.9240739345550537,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.015000000596046448
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.925436019897461,
"example_count": 600... |
4dd6e687-f8fa-458c-be64-1ecc7c277603 | easy | shan23chen | 2026-08-10 18:51:57.906804+00:00 | succeeded | 42b6843f7ec377ac5228fb7c88c22afea5928ca559e0290ad08682f2879ae860 | 41,591 | null | """Dual-root boundary-reset recurrent register Transformer.
One learned transition is applied once per requested outer step. The
transition receives an immutable decimal N register and the current learned
boundary state, but never receives T or a separate copy of the original X.
Within each outer application, scratch... | {
"id": "4dd6e687-f8fa-458c-be64-1ecc7c277603",
"created_at": "2026-08-10 18:51:57.906804+00:00",
"db_md5": "6f39fc10d25a043f44e7f4e942a22d17",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "81cc1722-a484-4fb3-9e24-1b5ced9b4187",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 2.264200210571289,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011874999850988388
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.260340929031372,
"example_count": 800,... |
4de6985d-adcc-48dc-af7a-fa6787959a57 | easy | Maxwells-Demons | 2026-08-13 14:01:13.381171+00:00 | succeeded | 23e7f8e8613c954a19d39d68f14add210cabbd2fdbaad0993678dc4734b173d9 | 121,281 | null | """One shared digit operator, its two free neighbours of x, and two learned maps.
The backbone is unchanged from v47/v96: the state IS the digit one-hot, one
application of the operator is
(state digits, N digits) -> next state digit logits
and the straight-through code of those logits is the whole state handed to... | {
"id": "4de6985d-adcc-48dc-af7a-fa6787959a57",
"created_at": "2026-08-13 14:01:13.381171+00:00",
"db_md5": "3155441c933270b9f447364413d08fb5",
"submitter": "Maxwells_Ayakashi",
"github_login": "Maxwells-Demons",
"run_id": "2c8ef11e-9332-4025-809d-7a3925ed8e3e",
"tier": "easy",
"dataset_id": "e3",
"st... | {
"score": {
"mean_loss": 3.824139915776734,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.8675000035762788
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.05494731747407,
"example_count": 800,
... |
4dec09ba-b69a-42db-96fa-b7e673ee30b3 | easy | erdavis0 | 2026-08-24 09:17:32.641373+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": "4dec09ba-b69a-42db-96fa-b7e673ee30b3",
"created_at": "2026-08-24 09:17:32.641373+00:00",
"db_md5": "74f468822dce7e749383257b9e669998",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "4663787c-7358-4f3d-a5d5-afe13b79f7b6",
"tier": "easy",
"dataset_id": "e9",
"status": "succeeded",... | {
"score": {
"mean_loss": 8.904873847961426,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03222222253680229
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.180625915527344,
"example_count": 90,
... |
4df59374-1250-4c62-8546-44263f78c832 | easy | richardcepka | 2026-08-13 22:20:04.186310+00:00 | succeeded | 99f45e8e9a3dd8a557d695390445686cf75e8d1126813bb8a5f2677325cab491 | 43,380 | null | """Recurrent digit-register model for One Layer Deeper.
A looped transformer whose recurrent state is a decimal digit register that is
re-quantised through a ten-entry codebook on every iteration. Nothing inside
the loop depends on the iteration index and no parameter is indexed by an
absolute slot position, so the e... | {
"id": "4df59374-1250-4c62-8546-44263f78c832",
"created_at": "2026-08-13 22:20:04.186310+00:00",
"db_md5": "58c1fe9141c182b0fc57643f3468ad69",
"submitter": "Richard Cepka",
"github_login": "richardcepka",
"run_id": "1dd74d35-c6f6-455d-8ec3-f6d0fa8482a2",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.213578069599362,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002083333373690645
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2145678997039795,
"example_count": 600... |
4df5e604-2cff-40e1-b9a5-e3b128afe60c | easy | viridale | 2026-08-15 06:59:37.532031+00:00 | succeeded | c544c255889ca02667f3ed156c6d2a4d52d1586c0ef62f6de3810952d76991cd | 18,554 | null | """Width-equivariant raw-measure translation-shared scan.
hypothesis: v633's Easy-to-M6 collapse is caused by the conjunction of false shared
supports, length-dependent reverse padding, and destruction of routed additive
mass; enforcing all three representation laws should transfer boundary acquisition.
axis: ... | {
"id": "4df5e604-2cff-40e1-b9a5-e3b128afe60c",
"created_at": "2026-08-15 06:59:37.532031+00:00",
"db_md5": "a798a57ba7c242e5079e947980b887ed",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "78640fb3-9471-4f1a-9942-f6af9537e909",
"tier": "easy",
"dataset_id": "e10",
"status": "succ... | {
"score": {
"mean_loss": 3.5519694089889526,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.15113131701946259
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.3499226570129395,
"example_count": 125... |
4dfdecfa-9b9d-45f5-8700-17c039c70a1b | easy | DDanlov | 2026-08-26 20:35:21.287137+00:00 | succeeded | 6b2a41d8a86ba26f5b2915c748dea4c27166d088f380d0036207a0318894d3ba | 34,345 | null | import math
import time
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from onelayerdeeper.api import Submission, OptimizerBundle, ModelSpec, OptimizerSpec
except ... | {
"id": "4dfdecfa-9b9d-45f5-8700-17c039c70a1b",
"created_at": "2026-08-26 20:35:21.287137+00:00",
"db_md5": "6698f98a9dbbb1997edeb6751f3e2d22",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "9548154d-9d9d-46d9-88d4-063dec52b3d5",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.3009389641095024,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2996903577727585,
"example_count": 600,... |
4dff2b66-4127-4541-9b40-c0b571527f0c | easy | DDanlov | 2026-08-19 20:11:58.611976+00:00 | succeeded | 43d106c7b01f1c9bc032c4ab93fa9ef1cfb5be32c23cca5b3ae6a13cfd171f68 | 17,727 | null | """
Dynamic Submission for One Layer Deeper API Benchmark
Config: dim=1696, num_heads=32, d_ff_mult=4, fixed_trec=192
"""
from __future__ import annotations
import math
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
... | {
"id": "4dff2b66-4127-4541-9b40-c0b571527f0c",
"created_at": "2026-08-19 20:11:58.611976+00:00",
"db_md5": "9f5a0b292534b3c9d55afd9999ac6df7",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "f5441cb1-fd69-4004-9650-3d2e68eaa0e6",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.4048448288737605,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012083333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4825532805224704,
"example_count": 60... |
4e05832d-a505-45f2-a4c8-94f5481c0956 | easy | pculaf | 2026-08-25 07:26:33.040798+00:00 | succeeded | f8f2a30f87664a77524dabd8bcc6f6ffda95379be981acccf6300e60f4098c28 | 7,752 | null | """Small single-pass post-RMSNorm Transformer for One Layer Deeper."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
... | {
"id": "4e05832d-a505-45f2-a4c8-94f5481c0956",
"created_at": "2026-08-25 07:26:33.040798+00:00",
"db_md5": "bc9d50be53478a313d394a2159d213c8",
"submitter": "Pavle Culafic",
"github_login": "pculaf",
"run_id": "1a0c95a8-e65b-48cc-96d6-7832cd04d0b9",
"tier": "easy",
"dataset_id": "e6",
"status": "succe... | {
"score": {
"mean_loss": 1.944625437259674,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03851351607590914
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9077816009521484,
"example_count": 60,
... |
4e0b8541-ec0b-4fad-894b-f71d86804030 | easy | liam-gb | 2026-08-13 02:27:54.987619+00:00 | succeeded | 5ae4da782d01d42a5e2630f18f3c7dc668f4f397431ff7645051fa3182e49267 | 16,651 | null | """rns_pm_gather_ct_widet_drank: dense loop RETAINED, plus a parallel
residue track supplying the CRT rank.
Base: widet with lookup tables at lr 3e-2.
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... | {
"id": "4e0b8541-ec0b-4fad-894b-f71d86804030",
"created_at": "2026-08-13 02:27:54.987619+00:00",
"db_md5": "1c479f1e07ec8e176359058d7ba96f62",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "f5bd5109-9414-41b3-8f50-05491dec99e7",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.346919008539066,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.037916666672875486
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.865854263305664,
"example_count": 600,... |
4e0e4184-c220-4230-9cda-79f827e5fd43 | easy | sapient-sapiens | 2026-08-04 17:09:20.058421+00:00 | succeeded | 5c50c8a299a40c8f61f8f383a7d2b7d76ff676e332f38f34bc0d8d754a7cc2e8 | 18,639 | null | """v2: weight-tied recurrent Transformer with depth-aware supervision.
Fixes relative to Phase 2–6:
- Per-row T parse; supervise only rows with T <= depth at each readout.
- Genuine hold terms beyond T; deep-curriculum under truncated BPTT.
- Power-iteration spectral normalization (contractive residual).
- Det... | {
"id": "4e0e4184-c220-4230-9cda-79f827e5fd43",
"created_at": "2026-08-04 17:09:20.058421+00:00",
"db_md5": "7a039edd75cfdb1c8501f84df71186bc",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "163a215b-406f-4400-b77d-b75e1f7286d8",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 1.9178665301151705,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07499999846021335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8708958625793457,
"example_count": 100... |
4e140318-d00a-45bd-81ad-81a885122fc8 | easy | DESU-CLUB | 2026-08-30 18:45:11.725206+00:00 | succeeded | bb98c13a9397ef0bf289bf14d935f94ca4580fdee8ebed0e55b46ef2d20f472d | 13,947 | null | """Discrete-state loop over a place-value tape, trained with orthogonalize-only Muon.
The recurrent state carried between squarings is the *digit string*, not a
residual stream. Each outer iteration reads (N, x) off an LSB-first place-value
tape, emits digit logits for one squaring, hard-argmaxes them (straight-throu... | {
"id": "4e140318-d00a-45bd-81ad-81a885122fc8",
"created_at": "2026-08-30 18:45:11.725206+00:00",
"db_md5": "47b7737ce8cf879f2bffb9c8e5d9fd94",
"submitter": "Low Keng Hoong, Warren",
"github_login": "DESU-CLUB",
"run_id": "f6a85a97-933c-4c83-baeb-7b59adfe42fd",
"tier": "easy",
"dataset_id": "e6",
"sta... | {
"score": {
"mean_loss": 1.1899407363101226,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6720720875907589
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.7592005133628845,
"example_count": 60,
... |
4e153efa-01f8-4203-9602-2bb6404e4e65 | easy | DDanlov | 2026-08-31 20:16:59.093345+00:00 | failed | 54ac43a6a5799e1cffc8d449db0011362751e03d1764f5a1afde23903c6cb761 | 61,375 | null | """
Pure Skipless Recurrent Deliberator with Causal Autoregressive Sequence Decoder
================================================================================
File: src/clean_slate_model.py
A principled, scalable deep neural architecture for discrete modular squaring (x^{2^T} mod N):
1. Input Representation:
... | {
"id": "4e153efa-01f8-4203-9602-2bb6404e4e65",
"created_at": "2026-08-31 20:16:59.093345+00:00",
"db_md5": "1c4c3468a2fd70cf37e7b0ccfa544210",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "d7b02d67-e6d9-4cdc-b3af-319726f72194",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
4e185b99-32a9-4bab-8383-b848f0510c9e | easy | liam-gb | 2026-08-16 23:02:18.488844+00:00 | succeeded | 82632945ab7e0c87d203dfc4e2f107a9e7f49bf180d7e486a194d56b896646c1 | 14,535 | null | """charblock: a from-scratch body whose per-prime step has a known closed-form
solution, trained in the regime where that solution is documented to be found
from random initialisation.
Five ingredients, declared as one ARCHITECTURE+REGIME bundle:
1. Per-prime two-layer blocks with a QUADRATIC activation. For each ban... | {
"id": "4e185b99-32a9-4bab-8383-b848f0510c9e",
"created_at": "2026-08-16 23:02:18.488844+00:00",
"db_md5": "a79f5e7cd40035de9d3d3adb1e10b7b5",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "27a7c667-c57b-4254-a9af-3d61e0d50903",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.340604305267334,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4195946007966995
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.496870517730713,
"example_count": 60,
... |
4e1e4f9d-3e9a-414e-a109-381818181981 | easy | yashkant | 2026-08-09 03:06:32.222000+00:00 | succeeded | 075b8b560e2bc7aa6189a86079343a506a8a99429e6db01db1e1aee5ebf84746 | 22,841 | null | """Round 211 common TRM-style x/y/z shared-transition template."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
from torch import... | {
"id": "4e1e4f9d-3e9a-414e-a109-381818181981",
"created_at": "2026-08-09 03:06:32.222000+00:00",
"db_md5": "6e76b66c89d2b63b0d67174df1ee1970",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "baae06a1-ca10-467e-a276-eda3a139b0fc",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 4.652785539627075,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04333333298563957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.0775017738342285,
"example_count": 100,... |
4e21bbaa-bcde-4065-bc37-d6ca0367ede8 | easy | kushasareen | 2026-08-27 03:51:39.305659+00:00 | succeeded | 6215121c4ac5d59973f6605c30ee6f41b9a2a8505c720b42ead6258c809571fa | 10,901 | null | """loop1_noT: digitcomp (4 blocks) where blocks 2..2 repeat T times; T token dropped.
At T = 1 this is exactly the digitcomp model. Muon recipe, 3 updates per batch.
Each number's token is a learned function of its digits: every place has
its own embedding table, the place embeddings are summed and passed
through a s... | {
"id": "4e21bbaa-bcde-4065-bc37-d6ca0367ede8",
"created_at": "2026-08-27 03:51:39.305659+00:00",
"db_md5": "cbe892f66eb2fada8763e1afcfe4a65b",
"submitter": "Kusha Sareen",
"github_login": "kushasareen",
"run_id": "ed50d6b7-96d5-4293-aea8-7072a5dc001d",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 2.847497880458832,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4214583598077297
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.304295539855957,
"example_count": 300,
... |
4e2a1154-7452-48ff-abde-95cb71ff0694 | easy | ddoan | 2026-08-14 08:56:43.833841+00:00 | succeeded | 7ae812a17a2fcfa1833c4c1a51076751430f9e8144203cdc36c40f717abf4e44 | 7,016 | null | """Batch-600 located recurrence with four shared transition steps."""
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,... | {
"id": "4e2a1154-7452-48ff-abde-95cb71ff0694",
"created_at": "2026-08-14 08:56:43.833841+00:00",
"db_md5": "255324b221cd0b468e8b5aaa915e4cdd",
"submitter": "Doug Doan",
"github_login": "ddoan",
"run_id": "9aa76e27-b0c1-4fbd-b547-c2803bfce2b7",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.1717322357814624,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000024835268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1733965809569766,
"example_count": 60... |
4e3336a8-d248-4a11-9170-4615d706a41f | easy | sapient-sapiens | 2026-08-09 15:00:40.900079+00:00 | succeeded | 9ba049d1c2df4d13aa5a119aec4ec3b5df44dd0311ae58892e3107ddc2d325bd | 13,048 | 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,
TokenLossBatch,
... | {
"id": "4e3336a8-d248-4a11-9170-4615d706a41f",
"created_at": "2026-08-09 15:00:40.900079+00:00",
"db_md5": "34b550e6189e2eed912f02c824ceca34",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "e2b67e74-9cff-45bd-8c2f-76d756d8f130",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.380681301493796,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.376476719307258,
"example_count": 600,
"exa... |
4e356d04-8b21-44b1-b5b8-a42dc8e67eb2 | easy | liam-gb | 2026-08-17 01:49:31.088857+00:00 | succeeded | 24fc878b19987564e88eccf1d98937ae5735fa8ab620fdef470b9eacfec14e2f | 16,740 | null | """rns_pm_gather_ct_widet_l32nore_r8: single change vs
rns_pm_gather_ct_widet_l32nore — the batch-reuse policy returns True, so the
evaluator reuses each fetched batch for the maximum eight optimizer updates.
WHY. On the granular Easy suites one batch is 85-97% of the training split,
so every step ends an epoch and th... | {
"id": "4e356d04-8b21-44b1-b5b8-a42dc8e67eb2",
"created_at": "2026-08-17 01:49:31.088857+00:00",
"db_md5": "46c5decfff7d97a850fb297981e9d310",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "e8eecc58-9ec4-4057-a52c-4af1d6456731",
"tier": "easy",
"dataset_id": "e9",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.868093967437744,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4544444568455219
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.594183921813965,
"example_count": 90,
... |
4e39f9ed-3a11-4e11-88e2-e8043b205ad0 | easy | AFeinfield | 2026-08-28 12:32:47.167338+00:00 | succeeded | da92817bff67514f5ac8d12736f033e1dcac5803429d3bad57384fb8faff1d24 | 11,889 | null | """191: LEARNED-TABLE EXACT-ROUTING MACHINE — the perfect-score design.
Every arithmetic operator is a learned classification table over a tiny,
fully observed discrete domain (all random init):
mul_lo/mul_hi (10x10x10): digit times digit -> low/high result digits
add_lo (10x10x10), add_carry (10x10x2): digit addi... | {
"id": "4e39f9ed-3a11-4e11-88e2-e8043b205ad0",
"created_at": "2026-08-28 12:32:47.167338+00:00",
"db_md5": "4a3a07f94de298edbcbef19dec129808",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "1cc628df-59c9-4261-baf0-5b02cc6c23d0",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.7961496114730835,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.784252405166626,
"example_count": 100,
"exac... |
4e40a14e-95e8-409a-868d-1540d6f2cbd7 | easy | DDanlov | 2026-08-27 18:42:30.197352+00:00 | succeeded | 537a29fe91bb5cb9f38134b32292297ead93a8479f47c664ae6d9bb8f54b04d1 | 38,067 | null | import math
import time
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from onelayerdeeper.api import Submission, OptimizerBundle, ModelSpec, OptimizerSpec
except ... | {
"id": "4e40a14e-95e8-409a-868d-1540d6f2cbd7",
"created_at": "2026-08-27 18:42:30.197352+00:00",
"db_md5": "4b54bcfb1bd066dc2caf11af40c0847b",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "0c43e9ec-8499-44f1-a8dd-047734a397e8",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.555666293938655,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.557476584686827,
"example_count": 600,... |
4e5a1372-39e0-47f2-8765-cdb08e1f893a | easy | anirudh-chakravarthy | 2026-08-12 01:49:59.920480+00:00 | succeeded | 605e49a0d55b8a24924f99953729f45483997526a7a7b0410b6b2e55eff2faa2 | 14,900 | null | """A learned state transition applied once per requested time step."""
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,
TokenLossBatch,
asser... | {
"id": "4e5a1372-39e0-47f2-8765-cdb08e1f893a",
"created_at": "2026-08-12 01:49:59.920480+00:00",
"db_md5": "c6bc5ee15aa7912268f604016e5ff861",
"submitter": "Anirudh S Chakravarthy",
"github_login": "anirudh-chakravarthy",
"run_id": "76799ce0-2998-4920-84a1-ab1f2804c53b",
"tier": "easy",
"dataset_id": "... | {
"score": {
"mean_loss": 2.2626039534762055,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2741987427253894,
"example_count": 60... |
4e5f16df-5996-4ff9-bc5a-db39ea5c9734 | easy | newjordan | 2026-08-09 20:03:11.911514+00:00 | succeeded | a729f6bb0df1dfe48b68658b6d747e9d765131484ce839d06a15dac32399a53d | 94,093 | null | """Helix-Crawler — a recursive-weight-shared "crawler" ALU for modular squaring.
The task is r_{i+1} = r_i^2 mod n, chained T times: a *recurrence* over a single
operator. This build embodies the crawler / Frugendorff lineage (spark:~/sota_crawler)
mapped onto that structure. n never enters as a table row (unseen modu... | {
"id": "4e5f16df-5996-4ff9-bc5a-db39ea5c9734",
"created_at": "2026-08-09 20:03:11.911514+00:00",
"db_md5": "c3aa11c6494d5acd02af87ecf0d026d9",
"submitter": "Frosty40",
"github_login": "newjordan",
"run_id": "36d7d7c0-f455-4629-9c9b-ae5af98c6fc9",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 6.429512001239777e-07,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.4404096099076e-07,
"example_count": 800,
... |
4e61c3c7-056b-4e1e-a4ab-2c78ab266892 | easy | jordanrubin | 2026-08-06 07:09:07.802603+00:00 | succeeded | bc544a598b08e36767e7ace2e9fd9978df141535f851a95bb210df89415b4574 | 15,972 | null | """Gauge-free pair-relational categorical looped Transformer.
This is a learned transition model, not an arithmetic program. The fixed
parts expose only the public prompt grammar and six representation priors:
right-origin decimal slots, a shared ordered digit geometry, unordered pair
tokens, immutable N recall, a so... | {
"id": "4e61c3c7-056b-4e1e-a4ab-2c78ab266892",
"created_at": "2026-08-06 07:09:07.802603+00:00",
"db_md5": "614e809a4d95bde69942b92c2b5c046b",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "37a25fd3-fd75-4323-9b70-d57933464236",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 1.885040819644928,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8441998958587646,
"example_count": 100,
"exac... |
4e689099-6e2b-4ebb-9d47-452e8fe4e7b9 | easy | ehonig | 2026-08-18 13:01:31.988966+00:00 | succeeded | eedc11bb7437c9f834e0c512cd2e08b0c79c24802c0195db6ba28d2414eac966 | 13,504 | 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": "4e689099-6e2b-4ebb-9d47-452e8fe4e7b9",
"created_at": "2026-08-18 13:01:31.988966+00:00",
"db_md5": "3319ebf686b078036467bd6a65cf6971",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "fbcd034b-4ff0-4324-9427-ba9a63598dfc",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.0637767600883237,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0655454565294358,
"example_count": 1200,
... |
4e6d115a-7041-4c6f-8ce3-6533ab98983f | easy | pculaf | 2026-08-15 17:45:23.586216+00:00 | succeeded | e70980cd7a5bb15acbc2937f672faaf799e80d2cc762c7066452ec0f58e0417a | 8,354 | null | """Muon-optimized recurrent Transformer for the E6 activation/depth sweep."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submissio... | {
"id": "4e6d115a-7041-4c6f-8ce3-6533ab98983f",
"created_at": "2026-08-15 17:45:23.586216+00:00",
"db_md5": "c607766f51f34aeb0a1c0e263a3ba3c5",
"submitter": "Pavle Culafic",
"github_login": "pculaf",
"run_id": "1e6e22fd-5cf3-4d37-aee9-93109176ea3f",
"tier": "easy",
"dataset_id": "e6",
"status": "succe... | {
"score": {
"mean_loss": 3.0862203240394592,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.025225226767361164
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.195132732391357,
"example_count": 60,... |
4e6ddbf9-d1d2-4b0d-b70b-81b87dcb959f | easy | Yalyenea | 2026-08-12 06:59:38.148418+00:00 | succeeded | cb55e3bded7f7aefd9cb0c6db777bfc4ac65d5acf8ffa09145324183efa49792 | 10,480 | null | """Right-aligned decimal-state recurrence for separate prompt/output data."""
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,
TokenLossBatch,
... | {
"id": "4e6ddbf9-d1d2-4b0d-b70b-81b87dcb959f",
"created_at": "2026-08-12 06:59:38.148418+00:00",
"db_md5": "d7ec34f50dc721579f80a7afffe9cd64",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "9e09d666-d5d6-4bd7-9da6-6635c6a521f1",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 3.2890308984966428,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5328240394592285,
"example_count": 60... |
4e70f7b4-620c-440a-be10-08eb7ba3133f | easy | DDanlov | 2026-08-20 12:25:22.305169+00:00 | succeeded | ed828e10e5b0d33e4a2942bbd073bf1433c9c06caab5174e9fcb8dd4839bf113 | 20,958 | null | from __future__ import annotations
import math
import time
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
except Imp... | {
"id": "4e70f7b4-620c-440a-be10-08eb7ba3133f",
"created_at": "2026-08-20 12:25:22.305169+00:00",
"db_md5": "ebd6ee088bfe5dabb5324283fc8d9060",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "385b2d8b-a417-4293-8ff7-5f7a06332d6e",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.533939166788179,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.540101511061459,
"example_count": 600,... |
4e7c23cf-4200-479f-a2d6-da492c3bea56 | easy | Ranoobaba | 2026-08-16 08:10:24.095171+00:00 | succeeded | d99a0514368a50256f91c7967c97fded19a9e902df3ed0c9110134fc2fe08e3e | 5,803 | null | """Looped transformer, iteration count conditioned on T read from the prompt.
Core idea: learn one latent "squaring step" operator and apply it T times.
Depth generalization comes from the loop structure itself; the model never
needs more parameters for deeper problems, only more iterations.
"""
from __future__ impor... | {
"id": "4e7c23cf-4200-479f-a2d6-da492c3bea56",
"created_at": "2026-08-16 08:10:24.095171+00:00",
"db_md5": "551da5941e8398d931cfc7bb3dbdcd11",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "e05395a3-7f98-4a61-a141-e62b741be5d4",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 6.871619626470932,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03833333184321722
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.093353748321533,
"example_count": 100,
... |
4e80c10e-ac1c-4f42-a9ba-c5e2eb8b2d26 | easy | jordanrubin | 2026-08-03 21:32:48.887923+00:00 | succeeded | 09586f90ffa296e3bdbb25e6a0b059c7579344900d2f7d9eb12b0e902c4848f6 | 15,390 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "4e80c10e-ac1c-4f42-a9ba-c5e2eb8b2d26",
"created_at": "2026-08-03 21:32:48.887923+00:00",
"db_md5": "ef7d67d872a65aa96e81ca53991ab27d",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "f3193c71-0be1-481e-926e-73a16bf91806",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 3.514654789442125,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.18041666356536248
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.519161993064393,
"example_count": 300,
... |
4e8dd872-7feb-4890-ade2-c0942d5717c3 | easy | blake-camp-surge | 2026-08-04 02:52:07.919990+00:00 | succeeded | 5c13aaa89e5dd7611ef95334b679a8bb1f65bfda81a9a07248c51b19fb89aacb | 10,024 | null | """Vector-state virtual neurons driven by one shared QKV-generating DAN."""
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,
TokenLossBatch,
... | {
"id": "4e8dd872-7feb-4890-ade2-c0942d5717c3",
"created_at": "2026-08-04 02:52:07.919990+00:00",
"db_md5": "21e910cd8c1e700d6268ec31232e9f69",
"submitter": "Blake Camp",
"github_login": "blake-camp-surge",
"run_id": "05d6b7ee-42ce-4168-b844-adf148d3443b",
"tier": "easy",
"dataset_id": "e1",
"status":... | {
"score": {
"mean_loss": 2.288510024547577,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.056666663847863674
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6947436332702637,
"example_count": 100... |
4e8eab58-d11b-40a9-9b5e-cd33aef7be5a | easy | jayyala | 2026-08-18 13:09:55.519113+00:00 | succeeded | e6c747c86aa58869f964c195b1efe51c52ccb8bbc525e3613aff1138a5113d25 | 24,443 | null | """One N-equivariant step: wrap a learned bit-pair product, then invert.
exp(2*pi*i*k/N) depends only on k mod N, so the integer product wraps into
the reduced residue. That wrapping is the transition, reused T times. A
tied coarse-to-fine decoder inverts the phase; it is not a table of (N, x).
Prompt reconstruction... | {
"id": "4e8eab58-d11b-40a9-9b5e-cd33aef7be5a",
"created_at": "2026-08-18 13:09:55.519113+00:00",
"db_md5": "4da96419acfd905f77442ec1acb5a1d0",
"submitter": "jayyala",
"github_login": "jayyala",
"run_id": "dd0d74dc-df79-45b4-8e0f-270e8dc4edb0",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 0.49800869822502136,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9833333194255829
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.5970149040222168,
"example_count": 100... |
4e9807cc-1727-4144-b39b-83361710836b | easy | Avyukth | 2026-08-19 15:11:02.411345+00:00 | succeeded | eb213f88d08e02ae28ea24086a0c4e173bfba4d93f2d9b5fa501688cba22b28b | 27,553 | null | """v16 — general-N T-dependent torsion-orbit augmentation (T-503 generalized).
T-503 (hardcoded N=323 kernels) scored 35.00% on hosted Easy e1 — the
strongest generalization signal of the campaign. This version computes the
same 2-power torsion kernels for ANY prompt-parsed modulus at runtime:
factor N (Miller-Rabin ... | {
"id": "4e9807cc-1727-4144-b39b-83361710836b",
"created_at": "2026-08-19 15:11:02.411345+00:00",
"db_md5": "9d1d69d7899e4068c402fabbc4605cb3",
"submitter": "Subhrajit Makur",
"github_login": "Avyukth",
"run_id": "abca54aa-c93a-4fc0-a17e-03077da4f4e6",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 4.947626173496246,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.39166666474193335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.48330307006836,
"example_count": 100,
... |
4e988d48-4f37-4d48-b582-2682fa5b821e | easy | pculaf | 2026-08-15 22:33:29.486183+00:00 | succeeded | 30baa8eb29da03e62a2a8b490d55922adf70f2a4d237c7fa916cebf44912b518 | 10,444 | null | """Muon-optimized recurrent Transformer for the E6 activation/depth sweep."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submissio... | {
"id": "4e988d48-4f37-4d48-b582-2682fa5b821e",
"created_at": "2026-08-15 22:33:29.486183+00:00",
"db_md5": "00f5e626a3c49b5ee613f40728be50e5",
"submitter": "Pavle Culafic",
"github_login": "pculaf",
"run_id": "3c7ebded-614a-455b-af82-28992dfeddc1",
"tier": "easy",
"dataset_id": "e6",
"status": "succe... | {
"score": {
"mean_loss": 1.9376614093780518,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333767950535
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9181727170944214,
"example_count": 60... |
4ea480b5-0300-40a3-bc37-35d8149119ab | easy | benjaminW2025 | 2026-08-20 23:14:07.415567+00:00 | succeeded | 6df80902199db40c838b6112a36e1cd98493ed1aee1d0e0d1c6d2295116a6297 | 5,532 | null | """Phase 1 U8: eight untied blocks with mixed top-quartile sequence CE."""
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_s... | {
"id": "4ea480b5-0300-40a3-bc37-35d8149119ab",
"created_at": "2026-08-20 23:14:07.415567+00:00",
"db_md5": "af0d0d6ba169f5993c04a03880bd3714",
"submitter": "benjawesome",
"github_login": "benjaminW2025",
"run_id": "043dde81-5af4-48e9-bede-2a2184480af5",
"tier": "easy",
"dataset_id": "e3",
"status": "... | {
"score": {
"mean_loss": 2.1735685328762875,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.016875000121071935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2140353222479074,
"example_count": 80... |
4ea7933a-de94-4df0-85f2-79ed726774e5 | easy | jordanrubin | 2026-08-11 01:27:57.886048+00:00 | succeeded | aa0c446a01b1c50a3739ac906b125fbe05d6b1cd9cf0a23d1a564e062285e8d6 | 21,670 | 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": "4ea7933a-de94-4df0-85f2-79ed726774e5",
"created_at": "2026-08-11 01:27:57.886048+00:00",
"db_md5": "344e901468d7fc4aff5006fca563aa6e",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "f6fbb393-4f2f-44b8-ba4a-96091a9e81e0",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 3.3900791997019173,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006249999860301614
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2349380681065703,
"example_count": 60... |
4ea87a30-333a-4345-94c6-fccd60ca16a1 | easy | viridale | 2026-08-04 22:46:11.699825+00:00 | succeeded | e19e464b6e5c86d5c319e79fdba4058e4fd9a63e2c818d5ceefe5a49a12b6d64 | 9,728 | null | """Clean compact product-quantized autonomous recurrent token model.
hypothesis: v178's 32x16 state memorizes E1 and halves updates; reducing to 16 groups
and 8 codes gives a fourfold cheaper, much lower-entropy hard state that may force
shared structure while retaining enough generic capacity for token transd... | {
"id": "4ea87a30-333a-4345-94c6-fccd60ca16a1",
"created_at": "2026-08-04 22:46:11.699825+00:00",
"db_md5": "078f7c883f6f6c3588c1d7c9c20c78d8",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "fc932b4e-0b1c-455b-9883-39a6a86b2354",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 4.071259140968323,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.046666666865348816
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.1425557136535645,
"example_count": 100... |
4ea92baf-03d0-462f-8a47-26a164d6c656 | easy | ehonig | 2026-08-19 13:30:41.435811+00:00 | succeeded | f1151cf7fb93a871711b7d29f5a64d0f28eb0a2817c1f9323f9fd01d1be1176c | 15,765 | 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": "4ea92baf-03d0-462f-8a47-26a164d6c656",
"created_at": "2026-08-19 13:30:41.435811+00:00",
"db_md5": "5a8b22e6d9eb37422b59a5a381a7467c",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "d6d3c901-eead-499f-82b3-37410d01e88e",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.426368467174135,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00499999995653828
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4660146236419678,
"example_count": 600,... |
4eb68999-233e-47ae-bb63-f98280edb0b9 | easy | shirvani-jr | 2026-08-28 17:28:52.396037+00:00 | succeeded | fcae023b367b40f4b44f23829739f2280cdce4c9194787a7b0bd9610bb3f6ba8 | 54,138 | 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": "4eb68999-233e-47ae-bb63-f98280edb0b9",
"created_at": "2026-08-28 17:28:52.396037+00:00",
"db_md5": "18f593bd2d110a271152d2e41f3e0b81",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "199ece87-a525-49c4-8f27-16edc07695ef",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.9512848854064941,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06666666641831398
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9288452863693237,
"example_count": 100... |
4eba0f0d-126d-44dc-b8de-6ed4155cf04d | easy | yashkant | 2026-08-23 21:56:58.703741+00:00 | failed | 59e402738126c2a7d9046172f5e9ca605f7d2f104cad24ae6e634ea9f377c5f9 | 11,360 | null | """One Layer Deeper submission: CRT residue-register machine with a learned squaring
transition and a Fourier "clock" bilinear reconstruction readout.
Idea (end-to-end learned; nothing about the task is hard-coded):
* State = per-modulus residue registers over a fixed bank of small primes. The input
integer x is... | {
"id": "4eba0f0d-126d-44dc-b8de-6ed4155cf04d",
"created_at": "2026-08-23 21:56:58.703741+00:00",
"db_md5": "00243cee1eea1842d43733cc6c3cf8e5",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "2ad3c8ef-197f-4660-b0af-b9a4b9a4b31c",
"tier": "easy",
"dataset_id": "e1",
"status": "failed"... | null
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.