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 |
|---|---|---|---|---|---|---|---|---|---|---|
4a94e8dc-0f5d-465b-a874-ed5f8de1d01b | easy | yashkant | 2026-08-09 05:05:06.249804+00:00 | succeeded | 6caa7c3c3ddbfb923ab7b9ea867a6f86b918a127bb36d7e74ee8936c050ea580 | 17,715 | null | """Round 252 clean Round 157 D288/H9/DEC576 scale variant."""
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 Tens... | {
"id": "4a94e8dc-0f5d-465b-a874-ed5f8de1d01b",
"created_at": "2026-08-09 05:05:06.249804+00:00",
"db_md5": "551a789abe8fcba00c304119c3050b22",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "c99f52f3-9422-4e90-bf77-1bb5cddfdbee",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 4.1148680448532104,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.15166666731238365
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.755451202392578,
"example_count": 300,... |
4a96dcd9-5ad3-456b-a69e-208857fa88d4 | easy | oupadhyay | 2026-08-11 08:56:45.229445+00:00 | succeeded | bf10cc4c4b8ba425d2f9b9b7fc30d17b3b5f09a730b67b5970b4bf5f654cbd77 | 7,607 | null | """One-block spatial-gating categorical register with a learned highway gate."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
PAD, N_MARK, X_MARK, T... | {
"id": "4a96dcd9-5ad3-456b-a69e-208857fa88d4",
"created_at": "2026-08-11 08:56:45.229445+00:00",
"db_md5": "4711d49e90b535425069ed44bc799fdd",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "caf753f4-4a4c-4f0e-a4ca-353787104097",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.148817707810021,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.144118748438198,
"example_count": 600,... |
4a982609-0eff-4370-93fc-f9cda950b704 | easy | oupadhyay | 2026-08-28 02:00:22.197130+00:00 | succeeded | 1f1f9f28fce308d6efb67467d83e17f2000dda55f5ef8c5af2218fc227ac2356 | 4,043 | null | """Literal polynomial cell."""
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
D,H=64,32
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_size,s.max_seq_len=vocab_size,max_seq_... | {
"id": "4a982609-0eff-4370-93fc-f9cda950b704",
"created_at": "2026-08-28 02:00:22.197130+00:00",
"db_md5": "a2453ef0c1f82a54936a296df54b8479",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "d966a78e-70d6-476d-81f8-0dba4a527781",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 4.124470233917236,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.11193693988025188
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.9563536643981934,
"example_count": 60,
... |
4aa81dd9-b8e4-4cc3-8681-1da18e111b16 | easy | oupadhyay | 2026-08-12 06:30:00.681133+00:00 | succeeded | e743136ce9b340961e09b62b8a69aeb9d393fe2d62edbb1d844723a7a8c19572 | 7,875 | null | """One-block axial spatial-gating categorical register for the Easy tier."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
PAD, N_MARK, X_MARK, T_MAR... | {
"id": "4aa81dd9-b8e4-4cc3-8681-1da18e111b16",
"created_at": "2026-08-12 06:30:00.681133+00:00",
"db_md5": "c4a4c0d6c1f4a74392d930c0fcc3d6fa",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "29daa277-afd9-44e1-b7e0-565589428a37",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.212347257151987,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000024835268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1565086061109877,
"example_count": 600... |
4aaacbde-019d-4245-94db-93e8632b832f | easy | DDanlov | 2026-08-14 06:33:19.409679+00:00 | failed | 5cb081a2faf8368ab5f915695ebcbef6f089a4ecb9412a7dc0f60ffb372d28bb | 21,858 | null | import math
from typing import Optional
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark import Submission, assert_model_state, OptimizerBundle
except ImportError:
try:
from client import Submission, assert_model_state, OptimizerBu... | {
"id": "4aaacbde-019d-4245-94db-93e8632b832f",
"created_at": "2026-08-14 06:33:19.409679+00:00",
"db_md5": "753db0f81abcbb9abe1f2f1f6e20fcf0",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "fb2b12e8-9fb0-4766-8c36-57cd53b1b9b8",
"tier": "easy",
"dataset_id": "e3",
"status": "failed",
... | null
|
4aaf9ab4-4c55-4010-ba8c-bd0b646fc2eb | easy | sapient-sapiens | 2026-08-17 17:50:26.623410+00:00 | succeeded | a387e80831f82cabb13a659056098928ec6f2a8924253b342777a9ca36eb41a8 | 22,019 | null | """E6 generalization candidate: width-512, d_h=4, 6-block ACT64.
Architecture changes relative to the hosted ACT max-4 source:
- selectable recurrence depths are 1 through 64;
- canonical cumulative-mass halting genuinely removes halted examples from
later recurrent applications;
- the T-field representation preser... | {
"id": "4aaf9ab4-4c55-4010-ba8c-bd0b646fc2eb",
"created_at": "2026-08-17 17:50:26.623410+00:00",
"db_md5": "70165151a25e6549361788e612ee10ec",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "36fb59a2-525a-4b51-b989-a6705c0c8fb1",
"tier": "easy",
"dataset_id": "e6",
"status": "su... | {
"score": {
"mean_loss": 5.849394824406872,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.081756759736989
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.168057441711426,
"example_count": 60,
... |
4ab30fa4-a7fa-4482-b863-fc72dd401b2a | easy | yashkant | 2026-08-17 00:11:08.708321+00:00 | succeeded | a8cb269a6582229972054809675b8d1c25ee4dcac4c6ac4b6f9783118bad11e1 | 41,111 | null | """Round2236 parsed-T tied masked mean/max encoder-state transition.
The complete Round2230 initialization stream is constructed, but only its
first width-320 masked mean/max encoder block remains live and is reused over
the full token field. Public T selects 2/3/4/5/6/7/8 physical calls for
T=1/2/4/8/16/32/64. Immuta... | {
"id": "4ab30fa4-a7fa-4482-b863-fc72dd401b2a",
"created_at": "2026-08-17 00:11:08.708321+00:00",
"db_md5": "07a10e65d9953e95ed8bfa271408354c",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "19253473-d262-4b60-814d-e25e1dbd08bf",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 3.227891206741333,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.053333332762122154
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.4432568550109863,
"example_count": 100... |
4ab7b25c-0835-4c9f-936b-1936e6f28c7b | easy | yashkant | 2026-08-15 11:58:32.114057+00:00 | succeeded | 1bae221b965dc02164b760f423fd9f13a432696042ed6e215f8f79e29e7e7834 | 44,981 | null | """Round1935 D273 four-digit-modulus residual expert.
The exact Round1917 D273 sparse-SAM8 carrier remains intact. A small bounded
residual is appended after all parent tensors and activated only when the
public N field contains at least four decimal digits. The candidate derives
that residual from the immutable N-f... | {
"id": "4ab7b25c-0835-4c9f-936b-1936e6f28c7b",
"created_at": "2026-08-15 11:58:32.114057+00:00",
"db_md5": "8f44323ab5b06ea9d57c404ddf246080",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "3a68a66b-8f54-40bf-916e-34a9eea2b207",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 2.138610120523163,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013125000060535966
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.140932259060126,
"example_count": 800,... |
4abb3565-e12f-4078-9c6b-df0e670ed039 | easy | shan23chen | 2026-08-08 02:04:56.420447+00:00 | failed | fcd0be99580314a10ec21ef2e8e0e07770e561160730c91ef6339c67b4ac7722 | 15,061 | null | """Context-reinjected stationary Neural-GPU over decimal registers.
The model decodes only the public N/X/T interface. One learned local map is
applied exactly T times, and only a soft decimal register crosses outer-step
boundaries. Immutable N context is reinjected into every convolutional block.
All supervised ter... | {
"id": "4abb3565-e12f-4078-9c6b-df0e670ed039",
"created_at": "2026-08-08 02:04:56.420447+00:00",
"db_md5": "43b6cd529cc5347ffc8f923e55085027",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "8996d5c4-b161-4bd0-be95-b36f685a458e",
"tier": "easy",
"dataset_id": "e5",
"status": "faile... | null
|
4abc9d3e-673d-469b-aae3-e61e18b97bb2 | easy | amirmeisami | 2026-08-16 17:58:05.234645+00:00 | failed | 272cfeacbaed85dea5bbda991a524781277b4d524bc5f6df74aa35ff5fb37d35 | 14,118 | null | """T021 Arm B: learned microprogram with a soft-routed operator population."""
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, assert_model_state
D_MOD... | {
"id": "4abc9d3e-673d-469b-aae3-e61e18b97bb2",
"created_at": "2026-08-16 17:58:05.234645+00:00",
"db_md5": "290b3cc8976f5a870c308d54ff0430ee",
"submitter": "Amir Meisami",
"github_login": "amirmeisami",
"run_id": "03c901be-3ef4-48f4-9048-efe700fbb2bb",
"tier": "easy",
"dataset_id": "e7",
"status": "f... | null
|
4ac96c4f-fdd1-4512-a0c6-f9f069cbb510 | easy | oupadhyay | 2026-08-29 18:37:53.843606+00:00 | succeeded | dc1d71fb86f129c3faec096f4e689b4db10dc8f129fd9d1d6b179570afd20be9 | 3,994 | null | """Conditional optimizer phase 3: warm8_lr9."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_... | {
"id": "4ac96c4f-fdd1-4512-a0c6-f9f069cbb510",
"created_at": "2026-08-29 18:37:53.843606+00:00",
"db_md5": "bd4a1e7a0ebf640cac0f834efc3f69f7",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "56567746-0754-4c4b-8dad-62434a1078d0",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 1.9534559845924377,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02860360499471426
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9307019710540771,
"example_count": 60,... |
4ad1cd9e-115c-45c3-852d-a1aafe3dfdda | easy | Bananafly | 2026-08-12 11:53:34.512415+00:00 | succeeded | 4c06c283121e2e960d1d9809435f31cb69e0dfc484955b562aff0cb93a8c3da9 | 14,169 | null | """T-aligned relative bilinear tape for the Easy E5 architecture screen."""
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,
... | {
"id": "4ad1cd9e-115c-45c3-852d-a1aafe3dfdda",
"created_at": "2026-08-12 11:53:34.512415+00:00",
"db_md5": "0d66f8607b5490548a298282c220fc3d",
"submitter": "Andre Kreidemann",
"github_login": "Bananafly",
"run_id": "19aac67c-bc98-4431-827c-75db311eb3b4",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.1624911244253857,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00999999995653828
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1416306495666504,
"example_count": 600... |
4ad249e8-5e54-4b8e-b878-7c16a387e8e1 | easy | 0Chris5R | 2026-08-02 16:41:23.528593+00:00 | succeeded | 46301343d1d65e8a6fd1a9cb9a8b2d8d7e5d0d98d6d44a12d7d7998b762bd7e0 | 24,605 | null | """Shared-digit Neural GPU with late full-objective EMA evaluation.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the sam... | {
"id": "4ad249e8-5e54-4b8e-b878-7c16a387e8e1",
"created_at": "2026-08-02 16:41:23.528593+00:00",
"db_md5": "d6eea82558adefce35fad3755a529c8e",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "0a95a31a-0c7d-48b1-9d0a-7759dc3d8b64",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.7737601611748035,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03333333334575097
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.626036842842273,
"example_count": 600,... |
4ad7be10-4b93-472c-9284-86ef1dc31a5a | easy | Ranoobaba | 2026-08-19 18:01:31.493492+00:00 | succeeded | b71314738962d767d7c6eeb1c9d67deabd95a17ff4fcfd2b69ef853953d51498 | 6,601 | 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": "4ad7be10-4b93-472c-9284-86ef1dc31a5a",
"created_at": "2026-08-19 18:01:31.493492+00:00",
"db_md5": "462bd1aee09ff4498696aeaa01594429",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "1ab9df3b-64f0-4713-9948-cc5b0039935a",
"tier": "easy",
"dataset_id": "e4",
"status": "... | {
"score": {
"mean_loss": 2.1572631475802186,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1726377846651785,
"example_count": 12... |
4adc8081-af95-4e6e-af4f-eb6665c540a4 | easy | shreyash-chonkie | 2026-09-01 00:55:18.247373+00:00 | succeeded | 40205d730e630b6dcf30ebd3eafac32ce25170b8c8ba9fd68a792f7af774e8ac | 9,434 | null | """Prompt encoder with a routed recurrent-state bottleneck."""
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,
assert_model_... | {
"id": "4adc8081-af95-4e6e-af4f-eb6665c540a4",
"created_at": "2026-09-01 00:55:18.247373+00:00",
"db_md5": "771bd8a27fa74582f1714f2d3fcfe1a1",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "3e01acf1-8257-4abb-81df-fab795ed475a",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 3.696975588798523,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06000000052154064
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.8506953716278076,
"example_count": 100,... |
4addb72d-53ec-4a35-a660-5ec444df3654 | easy | lpbb | 2026-08-11 15:03:59.181145+00:00 | succeeded | 86d711872e89fecbf69443380c6bf91086840137078bdea3f226186e22ae4e7b | 21,570 | null | """DIGIT-STATE DISCRETE FLOW (v2) — diffusion-style quantized recurrence, e1.
v1 (submissions/agents/diffusion) put the recurrent state in a learned
codebook. Flaw found on paper: nothing ties step(enc(x)) to enc(f(x)), so
the code space can fragment into per-chain private paths and the coverage
argument (98-99% of e... | {
"id": "4addb72d-53ec-4a35-a660-5ec444df3654",
"created_at": "2026-08-11 15:03:59.181145+00:00",
"db_md5": "645ba4ad888980c45daea62fb618e7f5",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "f55ad64f-001e-4a4a-af6b-421d49a6e06c",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 0.9079491645097733,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.7899999916553497
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.3281998932361603,
"example_count": 100,... |
4ae2ffd8-ef11-48ee-a12e-1e2fc0205bef | easy | shirvani-jr | 2026-08-21 21:25:59.519825+00:00 | succeeded | ee14953f9376158abdc1052e124d5cc1b81c3e34c0261799a94027633a704971 | 8,496 | null | """Looped digit-state squaring machine.
One recurrent application of a shared block corresponds to one modular
squaring. The recurrent state is the decimal digit string of the current
value, re-quantised every step with a straight-through estimator, so the
learned map composes exactly and depth extrapolates in T.
"""... | {
"id": "4ae2ffd8-ef11-48ee-a12e-1e2fc0205bef",
"created_at": "2026-08-21 21:25:59.519825+00:00",
"db_md5": "0696d988b1f7e73bcd598720749d4fe9",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "63c83837-19cf-4df5-97e7-cf0b264e4de4",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.65391206741333,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.532169818878174,
"example_count": 100,
... |
4af45f6a-b59c-4b45-a7bb-51473879a426 | easy | khushidahi | 2026-08-16 13:09:47.105688+00:00 | succeeded | 9e651eb360c5d93f762db89d8c516c43686d6baafba1583534cc6f12b9947e36 | 24,960 | null | """R109: weak residual-identity prior for functional-root learning.
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": "4af45f6a-b59c-4b45-a7bb-51473879a426",
"created_at": "2026-08-16 13:09:47.105688+00:00",
"db_md5": "4995b56bce1c75b9b961bec4908b64ef",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "1e97dd4b-9fba-484a-9ba8-bd4818e1e286",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 2.8792277861421014,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00625
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.875061543789239,
"example_count": 800,
"... |
4b075180-9459-45ba-8097-da37978c2c6f | easy | jordanrubin | 2026-08-14 01:08:16.430177+00:00 | succeeded | fc98447fc19a656845f48580dd88f53c0b2d7d00c35e32a8fe2d3dfff79a73f4 | 29,581 | 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": "4b075180-9459-45ba-8097-da37978c2c6f",
"created_at": "2026-08-14 01:08:16.430177+00:00",
"db_md5": "d08ba312cd6460ffc449c5d2476626a5",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "b919b7a2-9f77-4bc8-a6de-583d973cdcb2",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.7688235712962266,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.020833332867672045
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2169473642992865,
"example_count": 60... |
4b0d703c-e8ea-4c6d-b317-d8e94e9f3199 | easy | benjaminW2025 | 2026-08-19 05:37:53.003657+00:00 | succeeded | bd739008db9df0cd960bec65df6f11308f15e3897a93d22de3fe4fd8804e4501 | 4,712 | null | """Phase 0 U4: four untied full-token Transformer blocks."""
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
NUM_HEADS ... | {
"id": "4b0d703c-e8ea-4c6d-b317-d8e94e9f3199",
"created_at": "2026-08-19 05:37:53.003657+00:00",
"db_md5": "050481e04268242cf27219a4925ae8e9",
"submitter": "benjawesome",
"github_login": "benjaminW2025",
"run_id": "613bd8f8-fe76-43a7-9546-bb0251fd5f81",
"tier": "easy",
"dataset_id": "e5",
"status": "... | {
"score": {
"mean_loss": 2.1567634172037478,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0037500000248352687
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.164448882432262,
"example_count": 60... |
4b170834-edaa-4b23-b1ac-b035f25f720f | easy | khushidahi | 2026-08-22 01:42:48.192845+00:00 | failed | f4773ba2c2f15997250a3dde90451612e23e38a11cee9a3378f8448909850d2a | 18,994 | null | """R395: minimal Universal-Transformer functional-root control for M5.
This candidate adapts the released Universal Transformer to the competition's
separate-output interface. Numeric prompt symbols are only rearranged onto
equal-length, right-aligned categorical N and X tapes. A third tape supplies
parallel learned... | {
"id": "4b170834-edaa-4b23-b1ac-b035f25f720f",
"created_at": "2026-08-22 01:42:48.192845+00:00",
"db_md5": "1cc7d8f7a931b08ce183d3efcd6d09ce",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "6e53b230-b7bb-4b00-90d9-b1b1e1aaf8a0",
"tier": "easy",
"dataset_id": "e5",
"status": "fail... | null
|
4b2854fd-cba2-499a-9816-92933471cbe5 | easy | karanganesan | 2026-08-06 03:32:13.598483+00:00 | succeeded | 7864569594007b85b228515a710f29f07ca99b8e20e2bbbd3f7cbebe2b14baf9 | 23,794 | 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": "4b2854fd-cba2-499a-9816-92933471cbe5",
"created_at": "2026-08-06 03:32:13.598483+00:00",
"db_md5": "7afa89ee0d908a657bacd3207031ed19",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "2c139571-33b6-4ee4-996d-78b65fd1e2b3",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 5.94509744644165,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00541666685603559
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.026419639587402,
"example_count": 600,
... |
4b2a8bf2-f989-4e4d-a3b9-047ad5047403 | easy | sankalp1999 | 2026-08-04 12:24:23.856565+00:00 | succeeded | 2ec22c147f1ac287c02205722af72ad290e0c2e1383979c414d6955234fa79ac | 7,511 | null | """Latent recurrence: encode once, apply a shared squaring map T times, decode.
v4 put the loop inside a transformer and underfit badly (train acc 4% after
4,432 steps) because attention cost ~7ms/step at 13 tokens and bought nothing.
v2 fit the train set perfectly with a cheap flattened encoder but memorized,
because... | {
"id": "4b2a8bf2-f989-4e4d-a3b9-047ad5047403",
"created_at": "2026-08-04 12:24:23.856565+00:00",
"db_md5": "b8a39eeac4516153030bcc591e2af2ef",
"submitter": "Sankalp ",
"github_login": "sankalp1999",
"run_id": "dfa4a1cf-b147-4af1-9b12-e6a7b4aef42c",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 4.560024261474609,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.22500000149011612
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.352609157562256,
"example_count": 100,
... |
4b3192ec-05fb-4df8-bb2a-32a602efe1ca | easy | yashkant | 2026-08-14 15:33:15.009261+00:00 | succeeded | 026922a6d4619f6ea994f27ac0007ba2bb4651e9b838b20c90d91a3fa37e543b | 35,883 | null | """Round1815 truncated 2-typical Witt cocycle inside Dykstra memory.
The exact Round1792 AdamW5 W8/D2 carrier is retained. Each width-eight
correction is viewed as two length-four 2-typical Witt vectors. At both
existing correction seams, the candidate adds the cross-coordinate Witt
addition cocycle between the prio... | {
"id": "4b3192ec-05fb-4df8-bb2a-32a602efe1ca",
"created_at": "2026-08-14 15:33:15.009261+00:00",
"db_md5": "7d57b5b87b0aae10958030573207417f",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "455265ef-9058-4917-91b9-0a8032964ee2",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 2.1213542719856058,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013125000023283064
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.122505201629395,
"example_count": 800... |
4b32d560-f0ff-44c3-9fe0-b1249df28c94 | easy | arthurfeeney | 2026-08-07 07:39:34.923461+00:00 | succeeded | 38fbdf754749f42281af01ebd4a712f9d44fb06323c061f0a2373c36b0ed43ae | 12,958 | null | r"""
Notes
1. goal is basically learn y = G(x_0, T, N).
G is always a recurrence `x_t+1 = f(x_t, t) mod N`.
Shouldn't be possible in general since f isn't determinable from a dataset...
I.e., multiple f can generate the same training dataset.
2. can't really use any info on structure of f...
"""
from __future... | {
"id": "4b32d560-f0ff-44c3-9fe0-b1249df28c94",
"created_at": "2026-08-07 07:39:34.923461+00:00",
"db_md5": "a5b8337f459eca6252975f41f7d118d6",
"submitter": "Arthur",
"github_login": "arthurfeeney",
"run_id": "2988b752-ef94-4a09-992b-9d093c8a69e5",
"tier": "easy",
"dataset_id": "e1",
"status": "succee... | {
"score": {
"mean_loss": 4.445844769477844,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04666666593402624
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.949726104736328,
"example_count": 100,
... |
4b32f885-2240-4752-9d34-ca97cf49b50f | easy | DESU-CLUB | 2026-08-30 23:54:11.570953+00:00 | succeeded | bed75f65600253f54da31b2391cf05de66a96465df945b63b1ced9194d63dc6b | 20,142 | 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": "4b32f885-2240-4752-9d34-ca97cf49b50f",
"created_at": "2026-08-30 23:54:11.570953+00:00",
"db_md5": "c1a3fd5ec6291b82f9b2cf9aa78e3973",
"submitter": "Low Keng Hoong, Warren",
"github_login": "DESU-CLUB",
"run_id": "6a6fa730-be1e-46c5-9fd5-e5647eb315e2",
"tier": "easy",
"dataset_id": "e5",
"sta... | {
"score": {
"mean_loss": 2.2360772189790277,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1699870230103824,
"example_count": 600,
"e... |
4b334733-c664-4af1-bd35-0cda77f106ed | easy | TZZheng | 2026-08-06 17:50:53.485491+00:00 | succeeded | 8f591099189cd1b3e6a49b57bd7ef9b95c62a395e0294d6ee8ea17acb70b4af1 | 7,304 | null | """Self-contained standard Transformer scaling-sweep submission."""
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,
)
VARIANT = {'block... | {
"id": "4b334733-c664-4af1-bd35-0cda77f106ed",
"created_at": "2026-08-06 17:50:53.485491+00:00",
"db_md5": "2ed97e9246e110f3a4956b6645d51edd",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "832575c3-81d4-4a9c-a9e6-056e6c52f40c",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.822606921195984,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06666666455566883
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.722256898880005,
"example_count": 100,
... |
4b3453c1-2957-41bc-b8c0-06ec39852185 | easy | khushidahi | 2026-08-08 02:33:29.880688+00:00 | succeeded | 993a43d172309d5550c6945bbc7aab0e9eb4dc4be1c9801fac184705d6e0ab88 | 13,289 | null | """R42 Fourier-number representation experiment.
All input-dependent numeric parsing stays inside differentiable GPU tensor
operations. This is a learned-model representation, not a task solver: it does
not compute modular squaring or create intermediate labels.
"""
from __future__ import annotations
import math
impor... | {
"id": "4b3453c1-2957-41bc-b8c0-06ec39852185",
"created_at": "2026-08-08 02:33:29.880688+00:00",
"db_md5": "a56669bef3400e842d7089afe6af1b1b",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "401615e6-f0f7-4f21-af88-a1ac04e4180d",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.159631609916687,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666972450912
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1670713424682617,
"example_count": 600... |
4b34d85b-dd5a-418f-80d4-78450ebc5ab4 | easy | lzy54 | 2026-08-29 08:41:20.569434+00:00 | succeeded | 2198c3e036cb63ce07ed516c32c16517b7cd5fafc3e2aa65b0b27ed0fb8ccf49 | 4,032 | null | """Phase 2 C1: four untied blocks with additive input recall."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HE... | {
"id": "4b34d85b-dd5a-418f-80d4-78450ebc5ab4",
"created_at": "2026-08-29 08:41:20.569434+00:00",
"db_md5": "859baf35963cf1712b065b05e50b68ce",
"submitter": "lzy54",
"github_login": "lzy54",
"run_id": "48402fcf-8386-452b-b91d-d16ab1ca13aa",
"tier": "easy",
"dataset_id": "e8",
"status": "succeeded",
... | {
"score": {
"mean_loss": 3.0220736265182495,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.031105170026421547
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1343235969543457,
"example_count": 85... |
4b375617-89cb-4cf7-a618-9f0c811eba4e | easy | DDanlov | 2026-08-15 14:29:21.921908+00:00 | succeeded | 5f6c7aec446d7e02aff077e034a1e1b6186dec62b69f2fa9908915e0fa012795 | 19,086 | null | """
Official Submission for One Layer Deeper Challenge
Architecture: Pure No-Norm, No-Skip DKS Recurrent Core with Double Embedding (dim=512),
Input RMS Normalization only, 64-Channel Harmonic+Dyadic Torus, Power-of-10 Alignment, and RohanShampoo.
"""
from __future__ import annotations
import math
from typing import O... | {
"id": "4b375617-89cb-4cf7-a618-9f0c811eba4e",
"created_at": "2026-08-15 14:29:21.921908+00:00",
"db_md5": "00683d19f914fc191099ad35ed951342",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "234a3e33-8fd5-4ccd-9e03-b99268fe58d1",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 7.29005865790578,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03500000033527613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.3540018441072155,
"example_count": 100,
... |
4b391b53-1359-4267-9bbb-ceffcdf27063 | easy | andreamiele | 2026-08-06 09:08:37.194548+00:00 | succeeded | cf4a743058354bda0ad39d56bac02e417116299991206e5fafe3280126fcf211 | 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": "4b391b53-1359-4267-9bbb-ceffcdf27063",
"created_at": "2026-08-06 09:08:37.194548+00:00",
"db_md5": "a93f94d5139f1dd297df573bb70b427c",
"submitter": "Andrea Miele",
"github_login": "andreamiele",
"run_id": "296c8f76-efda-4308-bba2-0cb7b187d2a6",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 3.533193102623295,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.028333333532015482
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5408357595329853,
"example_count": 100... |
4b395650-9565-43b7-becc-aae0d2eb5e93 | easy | pepinu | 2026-08-28 07:43:11.288560+00:00 | succeeded | 74307ba76d035d71e8330e51db839434ca24ae8d3efbab3f4b12dd2ec11dd273 | 46,270 | null | """Squaring mod N as a learned permutation on CRT components.
WHY THIS SHAPE. (Z_N, x) is a commutative monoid, so every finite-dimensional
linear representation splits into multiplicative characters and squaring is
diagonal there -- the component basis is the only one in which this operation is
shallow. Two consequen... | {
"id": "4b395650-9565-43b7-becc-aae0d2eb5e93",
"created_at": "2026-08-28 07:43:11.288560+00:00",
"db_md5": "d7f34595d7f996ce3ab1db374501668a",
"submitter": "pepinu",
"github_login": "pepinu",
"run_id": "2b8b7b39-8be1-48d7-9b51-22c3eac8ffdf",
"tier": "easy",
"dataset_id": "e7",
"status": "succeeded",
... | {
"score": {
"mean_loss": 0.014692762866616249,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.014596362598240376,
"example_count": 85,
"... |
4b39b275-0a54-4807-8cd5-030dc8fcdbe9 | easy | sapient-sapiens | 2026-08-19 19:08:15.318601+00:00 | succeeded | ae09ee117d9f6ca8118a44f611fb9da6f820261d0c6982a6412bc9d6e5dc7e14 | 32,520 | null | """Depth-quantized recurrence with a mixture-over-depth objective.
The recurrent state is a bank of right-aligned digit slots. Every operator
application is followed by a soft quantization back onto the token simplex, and
the digit logits produced by that quantization are the answer logits at that
depth. There is no s... | {
"id": "4b39b275-0a54-4807-8cd5-030dc8fcdbe9",
"created_at": "2026-08-19 19:08:15.318601+00:00",
"db_md5": "37bcb99dbc238c32d0276353ff567907",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "90e24818-371d-4161-b6d3-4c147ed395a9",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.882528784367624,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.866528946187999,
"example_count": 600,... |
4b3a32dc-eb2b-4168-824e-2acbe664eb5c | easy | 0Chris5R | 2026-08-02 16:28:30.722299+00:00 | succeeded | 74e6ed9d7f752ee180781b632d957cf7dcec3aedbe035d8eff494d948065db4f | 26,135 | null | """Shared-digit Neural GPU with late full-objective EMA evaluation.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the sam... | {
"id": "4b3a32dc-eb2b-4168-824e-2acbe664eb5c",
"created_at": "2026-08-02 16:28:30.722299+00:00",
"db_md5": "fce314723b0b1328228714af61fdffe7",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "7ae76244-a21c-4558-af8e-838942c78f49",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.604606633890999,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02458333334575097
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4954237777556005,
"example_count": 600,... |
4b423f49-1fb2-4248-b5c0-462d6ba0fd51 | easy | KaustubhKumar05 | 2026-08-25 16:12:52.912260+00:00 | succeeded | 03d0c373e594aa5715323fd2ba901ecc5a0250f4deb9d9e32371e95dc8a36c27 | 9,417 | null | """abacus: faithful port of the Abacus recipe (McLeish et al., arXiv:2405.17399).
The load-bearing idea is the embedding. Each digit is indexed by its
SIGNIFICANCE WITHIN ITS OWN NUMBER (0 = least-significant), from a table
SHARED across N, x and T. Digit k of N and digit k of x therefore land on the
same embedding an... | {
"id": "4b423f49-1fb2-4248-b5c0-462d6ba0fd51",
"created_at": "2026-08-25 16:12:52.912260+00:00",
"db_md5": "80b03c6db927f2e12cf645f2453373df",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "3caacafa-b9cb-4154-a06e-1cc736ff7a31",
"tier": "easy",
"dataset_id": "e8",
"status": "succee... | {
"score": {
"mean_loss": 5.71536922454834,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05632798746228218
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.832545280456543,
"example_count": 85,
... |
4b46b1f1-c4b2-46db-aad7-a6238a90df9f | easy | erdavis0 | 2026-08-22 02:28:38.724036+00:00 | succeeded | ad3cfa4ed3a309cd88e9fbab62f457059a8a4be249920a3faee4d638b54d7bbd | 5,712 | null | """Padding-invariant bidirectional fused GRU with global context."""
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
HIDDE... | {
"id": "4b46b1f1-c4b2-46db-aad7-a6238a90df9f",
"created_at": "2026-08-22 02:28:38.724036+00:00",
"db_md5": "f6574666cb73e6fedd6ea419febb213d",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "aaf22b40-8013-4dfc-b1e6-e57e0b95edbd",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",... | {
"score": {
"mean_loss": 3.911033034324646,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00437499990221113
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.024497032165527,
"example_count": 800,
... |
4b493e92-22e0-48c0-853f-42613dd02329 | easy | sapient-sapiens | 2026-08-21 21:47:19.704762+00:00 | succeeded | eec305e8a05acdd8d39ecb8ee4ebbcbb0022226506449e4f642557677995b4aa | 33,458 | null | """Depth-quantized recurrence with a mixture-over-depth objective.
The recurrent state is a bank of right-aligned digit slots. Every operator
application is followed by a soft quantization back onto the token simplex, and
the digit logits produced by that quantization are the answer logits at that
depth. There is no s... | {
"id": "4b493e92-22e0-48c0-853f-42613dd02329",
"created_at": "2026-08-21 21:47:19.704762+00:00",
"db_md5": "0d343195168f1660f6799922d573c854",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "57b021ee-dcd7-408d-8658-e6bb3c0bd787",
"tier": "easy",
"dataset_id": "e9",
"status": "su... | {
"score": {
"mean_loss": 4.608728605794066,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03222222276031971
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.621289253234863,
"example_count": 90,
... |
4b4ac89a-674d-4500-84a8-a2faf11d91ac | easy | karanganesan | 2026-08-08 04:08:07.196968+00:00 | succeeded | 8d9eecdc8607c91b190336e24b49597738decc69ce7d4e24ee5930ae97cfcad7 | 27,127 | 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": "4b4ac89a-674d-4500-84a8-a2faf11d91ac",
"created_at": "2026-08-08 04:08:07.196968+00:00",
"db_md5": "3e8439ba8ab561733e7dc883fab4ecb2",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "b63d66dd-67f8-4ab0-9058-40e56d17f9bf",
"tier": "easy",
"dataset_id": "e3",
"status": ... | {
"score": {
"mean_loss": 9.620464324951172,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006874999962747097
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.063362121582031,
"example_count": 800,... |
4b4fa800-1312-4b17-b065-ace4596cf7e9 | easy | erdavis0 | 2026-08-22 07:17:26.460584+00:00 | succeeded | a1c22e3f3f7475c81762f81ac0379de4cfc7d9c3c89e6889339eb862577f0cc4 | 6,485 | null | """Field crossbar with learned row and column summaries."""
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 = 64
FIELDS = 6
HID... | {
"id": "4b4fa800-1312-4b17-b065-ace4596cf7e9",
"created_at": "2026-08-22 07:17:26.460584+00:00",
"db_md5": "007d6bd26662c86be5ce3a5c495153a4",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "6d0fccf7-f6fa-4bf8-83e5-f9edf2998e64",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",... | {
"score": {
"mean_loss": 5.453266620635986,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666767559946
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.844635486602783,
"example_count": 600,... |
4b52477b-aa7d-4dba-84b8-4f37b78b1110 | easy | lzy54 | 2026-08-26 14:26:25.511155+00:00 | succeeded | 5c7673b478a41c8d587f3109f31975f282681bd9188b1db726615413eb461124 | 3,845 | null | """Four-pass shared Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
NU... | {
"id": "4b52477b-aa7d-4dba-84b8-4f37b78b1110",
"created_at": "2026-08-26 14:26:25.511155+00:00",
"db_md5": "9fa1927aa50135c5e775508038488486",
"submitter": "lzy54",
"github_login": "lzy54",
"run_id": "c74ad53e-942e-45e2-95ab-08930cd4d69e",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",
... | {
"score": {
"mean_loss": 5.176345586776733,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0468468489125371
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.78737735748291,
"example_count": 60,
... |
4b54e042-1f64-415c-ae21-1cc2f697c555 | easy | shirvani-jr | 2026-08-30 19:49:55.177510+00:00 | succeeded | 0ab0edb088d10d4d8f6d047f406758c8863cef544a5f76736bf06e6c68b99520 | 22,697 | null | from __future__ import annotations
import math
import time
from contextlib import nullcontext
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, TokenLossBatch, assert_model_state
MODEL_WIDTH = 128
ATTENTION_HEADS =... | {
"id": "4b54e042-1f64-415c-ae21-1cc2f697c555",
"created_at": "2026-08-30 19:49:55.177510+00:00",
"db_md5": "daed02fa0d1d596ad94839dba0c5b838",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "0a7b200a-123e-44d0-b051-6d58110073e1",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 8.673224449157715,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00812499993480742
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.653485298156738,
"example_count": 800,
... |
4b5c9064-3e06-45a5-9cf0-30480d3e28d7 | easy | adambrzosko | 2026-08-30 16:03:49.157079+00:00 | succeeded | 8099174425643e4a977abc24404e6913e9ec97ed419106a7308f5c05d5a04c2c | 12,497 | null | """One Layer Deeper — digit-slot recurrent machine, v2.
Prompt: [N] digits(N) [X] digits(x) [T] digits(T); label: the decimal expansion
of x^(2^T) mod N, tail-aligned, so prompt position len-1-s carries the digit of
significance s.
The model parses the three fields onto a grid of slots indexed by (role,
significance)... | {
"id": "4b5c9064-3e06-45a5-9cf0-30480d3e28d7",
"created_at": "2026-08-30 16:03:49.157079+00:00",
"db_md5": "5e17379179dfc87bf1e3a4253be32791",
"submitter": "Adam Brzosko",
"github_login": "adambrzosko",
"run_id": "a8b485d0-3946-42cd-9fc4-444c15e63e88",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.4295737686140084,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666679084302
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.224971689985472,
"example_count": 600... |
4b61091c-6366-44c5-9c65-61597de1c382 | easy | erdavis0 | 2026-08-26 03:33:00.090509+00:00 | succeeded | f323ea27a95b25c17ef240c64fd37d3915d7f3e1d86c503b6dc262e4235c4217 | 30,352 | null | """Latent-role continuous REV C1: selector-oracle distillation.
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 and first e... | {
"id": "4b61091c-6366-44c5-9c65-61597de1c382",
"created_at": "2026-08-26 03:33:00.090509+00:00",
"db_md5": "1658fcec88828e061d8640031c475500",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "64a57b05-43d5-4929-85d4-8336c44130f3",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 4.53936505317688,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08243243768811226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.796785831451416,
"example_count": 60,
... |
4b6148b7-af61-4b72-b6b9-a330f6f8a2be | easy | sapient-sapiens | 2026-08-09 14:39:56.180806+00:00 | succeeded | 382275856d281ae070ad581e7f108a466e47498c941a9e0d381383c62c4ed2c7 | 12,956 | null | """Near-500M adaptive recurrence grown from the hosted-working baseline."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
D_MODEL = 6400
NUM_HE... | {
"id": "4b6148b7-af61-4b72-b6b9-a330f6f8a2be",
"created_at": "2026-08-09 14:39:56.180806+00:00",
"db_md5": "3569dd21c17477c92d16a8c1ea063b06",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "1e87c736-de0e-4061-8c8f-2fcfa205145b",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.488756696819002,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333333333333335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4924733193733233,
"example_count": 60... |
4b6a2fc1-1bf7-4db0-9951-efbfa22a5c0b | easy | isaac0804 | 2026-08-04 17:18:01.341851+00:00 | succeeded | b93fc4a823e4feee3764a8ad68762b121799c2a8b2aab29a0e79a458f8d97802 | 7,956 | null | """Weight-tied looped Transformer that reads T from the prompt to set its own depth.
The task is x_t = x_{t-1}^2 mod N for T steps. A single fixed-depth block has no
mechanism to represent "apply this operation T times" and cannot extrapolate to
depths unseen in training. This submission instead applies one shared blo... | {
"id": "4b6a2fc1-1bf7-4db0-9951-efbfa22a5c0b",
"created_at": "2026-08-04 17:18:01.341851+00:00",
"db_md5": "add486d69448b5f5e42186a2f32a3dec",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "851b1871-e2e7-4af1-ac3a-f82469de547b",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.9463517665863037,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05833333171904087
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9504294395446777,
"example_count": 100... |
4b6a41d4-3f6b-47a8-94fc-8fe7a4bffd06 | easy | velocizapkar | 2026-08-06 21:54:51.181157+00:00 | succeeded | 6b0b775b6c1bac3905c0d066cba95abbb647c9f8eb4588ba44ba5e00bcb6981a | 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": "4b6a41d4-3f6b-47a8-94fc-8fe7a4bffd06",
"created_at": "2026-08-06 21:54:51.181157+00:00",
"db_md5": "333e470deb937b069f35db4fcf2c64e1",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "383b7f7b-e4a1-4236-9b01-ed2874d05679",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 2.6990907192230225,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.022499999962747097
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7421581745147705,
"example_count": 80... |
4b730df2-f18e-46b0-b72c-fc5ddd780d7c | easy | alirezashirvani-jr | 2026-08-31 07:44:27.583101+00:00 | succeeded | c252ce15d535434ae9695254770a32b96e97a7120717f3216a16792d4ff23118 | 34,903 | null | from __future__ import annotations
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
TRACK_FLOOR = 3
TRACK_CEILING = 96
STAGE_B... | {
"id": "4b730df2-f18e-46b0-b72c-fc5ddd780d7c",
"created_at": "2026-08-31 07:44:27.583101+00:00",
"db_md5": "b9699e2f9827d45ad6df7d21eed39459",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "8f831380-52f7-4e48-a84f-84c94280c13c",
"tier": "easy",
"dataset_id": "e3",
... | {
"score": {
"mean_loss": 0.12547419033944607,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9874999821186066
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.19157446920871735,
"example_count": 80... |
4b82a217-8fc4-4205-9cb3-1e4fb97c5864 | easy | chad-atexpedient | 2026-08-27 03:31:16.178501+00:00 | succeeded | bf85189c8839892a2b288fe1ea7ab5238da6a01fb63a41169a18ebbf059228a9 | 18,658 | null | """X136: fixed-T-active first-transition denoising.
X135 moved Medium exactness to T2. X136 instead perturbs each encoded input state,
applies one tied transition, and matches the detached clean first transition. This
directly regularizes the leaderboard-critical map on every row while retaining
X131's corrected natur... | {
"id": "4b82a217-8fc4-4205-9cb3-1e4fb97c5864",
"created_at": "2026-08-27 03:31:16.178501+00:00",
"db_md5": "3323b4bf5fda0551e7ff552e419dde85",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "52ba8735-fd86-47de-b34e-376aeead335e",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.217065673138931,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010416666666666668
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.241479194752304,
"example_count": 600,... |
4b844997-d8e3-4e16-b2b3-9fc12c19cc63 | easy | amirmeisami | 2026-08-17 19:40:19.893625+00:00 | succeeded | 513b5a53d0fceb33d51e2b31e87abf668c54d2c9f69985c4d96b3db071b64afe | 12,658 | null | """T021 Arm D: algebraic position-aware recurrent network."""
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, assert_model_state
D_MODEL, POLY_WIDTH, IN... | {
"id": "4b844997-d8e3-4e16-b2b3-9fc12c19cc63",
"created_at": "2026-08-17 19:40:19.893625+00:00",
"db_md5": "85a4578d87e4cb8b9093e9a8f5a4744b",
"submitter": "Amir Meisami",
"github_login": "amirmeisami",
"run_id": "be165819-59a3-44ba-88d3-80f1f17486fd",
"tier": "easy",
"dataset_id": "e7",
"status": "s... | {
"score": {
"mean_loss": 4.176059668681624,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.046256684711113334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.651538109197849,
"example_count": 85,
... |
4b84b299-ec65-4a59-86cb-08f06b70da7a | easy | khushidahi | 2026-08-06 05:25:51.344471+00:00 | succeeded | e1f4e1d2a7afd9c29739bb8da5f4892cc59647cf51cddbd37dc806ff67cdb574 | 14,801 | 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": "4b84b299-ec65-4a59-86cb-08f06b70da7a",
"created_at": "2026-08-06 05:25:51.344471+00:00",
"db_md5": "82883afb4e9562f42c7d79e2b187c1bc",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "72d66e7a-2eea-49d1-8088-92c2fb41d24e",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.079130530357361,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006874999846331775
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.077475070953369,
"example_count": 1200... |
4b86a79d-ec14-4b81-914a-a500bbfdfcec | easy | sapient-sapiens | 2026-08-09 14:30:32.104060+00:00 | failed | 6bf4d4fd7ddeead05c0b9110c34eb0b021b8eb987a28340ca44a2306cd6bbc25 | 11,016 | null | """Near-500M adaptive recurrence grown from the hosted-working baseline."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
D_MODEL = 6400
NUM_HE... | {
"id": "4b86a79d-ec14-4b81-914a-a500bbfdfcec",
"created_at": "2026-08-09 14:30:32.104060+00:00",
"db_md5": "68da4a5fff2061e1ef98397d7ade0cac",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "cea9df84-b990-4c64-b922-0ca437b45660",
"tier": "easy",
"dataset_id": "e5",
"status": "fa... | null
|
4b8f1885-3407-4b55-9c6e-ae8653e5659b | easy | oupadhyay | 2026-08-25 07:03:36.331739+00:00 | succeeded | befbab1981219d58797083a0555780a6d04e96634477969d99daa632b43c09a9 | 4,092 | null | """Dynamic-width shared conditional quadratic recurrence."""
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
D,H=32,16
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_size,s.m... | {
"id": "4b8f1885-3407-4b55-9c6e-ae8653e5659b",
"created_at": "2026-08-25 07:03:36.331739+00:00",
"db_md5": "0921da4f1402e0703ddf168bd538e3a2",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "44feed98-8429-4796-82ca-34f0a1339a19",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.360641266414853,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009583333345750967
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.154431327041489,
"example_count": 600,... |
4b9013ed-6f7a-4d10-a9fb-e9099b4d55f4 | easy | MarcoF1 | 2026-08-14 01:42:39.497030+00:00 | succeeded | ba4a2b738925f28de5cf79ccf5bc191edb8617a18ca217f987a70e6b514d7627 | 17,095 | null | """Population-of-clocks: K independent residue-clock models trained in
parallel as batched weight tensors, ensembled at evaluation.
Rationale: hosted training is kernel-launch-bound, not FLOP-bound, so running
K=8 models costs almost no wall-clock. Each member gets its own init and its
own table-dropout probability (s... | {
"id": "4b9013ed-6f7a-4d10-a9fb-e9099b4d55f4",
"created_at": "2026-08-14 01:42:39.497030+00:00",
"db_md5": "5b1a93a727fc818e9aa8ff404ee32f2a",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "52998a74-fd2a-4f32-bfde-484675c2ec96",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.1903671121254638,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006041666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.23024902417402,
"example_count": 1200... |
4b98484a-6f00-403e-87aa-2746f82b74d4 | easy | erdavis0 | 2026-08-25 01:54:58.219967+00:00 | succeeded | 35ff275a7c933b6d0b5e1dc217d55253aa63433308fe14fd797471c38b44cb91 | 24,148 | null | """Matched categorical-boundary panel: C3 ST oracle selector.
Public delimiter syntax is used only to route opaque token identities onto
separate right-aligned tapes. One T-blind learned macro maps a categorical X
tape and a digitwise N tape to a new categorical tape through freshly reset
interaction, carry, and modu... | {
"id": "4b98484a-6f00-403e-87aa-2746f82b74d4",
"created_at": "2026-08-25 01:54:58.219967+00:00",
"db_md5": "5fb89a7c7ac87a7ea6e553c7a9be46f2",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "f780629c-cc82-4dee-8ed2-f2f63c1848d7",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 1.8705756664276123,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06036036275327206
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8511793613433838,
"example_count": 60,... |
4b99fe5d-697e-4306-8963-293646e6d77d | easy | lpbb | 2026-08-14 10:36:56.443357+00:00 | succeeded | bcd52612b15dd46fa9832d828f6e581c6074d2aedb359dadb775d3a2490237d6 | 9,137 | null | """CHEAT L1d — L1 (one squaring-mod step, hint given) trained with Grokfast:
does amplifying the slow gradient component move GENERALIZATION, not just fit?"""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpe... | {
"id": "4b99fe5d-697e-4306-8963-293646e6d77d",
"created_at": "2026-08-14 10:36:56.443357+00:00",
"db_md5": "27c3d33b033bad44eea54ad1306c5688",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "e5a62b3d-7674-4a9f-b3b0-e6bc7e8fef0d",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 6.524562866229989,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.11437500002793967
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.8279390122442125,
"example_count": 800,... |
4b9e5a1c-0166-480b-9260-ddf5f6c07031 | easy | G-AshwinKumar | 2026-08-13 20:38:05.711318+00:00 | succeeded | 73c52c1745201f21d97afbf391ec3bbd8f9733d874ed163ff5351f52fe881f50 | 28,296 | null | """Recurrent squaring operator with a discrete residue state between iterations.
Continuous recurrent states drift: small errors in the learned dynamics compound
over iterations, so a model correct at its training depth degrades when run
longer. The measured fix is to make the state between iterations *discrete* --
re... | {
"id": "4b9e5a1c-0166-480b-9260-ddf5f6c07031",
"created_at": "2026-08-13 20:38:05.711318+00:00",
"db_md5": "9df427e3b2b748d67ceeeba70c26b2f6",
"submitter": "Ashwin Kumar",
"github_login": "G-AshwinKumar",
"run_id": "9f57712b-1c69-43c4-842b-2b76761701dd",
"tier": "easy",
"dataset_id": "e3",
"status": ... | {
"score": {
"mean_loss": 2.1392724567016668,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00875
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1311390161213377,
"example_count": 800,
... |
4bac3910-4e45-49cb-8808-de37d9e2efa5 | easy | jordanrubin | 2026-08-07 16:12:26.486499+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": "4bac3910-4e45-49cb-8808-de37d9e2efa5",
"created_at": "2026-08-07 16:12:26.486499+00:00",
"db_md5": "671c15c2db39c821999bae96a7defa51",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "1535b1d0-264b-47eb-94db-11fd7337cd49",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 3.971229635533768,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333202948173
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.02164301917692,
"example_count": 600,
... |
4baf2573-45a6-48f3-9660-44c9d4508adb | easy | chad-atexpedient | 2026-08-14 13:21:54.988055+00:00 | succeeded | f626f5abeb5be73a4a23d4b7de56a3543c4f6b0be9f94633b78b1cb709dacb91 | 18,495 | null | """X18: E04 with compositional consistency loss.
Creates state-level supervision by matching T=2 intermediate states with T=1
final states when the T=2 output equals the T=1 input (string comparison only).
No arithmetic is computed - only evaluator-supplied values are compared.
"""
from __future__ import annotations
... | {
"id": "4baf2573-45a6-48f3-9660-44c9d4508adb",
"created_at": "2026-08-14 13:21:54.988055+00:00",
"db_md5": "f17e6236ccafc6e598d19504beb7d2db",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "c93283bd-b0d7-43ca-a082-25c1bc21672a",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.192641730150786,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666666666666
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.206438761655525,
"example_count": 600,... |
4bb6ba30-99e2-4906-b0fd-3568ceb8f546 | easy | rteehas | 2026-08-19 07:53:00.315686+00:00 | succeeded | 3d873c4965549d9ae383f4cac2a9d63c2664f991f434d42ab38abb5f2972d28c | 39,323 | 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": "4bb6ba30-99e2-4906-b0fd-3568ceb8f546",
"created_at": "2026-08-19 07:53:00.315686+00:00",
"db_md5": "c15ad75df0af85562b3241362bac4f00",
"submitter": "Ryan Teehan",
"github_login": "rteehas",
"run_id": "57bdca0e-0beb-4565-a14f-36d6135e26df",
"tier": "easy",
"dataset_id": "e1",
"status": "succee... | {
"score": {
"mean_loss": 7.4471964836120605,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333334140479565
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.647143840789795,
"example_count": 10... |
4bb8b6c1-f0bb-492f-a47e-408e7fe1e4cc | easy | khushidahi | 2026-08-20 02:09:14.106253+00:00 | succeeded | 49a2d3b90a9bb432235f11c9ea46f94bce99302d23694406d59379134b399cad | 37,338 | null | """R279: semantic quadratic-root carry/reduce machine.
This candidate targets functional-root recovery when the evaluator supplies
only composed endpoints such as T=2. It trains only from evaluator-owned final
labels and always executes the supplied recurrence depth. The model never
converts N or X to native numbers... | {
"id": "4bb8b6c1-f0bb-492f-a47e-408e7fe1e4cc",
"created_at": "2026-08-20 02:09:14.106253+00:00",
"db_md5": "96c4d7538b5453aa2d4a338708c14db2",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "bc25d53e-dd9f-4256-8ba3-a5166844bb11",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 2.2241938229691733,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000009313226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.208422931744163,
"example_count": 800... |
4bb93738-406d-4960-bf0b-c271492cf40d | easy | briheet | 2026-08-07 02:34:40.702564+00:00 | succeeded | 36859d5c72de853029ee07cb3fd1adc33b60cdb31ff0308436a438ac2eff1a44 | 3,878 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 96
NUM_HEADS = 4
c... | {
"id": "4bb93738-406d-4960-bf0b-c271492cf40d",
"created_at": "2026-08-07 02:34:40.702564+00:00",
"db_md5": "bbab0adab737275c3a576327c7933ad0",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "ab655e7c-0ae0-4021-9c52-d8e1660e84c5",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.7366042137145996,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.031666665337979794
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.9269282817840576,
"example_count": 10... |
4bc2992b-788f-43af-8282-925bff9e35a3 | easy | christopherhwood | 2026-08-05 14:13:56.387448+00:00 | succeeded | 1dc2b12fb617a32bb2c32d93800bb715beff009a616ba3da91321ced47c19bf2 | 17,758 | null | """Digit-register recurrent model for repeated modular squaring.
The learned outer cell is applied T times. T is used only as GPU-side control
flow; the learned path sees N and the current residue, but not the T tokens.
One outer cell contains four learned refinement phases over right-aligned digit
registers. No ari... | {
"id": "4bc2992b-788f-43af-8282-925bff9e35a3",
"created_at": "2026-08-05 14:13:56.387448+00:00",
"db_md5": "434320830923089d1a97af12e2f20fe4",
"submitter": "Chris Wood",
"github_login": "christopherhwood",
"run_id": "e207262a-b889-4d54-940f-0d741ca23740",
"tier": "easy",
"dataset_id": "e5",
"status":... | {
"score": {
"mean_loss": 8.079948656529456,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000037252903
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.3277668952941895,
"example_count": 600... |
4bc4b459-7069-4c98-823f-1f4287d6300c | easy | ArkinDharawat | 2026-08-13 17:10:40.634771+00:00 | succeeded | 7b2c60c7a75003826f7be86b252e106bb226b2b884eec771f8a09bb7c51ff7bc | 7,630 | null | """Universal-Transformer-style looped block, v2: step embed + pos offset + LR schedule + 6 loops.
Extends ``looped_ut_padded`` with four ideas drawn from
Graves 2016 (ACT), Dehghani et al. 2018 (Universal Transformer), and
Merrill & Sabharwal 2025 (log-depth transformers):
1. **Learned per-step (timestep) embedding**... | {
"id": "4bc4b459-7069-4c98-823f-1f4287d6300c",
"created_at": "2026-08-13 17:10:40.634771+00:00",
"db_md5": "3f38bcf16fa1567928756b9059a0d42a",
"submitter": "Arkin Dharawat",
"github_login": "ArkinDharawat",
"run_id": "ddb18c1f-a2ed-49c3-9228-0ba9656725ca",
"tier": "easy",
"dataset_id": "e3",
"status"... | {
"score": {
"mean_loss": 2.2799747371086516,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01875000004656613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3338491743133507,
"example_count": 800... |
4bd55e69-cd76-4b1e-a141-f0e7d69abae6 | easy | jordanrubin | 2026-08-13 17:51:57.305309+00:00 | succeeded | e2245c74dfcfe7d00f8aeee61890a07c07a0d89977c245e3844890ce7f77b424 | 27,175 | 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": "4bd55e69-cd76-4b1e-a141-f0e7d69abae6",
"created_at": "2026-08-13 17:51:57.305309+00:00",
"db_md5": "3f710550312ea469d13b008d83980242",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "c4122181-f3b1-4fcd-b46f-c7f55cdb830e",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.861979366927245,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.028333332234372694
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2232502843232433,
"example_count": 600... |
4be00b83-0bf2-4d35-b164-ac4e4e5371d3 | easy | 0xDEADD06 | 2026-08-12 13:45:10.820356+00:00 | succeeded | 33d47fc26ef1cd463298926cea20599a24d7d5e2746b9a134f26b9aa9e1d8bb9 | 3,834 | null | """CNN: 1D convolutional stack for algorithmic tasks."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn, Tensor
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 256
NUM_LAYERS = 4... | {
"id": "4be00b83-0bf2-4d35-b164-ac4e4e5371d3",
"created_at": "2026-08-12 13:45:10.820356+00:00",
"db_md5": "7dc894721ff651a27ab24810cbc46083",
"submitter": "0xDEADD06",
"github_login": "0xDEADD06",
"run_id": "49853180-ed93-486b-8c9b-c55e0faa416b",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 2.296796207215348,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002083333401630322
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2904319056213702,
"example_count": 600... |
4be07b8e-20dd-4b17-9fbe-566d8414d592 | easy | mnida | 2026-08-23 21:38:08.495843+00:00 | succeeded | 7d71e9f08ac4e5894cbbd83686f80a4c1e8346654953d3570682d3c885a09009 | 6,685 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
D_MODEL =... | {
"id": "4be07b8e-20dd-4b17-9fbe-566d8414d592",
"created_at": "2026-08-23 21:38:08.495843+00:00",
"db_md5": "72d9fe86c7c3551b8cd41e4cb202626e",
"submitter": "mnida",
"github_login": "mnida",
"run_id": "78e2fe64-6c86-4430-a9ae-1b80d4b21b9d",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.7636717557907104,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07027027569711208
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4956412315368652,
"example_count": 60,... |
4be20766-e280-49f9-9f5d-226a667fa137 | easy | oupadhyay | 2026-08-08 11:34:16.933374+00:00 | succeeded | 73884a7521b2ac75d4757ad068b0be29e0dcd0b791976908b7b38c259b8da6f1 | 9,001 | null | """Equal-mixture compressed-axial and factorized-pair place transition."""
import math
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle, Submission, assert_model_state
PAD, N, X, T, ANS, DIGIT = 0, 2, 3, 4, 5, 7
MAX_STEPS, C, D = 64, 32, 64
class Config:
def ... | {
"id": "4be20766-e280-49f9-9f5d-226a667fa137",
"created_at": "2026-08-08 11:34:16.933374+00:00",
"db_md5": "32c646b238fd219b11eda7ce925970c1",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "1125eaf9-30f5-410a-9fec-afa02b8ad8d9",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.161245893972656,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1494418306468313,
"example_count": 600... |
4bf192f2-0045-4943-a366-0a92f7bf8006 | easy | chad-atexpedient | 2026-08-04 21:05:02.172721+00:00 | succeeded | e6c1efb7dfa555597cf9331e57cd5877cf75eb673771f83e62fc52d64bde81cd | 3,884 | null | """R010: twelve applications of one shared Glint-style Transformer block."""
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 = ... | {
"id": "4bf192f2-0045-4943-a366-0a92f7bf8006",
"created_at": "2026-08-04 21:05:02.172721+00:00",
"db_md5": "0cfe0509bb90eb7fa989dbea87ee3433",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "43c2b33d-1ec0-4318-ab04-d20fffcb2d56",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 10.047175976815467,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 10.213527944590478,
"example_count": 60... |
4bf35807-b236-4508-90c2-bf78c77287e7 | easy | shan23chen | 2026-08-08 01:54:46.934994+00:00 | succeeded | 63e033670a1f7266950058e123662770999cd38a42088e31a4b933e0e1ceab3a | 20,440 | 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": "4bf35807-b236-4508-90c2-bf78c77287e7",
"created_at": "2026-08-08 01:54:46.934994+00:00",
"db_md5": "bf471bac266c9d3c1d2a06838974b65b",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "6fdc8f84-ba2e-4b63-affe-e565fd6b1407",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.6379934549331665,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00875000050291419
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.599524974822998,
"example_count": 600,... |
4bf5a48b-4ebc-4e47-8194-0c01e1c8ab0c | easy | erdavis0 | 2026-08-27 01:30:14.180323+00:00 | succeeded | 795df3f9652030e991e92998273fd7a4939c775c6b6c6c9395c3028f971d4e8b | 31,304 | null | """Typed-prefix reversible transition B4: oracle-weighted step training.
Public delimiter syntax routes opaque N, X, and T fields into three strict
functional roles. X initializes only an evolving auxiliary tape, N initializes
only an invariant environment lane, and T reaches only a learned monotone
prefix controller... | {
"id": "4bf5a48b-4ebc-4e47-8194-0c01e1c8ab0c",
"created_at": "2026-08-27 01:30:14.180323+00:00",
"db_md5": "86ddb36d905f79e81813767c4d60fc8f",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "b3491d07-199d-4f2e-b8a2-26d4e21ee91a",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",... | {
"score": {
"mean_loss": 5.955052035014736,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01125000001862645
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.990496827215019,
"example_count": 800,
... |
4bf6965d-9d22-4b1f-93a2-32ff48cd935f | easy | nikolageorgiev2000 | 2026-08-04 12:00:10.163094+00:00 | succeeded | c1b13be66e91880e4406b55e1c265bde4c276e3b124bf3e776c0db29627cb7dd | 26,218 | null | """T-composed categorical-digit reasoner with fully learned pair/reduction maps."""
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,
asser... | {
"id": "4bf6965d-9d22-4b1f-93a2-32ff48cd935f",
"created_at": "2026-08-04 12:00:10.163094+00:00",
"db_md5": "7dcea68a2c0c1f9bdc302fae41526398",
"submitter": "Nikola Georgiev",
"github_login": "nikolageorgiev2000",
"run_id": "19480a91-43d2-4f7d-8fe7-444545e2b322",
"tier": "easy",
"dataset_id": "e5",
"s... | {
"score": {
"mean_loss": 2.193757035512885,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0054166666790843015
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.193165434850171,
"example_count": 600... |
4bfb1cf8-ccbb-480a-a73e-5768f744f957 | easy | TZZheng | 2026-08-07 05:01:54.196435+00:00 | succeeded | 713525993c3317e54b09c3af123590efbfa2c17d83558c9785e4b48cd46e0bc7 | 12,527 | null | """Learned recurrent modular-transition model.
The mutable digit state is initialized from x. A learned square-oriented
self-mixing network and an N-conditioned reduction network form one transition,
and exactly the same transition parameters are reused T times. No arithmetic
operation is hard-coded; only the public p... | {
"id": "4bfb1cf8-ccbb-480a-a73e-5768f744f957",
"created_at": "2026-08-07 05:01:54.196435+00:00",
"db_md5": "61a1bdb9ed8a898453ec865004227fd9",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "9e6e6712-1b79-4e19-a050-f997f62d75b0",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 7.300878047943115,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03333333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.917374610900879,
"example_count": 100,
... |
4c06dbe8-eda0-4c5a-8990-5603fa6eac7c | easy | yashkant | 2026-08-19 01:30:21.912301+00:00 | succeeded | 862146a1e9acbe92c8744760279022cd13aebb4d9dfb3fb9bb96711be0e3c16d | 41,871 | null | """Round2223 two-independent-branch low-capacity ensemble.
This source keeps the complete Round1829 nonlinear branch architecture,
SAM/Compositional-Muon optimizer, EMA, loss, and evaluator contract. It
instantiates two independently initialized width-288 branches and always
returns their symmetric arithmetic-mean lo... | {
"id": "4c06dbe8-eda0-4c5a-8990-5603fa6eac7c",
"created_at": "2026-08-19 01:30:21.912301+00:00",
"db_md5": "277a40ae4c484e54a3b1e41decf19bf8",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "a426509b-ac53-429c-b7d7-f7198d527dc8",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 2.2233123779296875,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018333333544433117
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2453322410583496,
"example_count": 30... |
4c16ba9f-4dc2-45fd-a1f6-71fbf580bee9 | easy | gauravmishra | 2026-08-10 19:10:31.285365+00:00 | succeeded | bd4a80c6bac1379838daa2d2b73e6c3ae0e6f3e41b8da3e2398605150eefcae3 | 14,523 | null | """Capability reset: end-relative output queries with optional recurrent tape.
The only line changed between hosted Easy variants is ``VARIANT``. Every
variant is generic: it uses only input embeddings, positions, and the supplied
attention mask. It does not inspect token IDs, parse fields, or implement an
arithmeti... | {
"id": "4c16ba9f-4dc2-45fd-a1f6-71fbf580bee9",
"created_at": "2026-08-10 19:10:31.285365+00:00",
"db_md5": "c0890540e858ad423f787fffd32ea956",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "f6334103-f99e-47f5-a577-296992987dec",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.1562883853912354,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07999999821186066
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1469810009002686,
"example_count": 100... |
4c17bdfb-54ba-412b-80cd-f538d75cebc6 | easy | yashkant | 2026-08-17 01:06:14.207213+00:00 | succeeded | 9b4cfe8e0bf5b7cab3ca2ef38181134e6f5e70d2df37a63d1efa9605a7811cec | 43,554 | null | """Round2190 removes cross-step Dykstra correction carry.
The retained official carrier, primal recurrence, two learned projections,
antisymmetric set codes, parsed-T allocation, optimizer, and EMA are fixed.
The candidate reuses the learned initial A/B correction offsets on every
cycle instead of carrying hand-writte... | {
"id": "4c17bdfb-54ba-412b-80cd-f538d75cebc6",
"created_at": "2026-08-17 01:06:14.207213+00:00",
"db_md5": "0754f6fe90068b218bbcff849a9e897c",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "df93fa8f-3b43-43a3-89b0-7730f093c045",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.1515753508437063,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1573028735515782,
"example_count": 60... |
4c1b77b3-2258-42d6-b7a1-d9e458b35672 | easy | k-penchev | 2026-08-22 23:20:05.223314+00:00 | succeeded | 450bc14d47b28e1ec66f7b80a98dd1f7e15aa2176c755fb320f857024ae6b750 | 8,441 | null | """Gated width-96 recurrence optimized with self-contained Schedule-Free AdamW."""
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,
Subm... | {
"id": "4c1b77b3-2258-42d6-b7a1-d9e458b35672",
"created_at": "2026-08-22 23:20:05.223314+00:00",
"db_md5": "448720dadd0d64a27da871562f6b887b",
"submitter": "Kaloyan Penchev",
"github_login": "k-penchev",
"run_id": "24447464-c4f9-44b1-9074-88da044cbf0d",
"tier": "easy",
"dataset_id": "e3",
"status": "... | {
"score": {
"mean_loss": 2.2311944499664564,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000027939677
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2312724465926657,
"example_count": 80... |
4c1d9b1d-3b29-49ec-bb83-993f037289c7 | easy | karanganesan | 2026-08-04 01:00:33.978672+00:00 | succeeded | b11fe3c44169371832d1c46d4178604084fd4d983ceb54932302015e1d8a1f9c | 11,992 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "4c1d9b1d-3b29-49ec-bb83-993f037289c7",
"created_at": "2026-08-04 01:00:33.978672+00:00",
"db_md5": "ccbc15a40903511c310c88b5b634af5d",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "379227cd-04f7-48f4-8c8e-54af0ba0a221",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.056242823600769,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333334140479565
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.01247501373291,
"example_count": 100,... |
4c2ccc3b-b315-4a03-9728-cb85f36a3c64 | easy | DDanlov | 2026-08-10 08:04:34.922541+00:00 | failed | 357e87f0b074251124da419085477b247736219197e7146193c9b6102808e15d | 16,397 | 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": "4c2ccc3b-b315-4a03-9728-cb85f36a3c64",
"created_at": "2026-08-10 08:04:34.922541+00:00",
"db_md5": "20d6eb5989a4ea086cb6e62279a502a2",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "81bea115-07bc-406b-a344-cd097828ef61",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
4c2e8faf-97ed-401e-b7e8-f141a08f2f4d | easy | Cree0618 | 2026-08-29 07:21:21.057699+00:00 | succeeded | 45e8da47932c79bfb624a78503361247e85728cda49f14af1d27590eea262131 | 6,060 | null | """E017: E007 untied-head model with T-aligned recurrence.
Single-variable ablation on E007: replace the fixed four shared-block
applications with one application per requested squaring step T, parsed
from the prompt's `T <digits>` field. Rows finish early via torch.where
freezing, so a batch loops to max(T) and each ... | {
"id": "4c2e8faf-97ed-401e-b7e8-f141a08f2f4d",
"created_at": "2026-08-29 07:21:21.057699+00:00",
"db_md5": "06dadf6680eba74478624813cdce968e",
"submitter": "Cree0618",
"github_login": "Cree0618",
"run_id": "d2b766f2-ae84-47dc-b708-1ffa6ee03130",
"tier": "easy",
"dataset_id": "e5",
"status": "succeede... | {
"score": {
"mean_loss": 6.623894602593384,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.325232418128728,
"example_count": 600,... |
4c39872c-f13c-4b93-93a4-3f48c1363e23 | easy | oupadhyay | 2026-08-29 18:51:37.332571+00:00 | succeeded | 3d8b8825fc86f252d407f3c4d759c62452cf9dafd53f48ca4d39d611c89c42e1 | 4,206 | null | """Conditional structured-loss screen: smooth05."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,TokenLossBatch,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,ma... | {
"id": "4c39872c-f13c-4b93-93a4-3f48c1363e23",
"created_at": "2026-08-29 18:51:37.332571+00:00",
"db_md5": "c1a751d04471b2ecc91007eee85af6bf",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "82aec705-10f4-4e58-a829-9bf96a555bad",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 1.9051340818405151,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0468468489125371
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.87375807762146,
"example_count": 60,
... |
4c3db99f-0a3b-4bce-b2d0-95c2e562f85a | easy | 0xDEADD06 | 2026-08-12 13:49:30.475168+00:00 | succeeded | bd691ba713ead65fd24ecd37f9d9faa289eebb71b4ddc13a43732ec31fd2f7a8 | 3,883 | null | """CNN: 1D convolutional stack for algorithmic tasks."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn, Tensor
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 256
NUM_LAYERS = 4... | {
"id": "4c3db99f-0a3b-4bce-b2d0-95c2e562f85a",
"created_at": "2026-08-12 13:49:30.475168+00:00",
"db_md5": "90e5390cf7b2306fcbe0b7f0bdbd6d6a",
"submitter": "0xDEADD06",
"github_login": "0xDEADD06",
"run_id": "98eee3bf-e45e-4871-976a-61f86046ba7b",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 3.729916854475481,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013750000409781932
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.6756956171027215,
"example_count": 600... |
4c463054-6d14-47ea-a4da-58835691423e | easy | karanganesan | 2026-08-05 12:21:45.605504+00:00 | succeeded | 646a054ea7deeddec66b19823252ac47c4f3f282f2324bdf5771af44145e8b73 | 22,063 | 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": "4c463054-6d14-47ea-a4da-58835691423e",
"created_at": "2026-08-05 12:21:45.605504+00:00",
"db_md5": "214184c3194912fec1e2b1b37cdc453c",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "abc4fa24-e62a-44c7-abde-e251a1bd2569",
"tier": "easy",
"dataset_id": "e3",
"status": ... | {
"score": {
"mean_loss": 2.6147106885910034,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009374999906867743
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.047579288482666,
"example_count": 800... |
4c47f33c-116d-49ee-a644-415c7331d1a9 | easy | chad-atexpedient | 2026-08-09 12:41:43.030334+00:00 | succeeded | 1d906af061568e31a7ac6f840c5d03636073fa92cf291cb7a7f6b81a8e4324f4 | 18,204 | null | """X10: pair register with smooth worst-digit sequence loss.
The mutable register is initialized from x once. Every subsequent application
of the tied cell receives only the current LSD-first digit state and immutable
N digits. Requested T controls only the number of applications. The state
logits are also the ans... | {
"id": "4c47f33c-116d-49ee-a644-415c7331d1a9",
"created_at": "2026-08-09 12:41:43.030334+00:00",
"db_md5": "4d97c530c50e66045d7809a50a11564c",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "d42cfe60-ff8c-4c64-a2f8-9f499568a5a3",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.7197746601257773,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000080714623
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4073848724365234,
"example_count": 60... |
4c496a9e-805c-4dfb-9379-f7ee419397ae | easy | khushidahi | 2026-08-07 02:58:24.034417+00:00 | succeeded | 59cf762f0d8386f960d0c286432239f69f2e8d485b69ce71bcd06c5c753214e4 | 18,220 | 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": "4c496a9e-805c-4dfb-9379-f7ee419397ae",
"created_at": "2026-08-07 02:58:24.034417+00:00",
"db_md5": "ed17dfc91d22ccd05803eca204a60136",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "b49d946e-4283-4cef-92bd-3f1c8c4260b5",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1985669136047363,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666772216558
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1976749897003174,
"example_count": 60... |
4c497f89-2ebe-4fba-a5d9-ecdb7e8cb8bf | easy | shan23chen | 2026-08-08 03:24:07.480994+00:00 | succeeded | 9758c456215d8c6d5dffd7918423e11a09188df64fb7a51b5271bff180f8ecf9 | 25,144 | 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": "4c497f89-2ebe-4fba-a5d9-ecdb7e8cb8bf",
"created_at": "2026-08-08 03:24:07.480994+00:00",
"db_md5": "a577eff796af4add230a29a657ad2e6f",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "4df18de2-fc37-4989-bb67-4067ffac677a",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.9175904989242554,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166667005047202
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.749485969543457,
"example_count": 600... |
4c4b5db8-6cdf-40a1-afec-3a11fec12c8c | easy | sapient-sapiens | 2026-08-04 17:12:14.353682+00:00 | succeeded | 1e6d90279bf3c2d05121303bc3138928233d9ae9ddb7ec13104482f1b406ccd4 | 18,638 | 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": "4c4b5db8-6cdf-40a1-afec-3a11fec12c8c",
"created_at": "2026-08-04 17:12:14.353682+00:00",
"db_md5": "fce18abc8f30cf1c5e1479fa5d786f76",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "d5eec93d-1552-4c1c-989c-1a3b62216a90",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 2.010233463194602,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04666666651765505
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0182700157165527,
"example_count": 100,... |
4c50dda8-505d-4579-969f-8c6303b5b8f3 | easy | adwhit | 2026-08-13 19:54:39.989664+00:00 | succeeded | d1aec503006f41febf80947ffee49ed62738588f5012287713d3dfbf99e4e479 | 14,647 | null | """GaussNet: additive-character recurrent model for One Layer Deeper.
Structure (all parameters learned from random init):
- Learned parse: cross-attention queries extract LSB-first digit registers
for x, copy-supervised from the prompt's own tokens via the auxiliary
channel (self-supervision, no task math).... | {
"id": "4c50dda8-505d-4579-969f-8c6303b5b8f3",
"created_at": "2026-08-13 19:54:39.989664+00:00",
"db_md5": "94a37bf677ea0f93db3b93480a2bcf4c",
"submitter": "adwhit",
"github_login": "adwhit",
"run_id": "633fa946-9878-4c2d-a2d0-21f873fea848",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.8162745237350464,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.81801176071167,
"example_count": 300,
"exact... |
4c51426b-3ae7-41b3-82b0-0177c769ecf1 | easy | Koenneker | 2026-08-04 14:42:37.599848+00:00 | succeeded | fee387cca6c7e0403d800583b2caf6f881aba8f546601e30ae71e799cb4aa59e | 3,194 | null | """Vanilla single-layer LSTM with PyTorch AdamW at batch size 200.
Identical to `submissions/baseline_lstm` except for `batch_size=200`.
E1 has 600 training examples and the manifest sets `drop_last: true`, so
the manifest default of 512 uses one batch per epoch and discards 88
examples (15%) every epoch. 200 divides ... | {
"id": "4c51426b-3ae7-41b3-82b0-0177c769ecf1",
"created_at": "2026-08-04 14:42:37.599848+00:00",
"db_md5": "e103231cbe922a8a1845a6e6dc7b1f3e",
"submitter": "Ferdinand Könneker",
"github_login": "Koenneker",
"run_id": "855539c8-6610-4a60-bb97-5c7a04f5c24d",
"tier": "easy",
"dataset_id": "e1",
"status"... | {
"score": {
"mean_loss": 4.751608848571777,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06833333149552345
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.563373565673828,
"example_count": 100,
... |
4c5319a4-5e94-4413-bfb4-abd4bcfec8d3 | easy | gauravmishra | 2026-08-10 08:31:29.807077+00:00 | succeeded | e00d373cf6796980c4cdd7c8f7f0f9e71b6f569c771f1d75a6ec26822b674861 | 7,358 | null | """E48: deterministic bidirectional ALiBi on historical K13-S385.
Architecture parent SHA-256:
42dc862219390b8300e98260725c6c2621b27aae32f750f81902dea4bf95889d
ALiBi prototype SHA-256:
6e7e9cf562fc43db879027703bfa9eafd73054f0ae815bd9fae1a2d4a1065aef
"""
from __future__ import annotations
import torch
import torch.nn... | {
"id": "4c5319a4-5e94-4413-bfb4-abd4bcfec8d3",
"created_at": "2026-08-10 08:31:29.807077+00:00",
"db_md5": "7e3a3f81faccc30dab367c18bf54c948",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "6cd0edb6-0799-447b-9aef-6cea968912c4",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.4523298926126236,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5420737052712203,
"example_count": 60... |
4c53cdd9-4a87-4a38-8312-168594a94c4c | easy | erdavis0 | 2026-08-26 03:17:14.796678+00:00 | succeeded | 6fe6fdf46bcdc9c76ffef64fbdc6b302f2262f8879ac4d7daea4e62dccf8b390 | 30,354 | null | """Latent-role continuous REV C5: hybrid token CE and sequence mixture.
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 and... | {
"id": "4c53cdd9-4a87-4a38-8312-168594a94c4c",
"created_at": "2026-08-26 03:17:14.796678+00:00",
"db_md5": "14ff631f83062f36483f24b1bd94c4c9",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "1e9c4595-fb59-4c2a-829e-37f8889ceb7f",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",... | {
"score": {
"mean_loss": 3.437997496751026,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012916666691501936
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.447696385362223,
"example_count": 600,... |
4c572269-ee05-4e2f-b079-daf5664ef701 | easy | isaac0804 | 2026-08-07 15:58:35.358517+00:00 | succeeded | 2fce75519d220cb3c9bbd177b86e8c055c6718556c8a553bda31e73056984e72 | 5,701 | null | """Clean variable-depth loop: shared block applied T times, NO damped gate,
NO step-embedding. The two components individually proven (local, this
session) to block single-squaring learning (gate floors loss at ~1.9;
step-emb halves exact), while the bare loop wrapper alone fits the primitive
at plain-block quality ((0... | {
"id": "4c572269-ee05-4e2f-b079-daf5664ef701",
"created_at": "2026-08-07 15:58:35.358517+00:00",
"db_md5": "ab31c5f467da49446c08540cfdf68b37",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "fa2cb1b5-f6f1-41d3-ae93-3e51f0ddddc7",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 3.713789579526474,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011875000158324836
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1819988207803034,
"example_count": 800... |
4c64a0b9-0202-4ee8-9d7d-e1d8ddbfc38a | easy | ehonig | 2026-08-14 00:38:42.986805+00:00 | succeeded | fca47cafd987b653a91415f87795891f8602e53f86845c98ddaf02009785b1ce | 48,079 | null | """DigitLoop — a place-value recurrent model for repeated modular squaring.
The task is ``y = x^(2^T) mod N`` with the answer read back tail aligned, one
decimal place value per token position from the right. A stack of attention and
MLP layers has no primitive for the two operations this needs — a partial-product
co... | {
"id": "4c64a0b9-0202-4ee8-9d7d-e1d8ddbfc38a",
"created_at": "2026-08-14 00:38:42.986805+00:00",
"db_md5": "c53ba6fe90c21436e6bf779a285b0dc6",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "1f47ad3d-34e4-421d-865d-077e2928ea41",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 5.132465124130249,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07666666805744171
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.516584873199463,
"example_count": 100,
... |
4c676063-cfb1-46ad-a4e7-d4e6e387ba90 | easy | 0Chris5R | 2026-08-01 07:44:10.580323+00:00 | succeeded | 5fc47f8c1ea8ce4dad89999a7f454944feb0f329f343cfee2e8553c566b2b778 | 23,575 | null | """Shared-digit Neural GPU with late full-objective EMA evaluation.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the sam... | {
"id": "4c676063-cfb1-46ad-a4e7-d4e6e387ba90",
"created_at": "2026-08-01 07:44:10.580323+00:00",
"db_md5": "4a085cf5f7ea55663149f5fbcafe9c8d",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "645aac6e-493b-4e79-a730-6a4cba5daf2d",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.7337905789603165,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.027916666679084302
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5397806531110687,
"example_count": 60... |
4c681faa-7aab-4520-8658-d51dd057bc88 | easy | gauravmishra | 2026-08-09 21:52:46.494342+00:00 | succeeded | 42dc862219390b8300e98260725c6c2621b27aae32f750f81902dea4bf95889d | 6,079 | null | """Compound-scaled Transformer with distinct blocks and PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_st... | {
"id": "4c681faa-7aab-4520-8658-d51dd057bc88",
"created_at": "2026-08-09 21:52:46.494342+00:00",
"db_md5": "04065419cbd1372225c17a397da38713",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "4769a5d4-f470-49e1-a64c-b4f735aec6b9",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.400381088259545,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4640779185188193,
"example_count": 600... |
4c6c65ca-63d0-4699-b858-b6be87d0d09f | easy | khushidahi | 2026-08-15 14:48:17.641645+00:00 | succeeded | 785ba79d6e5e0727b72cc4541a3a81ce6b564254b1affda04b3a67fc82e29b9c | 16,527 | 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": "4c6c65ca-63d0-4699-b858-b6be87d0d09f",
"created_at": "2026-08-15 14:48:17.641645+00:00",
"db_md5": "6f56267b313b94f67c334d2678192296",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "2bcc4a2e-11b3-4b43-ac9f-151e70ae4474",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1840227842330933,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0050000002374872565
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.182644844055176,
"example_count": 60... |
4c714bb5-45e1-42c5-a803-cad7f48b0b4b | easy | sapient-sapiens | 2026-08-18 21:22:20.117125+00:00 | succeeded | 06b55d65bbd7ba59c01084cc0f014726712d75d32e9aa2fbf258bec2d7cb90dd | 31,893 | null | """Depth-quantized recurrence with a mixture-over-depth objective.
The recurrent state is a bank of right-aligned digit slots. Every operator
application is followed by a soft quantization back onto the token simplex, and
the digit logits produced by that quantization are the answer logits at that
depth. There is no s... | {
"id": "4c714bb5-45e1-42c5-a803-cad7f48b0b4b",
"created_at": "2026-08-18 21:22:20.117125+00:00",
"db_md5": "e54dd2e99d9c3ad95120d2603b535e7e",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "01071184-5869-4c72-be96-8738aac4e4bf",
"tier": "easy",
"dataset_id": "e9",
"status": "su... | {
"score": {
"mean_loss": 2.3137218952178955,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.017777778208255768
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3040921688079834,
"example_count": 90... |
4c7cc7d6-7488-4f74-996e-cbc6f54425d2 | easy | khushidahi | 2026-08-21 04:08:17.791841+00:00 | succeeded | 341594dc9ed8a33cb21a9dfa2ca2e50636fba93a3a7b88ee1670ea101a8d607d | 28,689 | null | """R338: local semantic-T Transformer with a learned ordered digit axis.
This submission tests the strongest rules-safe generic baseline suggested by
the public competition discussion. It does not implement squaring, modular
reduction, or any other task solver. Numeric prompt fields are only rearranged
onto equal-le... | {
"id": "4c7cc7d6-7488-4f74-996e-cbc6f54425d2",
"created_at": "2026-08-21 04:08:17.791841+00:00",
"db_md5": "a2740687e6df1a0466719c9914d77aae",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "cba82334-d354-40ec-89e7-814b21f4505b",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 0.9187978506088257,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.8166666924953461
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.5410994291305542,
"example_count": 60,
... |
4c7e7e61-1482-4d21-a734-376bf2990df0 | easy | TZZheng | 2026-08-06 21:13:02.884881+00:00 | succeeded | ef4a82d9d5b384d9825eda4333db737ff33346e7bf74adce91734b578d908509 | 7,316 | null | """Self-contained standard Transformer scaling-sweep submission."""
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,
)
VARIANT = {'block... | {
"id": "4c7e7e61-1482-4d21-a734-376bf2990df0",
"created_at": "2026-08-06 21:13:02.884881+00:00",
"db_md5": "b7860fc155dfc9af985d679b586609e1",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "92aa21e1-04ee-439a-9828-4375e55a89dd",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.8921883702278137,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06500000134110451
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8644767999649048,
"example_count": 100... |
4c863966-233f-46f9-ab18-5dfd7aa2a7f4 | easy | shirvani-jr | 2026-08-23 00:02:05.714318+00:00 | succeeded | c097e127c148cd71ab3c0b348ff16d2ba974c6c606a059d32a0e206d7b62253c | 22,358 | null | """Looped digit machine with learned primitive tables.
Architecture: the recurrent state is the decimal digit string of the current
residue and one loop application performs one modular squaring, so depth
extrapolates in T by construction. The dataflow — schoolbook digit-pair
multiplication folded through the constan... | {
"id": "4c863966-233f-46f9-ab18-5dfd7aa2a7f4",
"created_at": "2026-08-23 00:02:05.714318+00:00",
"db_md5": "e943afc2de76b7c8faeea3eb405d5a5f",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "893bdf9d-0a83-4915-a9e4-c4d47c68ce62",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 0.02058847770456703,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.020541224628686905,
"example_count": 600,
"... |
4c8776a8-da60-4024-9536-71cf295a65da | easy | mcleish7 | 2026-08-23 19:04:08.073076+00:00 | failed | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "4c8776a8-da60-4024-9536-71cf295a65da",
"created_at": "2026-08-23 19:04:08.073076+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Sean McLeish",
"github_login": "mcleish7",
"run_id": "b3691b4c-8944-41fb-bd74-2fa8bb5c5301",
"tier": "easy",
"dataset_id": "e6",
"status": "fail... | null
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.