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 |
|---|---|---|---|---|---|---|---|---|---|---|
3d3f21ae-a7de-4be0-9da3-c81ace5999e7 | easy | newjordan | 2026-08-19 23:38:33.411561+00:00 | succeeded | 5bd7955d4a20ed4b16bf3e95bca83b89989a7331f93942988d0c0aa4fb2981af | 138,612 | null | from __future__ import annotations
import math,os,time,torch,torch.nn.functional as F
from torch import Tensor,nn
from benchmark import ModelSpec,OptimizerBundle,OptimizerSpec,Submission,assert_model_state
R_DIGITS=int(os.environ.get('CRAWLER_R_DIGITS','8'))
N_DIGITS=R_DIGITS
MAX_T=64
LABEL_PRODUCT_DIGITS=R_DIGITS+1
LA... | {
"id": "3d3f21ae-a7de-4be0-9da3-c81ace5999e7",
"created_at": "2026-08-19 23:38:33.411561+00:00",
"db_md5": "947b7335706b1bc198ec67d621c76599",
"submitter": "Frosty40",
"github_login": "newjordan",
"run_id": "b1a605b4-766f-49ad-906d-268b65b752be",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 0.0005723506915322587,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0005713357637644717,
"example_count": 600,
... |
3d50641e-a501-41fd-b831-e80b17e288dc | easy | jordanrubin | 2026-08-11 01:26:36.184745+00:00 | succeeded | 67982e9f505a2cf74a8823191467477f81e0c11b7660f6c116f0fafa565fb3eb | 21,669 | 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": "3d50641e-a501-41fd-b831-e80b17e288dc",
"created_at": "2026-08-11 01:26:36.184745+00:00",
"db_md5": "5f41f74e5ab605af5d290f14916e4deb",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "7311d999-ba6c-4627-8828-ddc063e90a15",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.417020025771926,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666489715379
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.16344451169262,
"example_count": 600,
... |
3d554083-b34a-451a-b49e-614a765d0e79 | easy | 0Chris5R | 2026-08-01 07:34:48.575489+00:00 | succeeded | 8e8116d29c0552dc2d7e4d4b65c261a6639de56d7f14fe3ccda576883b32f59e | 23,577 | 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": "3d554083-b34a-451a-b49e-614a765d0e79",
"created_at": "2026-08-01 07:34:48.575489+00:00",
"db_md5": "4b11f7414f45a5609be7fe3e2b767d7c",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "34e0325e-1d16-4e84-a457-9a4db89c2547",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.6108970634572444,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03250000001241764
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3302303761110177,
"example_count": 600... |
3d56a1a1-d373-4433-bd48-d50fcd04d6a5 | easy | DDanlov | 2026-08-31 20:29:58.889470+00:00 | succeeded | ca3bad44d796c2fc5906a8bec3ffb0adcbdda6f349daf8b19e4d760689d72560 | 32,407 | null | """
Unified Continuous Hopfield Equilibrium Propagation Deliberation Engine
=======================================================================
Architecture:
- Deliberation: Continuous Hopfield Multi-Layer Energy System (L=4, D=768, d_ffn=2304, H=16)
- Integration: In-Place Strang-Split Conformal Symplectic Velocit... | {
"id": "3d56a1a1-d373-4433-bd48-d50fcd04d6a5",
"created_at": "2026-08-31 20:29:58.889470+00:00",
"db_md5": "4bc1c00f59046c3ed4dec9b94bf519db",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "3cf5e904-e362-42bf-914a-e47099fc3959",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.408804624133473,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00875
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4127452293262794,
"example_count": 800,
"... |
3d5f1761-4b2a-4dcb-b0de-d5340d624b89 | easy | DDanlov | 2026-08-22 23:57:09.401348+00:00 | succeeded | f4b3f0bf50757a74fdbd38263d27e3df074a7c7123f221479297b316fa4d214e | 28,898 | null | """
Unified Hybrid Masked Deliberation Model (75.8M Parameters) - Variant 2 (Pure Skipless, No Cross-Attention)
Architecture:
- Scale: dim=1536, num_heads=24, d_ff=6144 (Exactly 75,826,176 parameters)
- Hybrid Prefix-Bidirectional + Causal Target Attention Mask
- Token Layout: [Prompt (L_p), 4 Pause Registers, Target T... | {
"id": "3d5f1761-4b2a-4dcb-b0de-d5340d624b89",
"created_at": "2026-08-22 23:57:09.401348+00:00",
"db_md5": "d7a80acebc87cfd0655a182b901f3766",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "9d113f3b-6155-440e-9fe6-ae12d5b3dd18",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.9223400810221367,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0016666666915019354
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.7586044933763856,
"example_count": 6... |
3d62a371-310f-4858-bc29-1a727d878675 | easy | Ranoobaba | 2026-08-20 03:23:15.302019+00:00 | succeeded | 95af73714fb41ffe9f3a2aac17adb1c30081c92a1085bcd2c756b78d6aea103d | 6,854 | 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": "3d62a371-310f-4858-bc29-1a727d878675",
"created_at": "2026-08-20 03:23:15.302019+00:00",
"db_md5": "357e8a27672954dcf4ddedecea26a9e8",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "cc543b3d-a3a5-438f-83c9-c39427d48593",
"tier": "easy",
"dataset_id": "e3",
"status": "... | {
"score": {
"mean_loss": 6.607710915168354,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.745126411099368,
"example_count": 800,
"exa... |
3d6683fb-5b00-4163-8bd3-9a0062d2146f | easy | lpbb | 2026-08-21 13:18:23.529881+00:00 | succeeded | 69b29e5a6a9fb7054817c1b1ec8bfbdd47261be2085ab750a05ab64b587792ad | 12,545 | null | """S0 + RoPE + abacus + looped block, with *input-adaptive* depth: instead of
a fixed `num_repeats`, each sample's T is parsed from its own token stream
(the digits after the `T` marker) and the shared block of `num_layers` layers
is applied exactly T times for that sample. One block pass = one modular
squaring step, s... | {
"id": "3d6683fb-5b00-4163-8bd3-9a0062d2146f",
"created_at": "2026-08-21 13:18:23.529881+00:00",
"db_md5": "636773373a8aed221f4253b2deae58f5",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "b84571a6-0842-4a6d-90b0-c1b9029a2e9a",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 4.750730973526724,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06166666492819786
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.088390827178955,
"example_count": 100,
... |
3d69599e-a9ed-4001-948f-51e7a156e742 | easy | Yalyenea | 2026-08-12 07:14:19.335992+00:00 | succeeded | 1fd5b88d1cd8820b97ed8a397ef89c59de2f32c93c36e896b94f2545e2b9d588 | 10,755 | null | """Compact right-aligned decimal recurrence with one-step supervision."""
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,
as... | {
"id": "3d69599e-a9ed-4001-948f-51e7a156e742",
"created_at": "2026-08-12 07:14:19.335992+00:00",
"db_md5": "2f8694957b4172ba8428f9fe09564ffa",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "f494769f-aeb2-4e90-a93d-c832a5127403",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 6.1049740118850595,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014166666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.8420640962540835,
"example_count": 60... |
3d697ba1-d39e-4f40-a10f-b4c5d6e71bb7 | easy | DDanlov | 2026-08-24 20:55:46.304648+00:00 | succeeded | ec5ed5fe890da88502f60700a7b382b60f4d580e64065ca22a7eadde66178f3a | 46,195 | null | """
Unified Learned Query Slot Deliberation Model (91.2M Parameters) - Pure Skipless No-Norm Slot Transformer
Architecture:
- Deliberation Core: dim=1536, num_heads=24, d_ff=6144 (Pure Skipless SO(2)^{D/2} Givens Recurrence)
- Token Layout: Recurrent loop processes ONLY [Prompt (L_p), 4 Pause Registers] (Zero target to... | {
"id": "3d697ba1-d39e-4f40-a10f-b4c5d6e71bb7",
"created_at": "2026-08-24 20:55:46.304648+00:00",
"db_md5": "b791f6f5117bf9f6cd45451421e54979",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "12a3ceca-ad81-4adc-b2d3-c183cb72b5f8",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.6906606279961007,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7385224323101642,
"example_count": 60... |
3d72630d-a3cd-4351-8160-a64d4b8e8154 | easy | shirvani-jr | 2026-08-29 02:43:50.083738+00:00 | succeeded | 93a1b471fddf1c663aa5c404de9da713f22babe968ac16312fd37530b2697706 | 49,652 | null | """Adaptive learned discrete recurrent fabric for One Layer Deeper.
A gradient-trained, operation-free recurrent computational machine. The
model learns, from the competition's endpoint supervision: how to parse the
prompt into a bank of categorical digit registers, a reusable transition
circuit applied serially, a l... | {
"id": "3d72630d-a3cd-4351-8160-a64d4b8e8154",
"created_at": "2026-08-29 02:43:50.083738+00:00",
"db_md5": "47e25796f314d531ac9bda32d631c27b",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "5737f9e0-3514-4dbc-aa45-2532e4d85f6d",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.837656319141388,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06499999947845936
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.7731575965881348,
"example_count": 100,... |
3d731629-c553-4306-a207-4ebefe1adf3b | easy | DDanlov | 2026-08-28 18:37:00.467329+00:00 | succeeded | bd34940d1994556665c5744f375de588956a50ae2385910b23568edd0c8181f8 | 15,209 | null | """
Model A Variant: Baseline Arch1 (K=1, continuous z0 + zt injection, DKS, RohanShampoo)
Config: D=512, H=8, d_ff=2048, T_rec=64, lr=0.0025
"""
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
... | {
"id": "3d731629-c553-4306-a207-4ebefe1adf3b",
"created_at": "2026-08-28 18:37:00.467329+00:00",
"db_md5": "b26a029df6845aebe9538f1da599b607",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "22ed256f-ca24-449f-9c12-323a495b7a97",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.328933045064133,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.33771060520758,
"example_count": 600,
... |
3d8c23f1-ebb9-40cf-8751-914d63b8b7aa | easy | MarcoF1 | 2026-08-13 08:22:48.342329+00:00 | succeeded | fbe213b38fad5332ced983962119ff0670c520f28a36e11de082b55e17c1ea24 | 14,397 | null | """Residue-clock model for fixed-N generalization.
Representation: parsed integer values of x and N feed (a) exact residue-phase
features over a bank of small integer periods, (b) learned double-hashed value
embeddings (per-residue identity), and T feeds cyclic phase features
(T mod 2..48) plus saturating channels — p... | {
"id": "3d8c23f1-ebb9-40cf-8751-914d63b8b7aa",
"created_at": "2026-08-13 08:22:48.342329+00:00",
"db_md5": "498f495d8a0b910efb7ca633ca25eee2",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "673aebc3-d78a-438f-97f2-e9a12584d903",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 3.1369026361260923,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.10166666711370151
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.7759416103363037,
"example_count": 100... |
3d8d33ce-b455-4467-9fdb-49e095f21148 | easy | adwhit | 2026-08-14 07:37:23.892667+00:00 | succeeded | f1cd4f1523ee4209806ca32414e4a623d8c1c6720135e03d2f3b6d6674a23821 | 14,917 | 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": "3d8d33ce-b455-4467-9fdb-49e095f21148",
"created_at": "2026-08-14 07:37:23.892667+00:00",
"db_md5": "4de04874d4c9d19755141277a079a004",
"submitter": "adwhit",
"github_login": "adwhit",
"run_id": "0e247b3e-2a7e-4c72-b97e-2ccf4e8788fe",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.501070261001587,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5043156147003174,
"example_count": 300,
"exac... |
3d9ec184-77c2-4d0c-be99-5185b7ab70d8 | easy | velocizapkar | 2026-08-06 21:50:26.408995+00:00 | succeeded | 1f9cc8aee2d18b6da2e783e564bfb8fac6983382c52459ff94cba052ec9b761c | 53,084 | 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": "3d9ec184-77c2-4d0c-be99-5185b7ab70d8",
"created_at": "2026-08-06 21:50:26.408995+00:00",
"db_md5": "f0a3b50c86e6594ed8facd75b0fe98f3",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "801921fd-2e8f-4f08-ab99-81cb62b0de8d",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 3.055337429046631,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014374999329447746
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.0896196365356445,
"example_count": 800... |
3da3d398-2997-440a-9c8f-a9c93897cceb | easy | MarcoF1 | 2026-08-14 18:18:24.003365+00:00 | succeeded | 5f9400db499673e08fdc956d355272d7c9d5d301dafe45e0df2f2a20e60cf15d | 15,558 | null | """Residue-clock model for fixed-N generalization.
Representation: parsed integer values of x and N feed (a) exact residue-phase
features over a bank of small integer periods, (b) learned double-hashed value
embeddings (per-residue identity), and T feeds cyclic phase features
(T mod 2..48) plus saturating channels — p... | {
"id": "3da3d398-2997-440a-9c8f-a9c93897cceb",
"created_at": "2026-08-14 18:18:24.003365+00:00",
"db_md5": "59b005e9ef7169f8ff423eca3cec89f5",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "aece0668-ec34-44b2-b9c6-6e141a070d23",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.1155192852020264,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.021666666492819786
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0115432739257812,
"example_count": 10... |
3dac8391-c5e1-4903-8452-8e98bc0c6343 | easy | yashkant | 2026-08-09 02:32:21.293133+00:00 | succeeded | da39b642da1afd6e1de460cec068bfa4cf1d92197848ffb327a0255903cfa95d | 22,836 | null | """Round 211 common TRM-style x/y/z shared-transition template."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
from torch import... | {
"id": "3dac8391-c5e1-4903-8452-8e98bc0c6343",
"created_at": "2026-08-09 02:32:21.293133+00:00",
"db_md5": "d191e6ac3e82ab0f77439af3060169fc",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "cd451bd7-a218-4975-bf92-a95257c2a840",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 4.372560083866119,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.3012500088661909
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.03543758392334,
"example_count": 300,
... |
3daf6286-9b28-416b-8d03-d37cdeb7fc2b | easy | nikolageorgiev2000 | 2026-08-04 00:32:31.741742+00:00 | succeeded | 05f4fa32e888cf1e9ad30077c10a2941fe5f038f079a807a9bd6912b9a054b99 | 8,688 | null | """K=1 recurrent graph reasoner with explicit differentiable writable memory."""
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_m... | {
"id": "3daf6286-9b28-416b-8d03-d37cdeb7fc2b",
"created_at": "2026-08-04 00:32:31.741742+00:00",
"db_md5": "7e1d1c8ae322ba9b9a3a6be4bf3babe6",
"submitter": "Nikola Georgiev",
"github_login": "nikolageorgiev2000",
"run_id": "28062f56-50d7-4965-9d62-d11a838efa17",
"tier": "easy",
"dataset_id": "e1",
"s... | {
"score": {
"mean_loss": 3.4016566276550293,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.053333332762122154
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.2245967388153076,
"example_count": 10... |
3dafaffe-40db-4558-99a3-971c2da787f4 | easy | viridale | 2026-08-05 00:18:38.032947+00:00 | succeeded | 27b3056e624fe19c54f1e62219f602851a40376adfc95f0f66098b79d973d8bf | 12,449 | null | """Clean recurrent model with an associative matrix-product mixer.
hypothesis: v174 memorizes because dense recurrent attention can learn unrelated prompt
geometries; a small noncommutative prefix/suffix product forces every sequence cut
through one shared compositional communication law with bounded state dim... | {
"id": "3dafaffe-40db-4558-99a3-971c2da787f4",
"created_at": "2026-08-05 00:18:38.032947+00:00",
"db_md5": "6381a6372356a94c3de6c68e0bbfd98d",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "09e4ed34-bd07-437c-8b1c-fc3a2d26bbe3",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 3.8084956407546997,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.046666666865348816
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.8961856365203857,
"example_count": 10... |
3db76dc1-3c47-492e-8a58-4f4d731fd0a0 | easy | yashkant | 2026-08-15 11:40:32.616760+00:00 | succeeded | 56c58ff04a92d5bd9a48a440254e01c0fea296a76379d2c18c5d52fc96c92184 | 45,947 | null | """Round2001 dual-timescale EMA merge on frozen Round1939.
The live Round1939 architecture, initialization, sparse-SAM8 optimizer, and
learning-rate schedule are unchanged. Both matched arms store and update the
same .995 and .999 EMA banks after committed optimizer updates. At evaluation
they eagerly form the fast-on... | {
"id": "3db76dc1-3c47-492e-8a58-4f4d731fd0a0",
"created_at": "2026-08-15 11:40:32.616760+00:00",
"db_md5": "470bdef721f901c5badd76e25f4b1c2b",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "63991cf1-940d-456c-bd46-7db626fcc55f",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 4.103438377380371,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.13416667096316814
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.858675003051758,
"example_count": 300,
... |
3db9c277-f7ac-40c5-8049-216b0bc6888a | easy | MarcoF1 | 2026-08-13 18:45:14.681048+00:00 | succeeded | e19b10b114cdd819ee77a356fac67077f05e0a2356b1fdf1243f53b4b80fedb3 | 14,395 | null | """Residue-clock model for fixed-N generalization.
Representation: parsed integer values of x and N feed (a) exact residue-phase
features over a bank of small integer periods, (b) learned double-hashed value
embeddings (per-residue identity), and T feeds cyclic phase features
(T mod 2..48) plus saturating channels — p... | {
"id": "3db9c277-f7ac-40c5-8049-216b0bc6888a",
"created_at": "2026-08-13 18:45:14.681048+00:00",
"db_md5": "6559ddbd63a12a6a3049e24dc9eb8826",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "6c2e1d5b-e7f4-44c8-8657-eab80a80234c",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 4.9524999782628845,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08833333387970924
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.903562545776367,
"example_count": 100,... |
3dc4964d-338e-41ce-8957-b957de71a0ad | easy | 0Chris5R | 2026-08-08 09:31:09.584501+00:00 | succeeded | 7861e2df57b8146f35fbd404bcda093ba26f8603704c43f8a365c9a889bb54c3 | 20,999 | null | """Shared-digit Neural GPU with a compute-matched operator curriculum.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the ... | {
"id": "3dc4964d-338e-41ce-8957-b957de71a0ad",
"created_at": "2026-08-08 09:31:09.584501+00:00",
"db_md5": "730ff1095689a28f2b458f997b3f1342",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "9a8467b7-5196-4307-a50e-1057e92c3d1c",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.9820216769610735,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04125000004967054
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8950481724845987,
"example_count": 600... |
3dcf5155-cded-4177-8831-1b33549fd581 | easy | repst | 2026-08-25 21:37:25.106559+00:00 | succeeded | a1b0c3f946d9ff4b8785b05ca91235e802c21b93e185b1a1bcb1e2710ecbd643 | 98,772 | null | # R704: label-free compositional program-key geometry.
#
# R703 localized semantic retrieval failure to anisotropic program embeddings.
# R704 adds an LM-like vocabulary-geometry objective to hosted-safe R699: a fixed
# target-agnostic sample of grammar programs must retrieve itself under cosine
# score plus the existi... | {
"id": "3dcf5155-cded-4177-8831-1b33549fd581",
"created_at": "2026-08-25 21:37:25.106559+00:00",
"db_md5": "1509dc3aad9234d5013973487a7ce6c3",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "6217a270-3545-4cc8-87d3-dd6b052abfda",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 2.285592642210963,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333333333333335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2757153511047363,
"example_count": 10... |
3dd5abf1-f0a4-49c0-b536-d90ec6e2246d | easy | jayyala | 2026-08-16 14:39:37.175238+00:00 | succeeded | 9fc949ad625571426ff8c826fe5c587fe3c0c7c2282aac7b6d960fe02edb671a | 13,241 | null | """A recurrent machine over structurally encoded byte slots.
Integers use little-endian byte slots for compactness, but a byte is never an
atomic learned symbol: both encoding and decoding factor through its eight
bits. One tied transition cell is applied exactly T times. Larger integers
add slots and recurrent comp... | {
"id": "3dd5abf1-f0a4-49c0-b536-d90ec6e2246d",
"created_at": "2026-08-16 14:39:37.175238+00:00",
"db_md5": "a6921991f25718e73fbf9514784fecb8",
"submitter": "jayyala",
"github_login": "jayyala",
"run_id": "0462f671-436b-4ce9-97b7-c15bdecab248",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 35.90461206275064,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333277453978
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 35.85201644897461,
"example_count": 600,... |
3dea6c54-4b31-4830-ba78-a8e81b594957 | easy | DDanlov | 2026-08-17 11:39:18.606644+00:00 | succeeded | 26efa6e7faabf0ae4bbf3bf5133340d8410f56d6941df3cc53c9dbea1886c198 | 30,517 | null | """
Scalable Recurrent Deliberation Transformer (ScalableDEPHNet)
============================================================
Architecture:
- Pure No-Norm: Single unit RMS normalization strictly isolated to the input token embedding layer.
Zero LayerNorm / RMSNorm inside recurrent deliberation blocks.
- Pure No-Skip... | {
"id": "3dea6c54-4b31-4830-ba78-a8e81b594957",
"created_at": "2026-08-17 11:39:18.606644+00:00",
"db_md5": "3835bf9bbf48c4ee4bcda6b3bf46fa09",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "5eaf0f0a-0685-4042-8788-da73937e32b3",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.9648160075308976,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04333333276212215
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.022378921508789,
"example_count": 100,... |
3dedc024-0536-4abd-ab3b-f4e63f23ad3e | easy | DDanlov | 2026-08-23 22:51:20.818857+00:00 | failed | a9196ca1ee3874eb01d5ec25a0cfa5714be71d4da0c860c2f843cf5b25150a4e | 33,101 | null | """
Unified Hybrid Masked Deliberation Model (40.4M Parameters) - Pure Skipless, No Cross-Attention
Architecture:
- Scale: dim=1536, num_heads=24, d_ff=6144 (Exactly 40,437,888 parameters)
- Hybrid Prefix-Bidirectional + Causal Target Attention Mask
- Token Layout: [Prompt (L_p), 4 Pause Registers, Target Tokens (L_tgt... | {
"id": "3dedc024-0536-4abd-ab3b-f4e63f23ad3e",
"created_at": "2026-08-23 22:51:20.818857+00:00",
"db_md5": "111f29e0fe4fd32b85292bda48a77415",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "c4869183-aab2-4c8c-9010-fcd8d292e760",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
3deff55c-ab83-46e1-9100-77b38d6a46f0 | easy | chad-atexpedient | 2026-08-20 15:37:45.929572+00:00 | succeeded | bdffb3e42f6408fd018d6f3477c9ae2ac31774f623335163ab3335f03bd8fecf | 23,626 | null | """X115: X105 with sum-plus-extreme routed aggregation.
Signed pair features can cancel when X105 sums an anti-diagonal, especially as
unseen lengths add interactions. X115 preserves the exact X105 sum route and adds
a separately initialized low-amplitude signed-extreme summary: for each channel,
the contribution with... | {
"id": "3deff55c-ab83-46e1-9100-77b38d6a46f0",
"created_at": "2026-08-20 15:37:45.929572+00:00",
"db_md5": "1fdef83418bf478dad7d397b71a1adc2",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "1cf480d2-5ed2-4574-9712-43d8a621745f",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.3954271417417723,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009583333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.479247048296736,
"example_count": 600... |
3dff2a8f-d32d-4c4b-9c7d-5b362d6d1484 | easy | sapient-sapiens | 2026-08-23 21:34:46.309415+00:00 | failed | 50281034562e83fd1f92b97c13250caf44700b72f462b244927f46841db34f77 | 18,319 | null | """42M Q-gated SwiGLU recurrent Transformer for the E5 campaign.
The arithmetic stream uses categorical token embeddings plus marker-derived
field and left-relative embeddings. N, X, and T are never decoded as numbers.
T-field values are excluded entirely from the arithmetic and halting streams.
All recurrent readout... | {
"id": "3dff2a8f-d32d-4c4b-9c7d-5b362d6d1484",
"created_at": "2026-08-23 21:34:46.309415+00:00",
"db_md5": "2e7b9283992424a604293b4b34d67b61",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "5d0c84f8-c234-4c86-a554-b216595c53c2",
"tier": "easy",
"dataset_id": "e5",
"status": "fa... | null
|
3e002017-fcbc-478d-9eed-54d38fd14b42 | easy | bernettorlando | 2026-08-04 02:25:35.523218+00:00 | succeeded | 4fc76bf403bb27f0b4d8d7c4256c5994e9e4e7b37d3a9695c1addf32b2f75cc5 | 5,120 | null | """Beginner-friendly recurrent Transformer submission.
The same learned block is applied several times, so increasing latent depth does
not increase the parameter count. Change only the constants in the
"Experiment knobs" section when you begin experimenting.
"""
from __future__ import annotations
import torch
impo... | {
"id": "3e002017-fcbc-478d-9eed-54d38fd14b42",
"created_at": "2026-08-04 02:25:35.523218+00:00",
"db_md5": "ae947be7889c22b21b3941fdedbec6c0",
"submitter": "bernettorlando",
"github_login": "bernettorlando",
"run_id": "93b74df5-a174-49b3-bbe3-c9c23c86d566",
"tier": "easy",
"dataset_id": "e1",
"status... | {
"score": {
"mean_loss": 1.8898001909255981,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.041666666977107525
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8298262357711792,
"example_count": 10... |
3e02e55b-34ce-4ad4-b5f5-1830cae638d2 | easy | ddoan | 2026-08-14 10:23:04.028687+00:00 | succeeded | 20b5946d4ab8c8d710824da2c5903ec62e03f2bb08d2dad4fe02f33130f3edf0 | 6,928 | null | """A compact diagonal Neural GPU active-memory model."""
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,
)
MAPS = 9... | {
"id": "3e02e55b-34ce-4ad4-b5f5-1830cae638d2",
"created_at": "2026-08-14 10:23:04.028687+00:00",
"db_md5": "78b5a13f0a2a3aa48e0b6198775133e5",
"submitter": "Doug Doan",
"github_login": "ddoan",
"run_id": "01d98a15-e2be-44f3-9275-a7bbf36d0731",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.188201973198567,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666666666666
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.174349486560565,
"example_count": 600,... |
3e044b8b-8d08-4165-bb57-5d530ebf37f5 | easy | viridale | 2026-08-06 09:31:39.786571+00:00 | succeeded | ee2bd4e66b39ccd1292de74c5412c4c7e6e092e18205d3d6e8a34edb3cec9462 | 11,935 | null | """Causal zonotopic version-space recurrence with exact robust margins.
hypothesis: a compact learned convex set can preserve multiple plausible latent states
while exact worst-case vocabulary margins prevent endpoint-specific collapse.
axis: arch/loss
target: one frozen official Easy E1 gate after exact corner, c... | {
"id": "3e044b8b-8d08-4165-bb57-5d530ebf37f5",
"created_at": "2026-08-06 09:31:39.786571+00:00",
"db_md5": "87617b8b46de3c0ebb896e2394382874",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "a9874749-74c0-41fc-b64b-433912a449ed",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 3.1646814346313477,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03333333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.9459033012390137,
"example_count": 100... |
3e097a45-a20d-4902-8a01-975370af10b7 | easy | sapient-sapiens | 2026-08-23 10:01:38.689197+00:00 | succeeded | ad8068b8febb6aba170bb31606ad70c135ebfd0e3cd9be76c6d98a8dee469836 | 21,126 | null | """Approximately 10x-scale adaptive child of the D6 pair/quotient architecture.
The architecture uses the required learned additive token/group/left-relative
representation. Marker-derived field lengths additionally route categorical
digit embeddings into right-aligned lanes; this alternative place alignment is
an ex... | {
"id": "3e097a45-a20d-4902-8a01-975370af10b7",
"created_at": "2026-08-23 10:01:38.689197+00:00",
"db_md5": "ec116060c382e9fbaa087c0307927f65",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "9cdf3aa8-a993-4f63-8920-a9fe48ee7ef7",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.143978744759552,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1461742605863665,
"example_count": 600... |
3e0cb516-7a74-4636-83ef-6fe54dced42a | easy | sapient-sapiens | 2026-08-30 22:24:15.295376+00:00 | succeeded | 615907427eeffceebabc44feb43315bf90634b92684f4ed3c84b976e5536084e | 17,155 | null | """Eligible E5 direct-Horner model with two latent S writes before every A write.
One approximately 30M-parameter Transformer block is tied across all calls.
At each of four Horner positions it advances the latent workspace twice while
holding A fixed, then consumes the resulting S to commit A. N and X are
immutable,... | {
"id": "3e0cb516-7a74-4636-83ef-6fe54dced42a",
"created_at": "2026-08-30 22:24:15.295376+00:00",
"db_md5": "e79d413f744bb5f949892df0bdb886a6",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "1cc452ce-ba47-4191-b633-93587f06bdc9",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1394569992435772,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1405276140289993,
"example_count": 600,
... |
3e10f793-e298-43ea-9d92-2efcca216458 | easy | DDanlov | 2026-08-26 19:26:45.320441+00:00 | succeeded | 7a067b7c8d4bec59fbcd0a6b54ad05901d0310d69226eb759e8005d2ab36b230 | 31,772 | null | import math
import time
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from onelayerdeeper.api import Submission, OptimizerBundle, ModelSpec, OptimizerSpec
except ... | {
"id": "3e10f793-e298-43ea-9d92-2efcca216458",
"created_at": "2026-08-26 19:26:45.320441+00:00",
"db_md5": "8c57578a877790ff5191a85774bf8b45",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "42e901c4-27b6-43d8-98b8-409d4709c107",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 5.0651024878226725,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003333333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.09170713125323,
"example_count": 600,... |
3e1d05de-d2e3-4a49-8fd9-e2453ebdd555 | easy | himalalps | 2026-08-09 11:42:42.989033+00:00 | succeeded | 607f4a1e9dd3a63de83f7d26f7ca0ca4fdb7b12219338d0c5f1df1b3fc718c2d | 79,270 | null | """Algorithmically constrained latent recurrence for One Layer Deeper.
One masked cross-attention pass creates two independent representations:
``state_0`` from X only and an immutable context from N only. T is excluded
from both representations and is used solely to choose how many times the
learned recurrent MLP tr... | {
"id": "3e1d05de-d2e3-4a49-8fd9-e2453ebdd555",
"created_at": "2026-08-09 11:42:42.989033+00:00",
"db_md5": "613169a1ea865d05544a1ba2bf2f713b",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "d76afa72-ac79-4e90-aab6-6ae6e4b8f0ee",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.1779183791363024,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010833333370586236
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.161147356033325,
"example_count": 600... |
3e1d3e04-7c90-44a0-8367-8bccdf69f71c | easy | sapient-sapiens | 2026-08-07 07:41:52.106917+00:00 | succeeded | 5fbe6e45f37f22e679466d9500369cbfa03784e8e4a9e9ffde2e3f25019132b2 | 11,764 | null | """Digit-space recurrent operator for repeated modular squaring.
Nothing here converts a token into a decimal number. Digits stay as embeddings,
so what a digit is worth, and every carry and reduction, is learned. That also
sidesteps a hard numeric wall: at Hard's modulus scale `x^2` runs past the 53-bit
exact-integer... | {
"id": "3e1d3e04-7c90-44a0-8367-8bccdf69f71c",
"created_at": "2026-08-07 07:41:52.106917+00:00",
"db_md5": "b3089682bf57468cde33113ccc21f3d1",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "cc2e356e-8d53-42f7-bdfd-054aaa8dc176",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 3.849272382229641,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07499999990065892
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.7990498542785645,
"example_count": 100,... |
3e1d51c2-bac9-41ab-bca8-063bee5094f4 | easy | viridale | 2026-08-06 10:56:41.510506+00:00 | succeeded | 707de2db4eca7e6d6ed25fe5b0bf0c8962a43f18e7ff9bb8050629e0759761a2 | 9,314 | null | """Shared-attention recurrence with position-indexed learned MLP circuits.
hypothesis: arithmetic models contain causally distinct MLP circuits for different
output positions; explicit spatial parameter specialization after shared attention
should let endpoint CE discover those circuits under the short trainin... | {
"id": "3e1d51c2-bac9-41ab-bca8-063bee5094f4",
"created_at": "2026-08-06 10:56:41.510506+00:00",
"db_md5": "062fc55bd9000d8ee16b6eaa0b69cf91",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "d75a8195-8041-4676-ad4e-65d524fbfe4f",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.9610280990600586,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06500000134110451
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9254528284072876,
"example_count": 100... |
3e260048-9e63-4b39-bc6a-b6fa37d011dd | easy | TZZheng | 2026-08-07 05:29:43.780984+00:00 | succeeded | f08fd05b7074612ce3672b1a01797678731347344d792100ce91c2247aa009d3 | 12,532 | 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": "3e260048-9e63-4b39-bc6a-b6fa37d011dd",
"created_at": "2026-08-07 05:29:43.780984+00:00",
"db_md5": "12ae608771b739cfdc6ff492964f4a39",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "34eeb43e-0698-49b7-8e47-9e9a7091cee0",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 7.632482290267944,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04166666604578495
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.638937950134277,
"example_count": 100,
... |
3e261efe-5397-428d-8af8-b49aa1322185 | easy | khushidahi | 2026-08-15 22:47:19.304334+00:00 | succeeded | 57684e0a74484484860af7a4f0604e3a20ca8f03898f92966600c5688da3600e | 20,643 | null | """R75: full digit-pair Neural GPU without a direct state shortcut.
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": "3e261efe-5397-428d-8af8-b49aa1322185",
"created_at": "2026-08-15 22:47:19.304334+00:00",
"db_md5": "571f106927f476ac8f55ef07fc89298d",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "2fe6569d-14eb-46f7-930e-4d6e9b271541",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 1.3796854615211487,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.5031531751155853
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.3364574909210205,
"example_count": 60,
... |
3e35ba6e-35dc-4565-ab9e-289ca22924ab | easy | jamaliki | 2026-08-04 13:35:51.841473+00:00 | succeeded | acc6e426afb135d937ce02311a7048b1c6a265ee7a0fb8f1383d80e68a7ece7c | 14,331 | null | """One Layer Deeper submission: clocked latent loop with FSQ restoration.
Design: an encoder reads the input once into a memory. A weight-tied core is
applied m times per macro-step over [register | scratch] slots, cross-attending
to the memory (input re-injection). At each clock edge, registers are snapped
through fi... | {
"id": "3e35ba6e-35dc-4565-ab9e-289ca22924ab",
"created_at": "2026-08-04 13:35:51.841473+00:00",
"db_md5": "984a72c645e5830feef3f85df2446089",
"submitter": "Kiarash Jamali",
"github_login": "jamaliki",
"run_id": "710a7869-b6e9-416d-a322-06f89e5284d9",
"tier": "easy",
"dataset_id": "e2",
"status": "su... | {
"score": {
"mean_loss": 2.3872722387313843,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006458333693444729
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4006614685058594,
"example_count": 30... |
3e39d951-857d-470a-90c7-de0ecc2cf0fa | easy | poissonali137 | 2026-08-05 02:21:57.180925+00:00 | succeeded | 3d1cbc6e6b35a32b63ccefb70faf59827a85d6c927a150707934f24d2ebb373b | 10,247 | null | """JordanAttention looped transformer + Muon — fully bounded loop, no norms.
Variant of jordan_loop: the loop block contains NO normalization. Stability is
structural: QK cone projection bounds scores in (0,1], sum-to-1 attention makes
the attention output a convex combination of values, and the residual state is
soft... | {
"id": "3e39d951-857d-470a-90c7-de0ecc2cf0fa",
"created_at": "2026-08-05 02:21:57.180925+00:00",
"db_md5": "f39ee2adadaad32d301c5cabf2110529",
"submitter": "Ali Abdul Rahim",
"github_login": "poissonali137",
"run_id": "4e5b02ac-d5f7-4033-a7c1-64c5008ede66",
"tier": "easy",
"dataset_id": "e1",
"status... | {
"score": {
"mean_loss": 16.250516891479492,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.10500000044703484
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 16.59031105041504,
"example_count": 100,... |
3e419968-7d8c-4aaa-b489-2367471efcc2 | easy | erdavis0 | 2026-08-25 02:37:16.234001+00:00 | succeeded | 150d173c75338db2a7113c981ff2f114e97e4da3a7ca5424065633b1be6b8905 | 26,183 | null | """Reversible augmented-state transducer over aligned learned digit tapes.
Public delimiter syntax only routes opaque token categories into distinct,
right-aligned N, X, and T tapes. Learned local cells compile separate
interaction, carry, and modulus-conditioned workspaces. N is isolated from
interaction and carry ... | {
"id": "3e419968-7d8c-4aaa-b489-2367471efcc2",
"created_at": "2026-08-25 02:37:16.234001+00:00",
"db_md5": "fad6ce63bd264638f733d308596ed7c1",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "57e681e9-b21d-4f80-97c0-ffb44f5c18d6",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 5.1357808113098145,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.18400901556015015
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.028593063354492,
"example_count": 60,
... |
3e46ab44-e372-4b2b-bc32-4e6aec527fc0 | easy | int3 | 2026-08-18 10:31:12.100705+00:00 | succeeded | d1d2d391029fd381e98ff5eb77e3f892d2ee16b7677ac68126984b4ee6e8ba61 | 6,779 | null | """Tied-weight recurrent transformer: iterate a learned squaring step T times.
Family A (depth extrapolation by construction): an encoder parses the prompt,
one weight-tied core block is applied T times (T read from the prompt's digit
tokens, per-example halting masks keep everything batched), and a decoder emits
the ... | {
"id": "3e46ab44-e372-4b2b-bc32-4e6aec527fc0",
"created_at": "2026-08-18 10:31:12.100705+00:00",
"db_md5": "9c964370aad310ddd376776d65c5c795",
"submitter": "Jez Ng",
"github_login": "int3",
"run_id": "351f9b04-b2c0-4f71-bcbc-43fa699fa4e4",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 6.390940189361572,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04833333194255829
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.373433589935303,
"example_count": 100,
... |
3e49969d-4582-47cf-892f-79e86a94d12e | easy | khushidahi | 2026-08-17 16:00:48.523211+00:00 | succeeded | 948cc6bf4eee71b852f412362116f18a42655ec0b3b93cb37bc51ff4ad32cde6 | 23,789 | null | """R208: fixed-80-update batch-256, depth-7 factorial control.
This is the first rung of a matched adaptive-recurrence ladder. It isolates
the substrate before input reinjection, dense trajectory losses, learned
halting, or a specialized deep optimizer are added.
The model keeps N and X categorical. A randomly init... | {
"id": "3e49969d-4582-47cf-892f-79e86a94d12e",
"created_at": "2026-08-17 16:00:48.523211+00:00",
"db_md5": "7833c60296d2b546f39f63f485128f69",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "bf3fd3a0-1091-4dfa-812e-339a03a01dc4",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 2.423026442527771,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006756756920367479
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5011093616485596,
"example_count": 60,... |
3e4c7723-30cb-4aa3-9c42-0a60901da9ab | easy | Dandandan | 2026-08-07 03:13:45.193881+00:00 | succeeded | 389b877b2fb82b9f4c7d0348b7c125773776ef1843c650a66cf90339f110fd85 | 99,108 | null | """Generalized exact recursive learner for One Layer Deeper.
The learned transition is a population of ordinary modular neural units with
several generic arithmetic activations. A single learned evidence selector
chooses one transition, which is then reused for every input-requested outer
step. The population covers ... | {
"id": "3e4c7723-30cb-4aa3-9c42-0a60901da9ab",
"created_at": "2026-08-07 03:13:45.193881+00:00",
"db_md5": "ce1e78c12ba9dfaa80226e9e4347495d",
"submitter": "Daniël Heres",
"github_login": "Dandandan",
"run_id": "2a9eb9b0-cc80-4dd6-b508-0952b00ee49f",
"tier": "easy",
"dataset_id": "e1",
"status": "suc... | {
"score": {
"mean_loss": 164.59204864501953,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.019999999552965164
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 159.17910766601562,
"example_count": 10... |
3e51e351-e183-40af-96e3-546a47a6a021 | easy | shirvani-jr | 2026-08-30 23:47:54.922460+00:00 | succeeded | 5e04560fff6d76c6350ed134dce34bbeb5ddf03992d210d9cdc6dda6ff09212c | 17,544 | 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,
)
PERIOD_MINIMUM = 11
PERIOD_MAXIMUM = 100
E... | {
"id": "3e51e351-e183-40af-96e3-546a47a6a021",
"created_at": "2026-08-30 23:47:54.922460+00:00",
"db_md5": "cfec548a5587d3a2942b9c10220452bc",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "27fc0a6f-7204-4e69-9d66-daf4c541daf7",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.811719059944153,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.19749999418854713
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5472452640533447,
"example_count": 800,... |
3e552c96-9fab-44db-9091-931228b5e470 | easy | sapient-sapiens | 2026-08-30 22:26:26.567406+00:00 | succeeded | 5aef17aa727b8edd71a178dc1898fc1e14893f68307565827ea855db9db1c470 | 17,155 | null | """Eligible E5 direct-Horner model with two latent S writes before every A write.
One approximately 30M-parameter Transformer block is tied across all calls.
At each of four Horner positions it advances the latent workspace twice while
holding A fixed, then consumes the resulting S to commit A. N and X are
immutable,... | {
"id": "3e552c96-9fab-44db-9091-931228b5e470",
"created_at": "2026-08-30 22:26:26.567406+00:00",
"db_md5": "bc6be38c6d5ae5e07fe30e2a63281aff",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "f17b21eb-b5d5-49d7-906c-df75fc596191",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.13858469543181,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.138360361347284,
"example_count": 600,
... |
3e5ac250-7109-45c5-ad6c-5d25657cfc7e | easy | karanganesan | 2026-08-05 10:03:22.100404+00:00 | succeeded | 7af54e899459a26200a71b0c876d5e3d942c2eb7e1c418be459891c847072686 | 22,081 | 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": "3e5ac250-7109-45c5-ad6c-5d25657cfc7e",
"created_at": "2026-08-05 10:03:22.100404+00:00",
"db_md5": "255e2f927dd274a2579557bd8a91931d",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "11603fc7-c4af-4f13-97a3-66cb727700e7",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 8.443020343780518,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.021666666492819786
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.21383285522461,
"example_count": 100,
... |
3e5e40d6-b1a8-4fdb-bb78-bb58cdda7f53 | easy | khushidahi | 2026-08-01 23:13:15.149903+00:00 | succeeded | a806ee11d8d698da19060605abfb048402d5f951e4ec0aa4de6b4e36e6d5a2b2 | 7,233 | null | """Compact recurrent local/global model for One Layer Deeper.
R0 optimizer ablation:
- train for 400 optimizer steps;
- inject immutable prompt context at every shared-core update;
- combine global attention with a cheap learned local digit mixer;
- use stable outer normalization;
- train/evaluate at four recurrent up... | {
"id": "3e5e40d6-b1a8-4fdb-bb78-bb58cdda7f53",
"created_at": "2026-08-01 23:13:15.149903+00:00",
"db_md5": "429b8f6f013b1861ab6a986e07784e6a",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "57c8e246-379f-4f42-b512-f3814ffae092",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.176177144050598,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666767559946
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1823132038116455,
"example_count": 600... |
3e67aae7-4938-4663-a120-cde4810aaaf1 | easy | JackCai1206 | 2026-08-13 04:18:10.471405+00:00 | succeeded | 888fa58db3e0499b619d8effdb8dff786f9a279d130ee51ed002a37d5d121f55 | 7,689 | null | """root — transformer++ iterated encoder, the tree's baseline node.
Principal directives baked in:
- STANDARD token embedding in, STANDARD linear unembedding over the vocab
at every position (loss masked by the evaluator's valid_mask). No numeric/
abacus/base-100 encoders here — those are CHILD nodes measure... | {
"id": "3e67aae7-4938-4663-a120-cde4810aaaf1",
"created_at": "2026-08-13 04:18:10.471405+00:00",
"db_md5": "89e2b0aace924998192d9295e1e13baa",
"submitter": "Ziyang Cai",
"github_login": "JackCai1206",
"run_id": "48004afc-b202-46f5-8842-e68a35fcc798",
"tier": "easy",
"dataset_id": "e2",
"status": "suc... | {
"score": {
"mean_loss": 2.487507700920105,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666772216558
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.476604461669922,
"example_count": 300,... |
3e6c1746-e690-4436-acb4-b28468e0223d | easy | shirvani-jr | 2026-08-31 01:39:39.979672+00:00 | succeeded | f36b045c50800cbdceef864d677daf420a793727133eaedab87b1505981fd664 | 21,222 | 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,
)
PERIOD_MINIMUM = 3
PERIOD_MAXIMUM = 100
EX... | {
"id": "3e6c1746-e690-4436-acb4-b28468e0223d",
"created_at": "2026-08-31 01:39:39.979672+00:00",
"db_md5": "52d288f37eb9ac99c594e70a5f2357e8",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "224ac52e-10d5-43b2-bf42-6d566b36957e",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 0.009462370187975466,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.996666669845581
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0004469298291951418,
"example_count": ... |
3e712dd5-a01b-4e6e-8144-cf1e983696bb | easy | ArpBansal | 2026-08-14 19:25:21.187643+00:00 | succeeded | 99aa7f0093e32b405f6bd7d9d51ed9f1a4fb29d0d866a409657236a6211ec740 | 7,365 | null | """Light recurrent encoder-decoder Transformer for the public Easy datasets.
This submission adapts the compact architecture from "Learning Modular
Exponentiation with Transformers" to the evaluator's separate-output contract.
The encoder reads the complete prompt. The decoder starts from learned output
queries and pr... | {
"id": "3e712dd5-a01b-4e6e-8144-cf1e983696bb",
"created_at": "2026-08-14 19:25:21.187643+00:00",
"db_md5": "dbd5807565fe6f0bb73efa5c6faf9438",
"submitter": "Arpit Bansal",
"github_login": "ArpBansal",
"run_id": "107d0aa1-aad8-468c-a199-1a3a591a89a4",
"tier": "easy",
"dataset_id": "e1",
"status": "suc... | {
"score": {
"mean_loss": 3.0567097663879395,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05999999865889549
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.6037909984588623,
"example_count": 100... |
3e726c0e-3ac5-4c9c-982d-038629545556 | easy | briheet | 2026-08-07 04:54:56.322681+00:00 | succeeded | e6142816a43f9a270cea658d494c255a72612f38764531661673db3c6f240832 | 8,382 | null | """Small Universal Transformer with shared refinement passes."""
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 = 32
NUM_HEA... | {
"id": "3e726c0e-3ac5-4c9c-982d-038629545556",
"created_at": "2026-08-07 04:54:56.322681+00:00",
"db_md5": "e13f179948550892510215c59ad8ce3a",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "40812104-6a01-427c-b630-1102ca5fc652",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 1.9481772184371948,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9211578369140625,
"example_count": 10... |
3e76d973-e93e-4c45-8e66-37f55b8e92f4 | easy | sapient-sapiens | 2026-08-06 21:18:34.596511+00:00 | succeeded | d5c084251c41a1a825f190ef3b27e86b7e5d5428b4ed9b19f9c3b2e663332d5f | 12,431 | null | """Recurrent single-operator model for repeated squaring.
The network learns one transition operator and applies it once per requested
time step, decoding an intermediate decimal state between applications. Depth
therefore comes from serial reuse of a single learned map rather than from a
closed-form expression, and n... | {
"id": "3e76d973-e93e-4c45-8e66-37f55b8e92f4",
"created_at": "2026-08-06 21:18:34.596511+00:00",
"db_md5": "9e84e510e2afd06b8b1027cc592e6d66",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "1b7b84f0-b3fb-48c0-be33-a2eb5cd2780c",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 1.2324436902999878,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9116666615009308
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.234283208847046,
"example_count": 100,
... |
3e77f0d0-d6f2-4ab1-9283-49f4c74636f8 | easy | karanganesan | 2026-08-08 05:05:01.941608+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": "3e77f0d0-d6f2-4ab1-9283-49f4c74636f8",
"created_at": "2026-08-08 05:05:01.941608+00:00",
"db_md5": "3e8439ba8ab561733e7dc883fab4ecb2",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "d8fb26f7-3a97-443d-961f-73ca1a1f8d6f",
"tier": "easy",
"dataset_id": "e4",
"status": ... | {
"score": {
"mean_loss": 7.06977391242981,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0024999999441206455
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.99122428894043,
"example_count": 1200,... |
3e81593f-f925-47da-92a4-08b455ceb805 | easy | richardcepka | 2026-08-13 22:14:42.586084+00:00 | succeeded | 02a8d4cede7b39071ad8381c1ac2507faa7885ebab407746159eff86ec140af5 | 43,380 | null | """Recurrent digit-register model for One Layer Deeper.
A looped transformer whose recurrent state is a decimal digit register that is
re-quantised through a ten-entry codebook on every iteration. Nothing inside
the loop depends on the iteration index and no parameter is indexed by an
absolute slot position, so the e... | {
"id": "3e81593f-f925-47da-92a4-08b455ceb805",
"created_at": "2026-08-13 22:14:42.586084+00:00",
"db_md5": "200b6f7ff4ed677ea066f2a34aaaa48b",
"submitter": "Richard Cepka",
"github_login": "richardcepka",
"run_id": "87a8eb01-c384-4bf1-93a4-8dbb007869ed",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.2464419124094217,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666883975268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2439069747924805,
"example_count": 60... |
3e8174e4-e762-4225-a3dd-a98ec4728130 | easy | newjordan | 2026-08-09 16:09:49.805664+00:00 | succeeded | 09e317114058d59619022cb0e4953f76718231ddd677f3859577238a9096ccf5 | 93,405 | null | """Helix-Crawler — a recursive-weight-shared "crawler" ALU for modular squaring.
The task is r_{i+1} = r_i^2 mod n, chained T times: a *recurrence* over a single
operator. This build embodies the crawler / Frugendorff lineage (spark:~/sota_crawler)
mapped onto that structure. n never enters as a table row (unseen modu... | {
"id": "3e8174e4-e762-4225-a3dd-a98ec4728130",
"created_at": "2026-08-09 16:09:49.805664+00:00",
"db_md5": "02e2eb0c60a05d858148e41a3d5e435e",
"submitter": "Frosty40",
"github_login": "newjordan",
"run_id": "6d0326e3-ea35-4ce7-9adf-5357ba8e63f2",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 0.0008539318489857258,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0008534474233122285,
"example_count": 800,
... |
3e874dc1-e762-4204-a3b9-f743195f67d0 | easy | achyutreddy24 | 2026-08-04 23:16:11.143372+00:00 | failed | 3243bd5dea3b518bab939bb379ed4d3a41767a5a5d833c98d48fa3e72e403dd2 | 11,637 | null | """A looped transformer with learned, input-conditioned depth.
WHAT THIS IS
============
This submission contains no task-specific arithmetic: no modular structure, no
hand-set initialization, no token-map constants. It is a single generic
transformer block applied repeatedly to a latent state, with a learned halting... | {
"id": "3e874dc1-e762-4204-a3b9-f743195f67d0",
"created_at": "2026-08-04 23:16:11.143372+00:00",
"db_md5": "0a62462d183ea04ef33ccdd2a1f40260",
"submitter": "Achyut Reddy",
"github_login": "achyutreddy24",
"run_id": "37e72dc9-2389-4734-baad-582d6d4012d9",
"tier": "easy",
"dataset_id": "e1",
"status": ... | null
|
3e87f771-140e-45ba-bdc1-63fb7b1ff3e0 | easy | Mantissagithub | 2026-08-19 14:38:59.389999+00:00 | succeeded | 43a9807791d82395cbc0e40dee39e23a721b003bbf5ac594a8978be54b0c4adf | 32,123 | null | """attempt_035 + a contentless T field (depth enters only as iteration count).
ADDENDUM (attempt_042_gated): attempt_040_residual with the cross-iteration
carry scaled by a single learned scalar gate, sigmoid(carry_logit), init 0.5.
This is one scalar parameter (+1 element against the model-state ceiling) and
one mult... | {
"id": "3e87f771-140e-45ba-bdc1-63fb7b1ff3e0",
"created_at": "2026-08-19 14:38:59.389999+00:00",
"db_md5": "a47353e05bfbde0454be06102e161cd0",
"submitter": "Pradheep P",
"github_login": "Mantissagithub",
"run_id": "e5814361-e75c-4e33-afa4-b9a8922d8164",
"tier": "easy",
"dataset_id": "e2",
"status": "... | {
"score": {
"mean_loss": 5.026310205459595,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.3579166845884174
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.318873405456543,
"example_count": 300,
... |
3e8cbd03-1212-45ef-8a03-ea9aaf7c9554 | easy | DDanlov | 2026-08-20 12:27:01.908487+00:00 | succeeded | 8fa728e3b1030575467ed098ee9787f9e95e93eadcce5aeb9cdde42436c5f19b | 20,959 | null | from __future__ import annotations
import math
import time
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
except Imp... | {
"id": "3e8cbd03-1212-45ef-8a03-ea9aaf7c9554",
"created_at": "2026-08-20 12:27:01.908487+00:00",
"db_md5": "6ff41db3d37f04d99fe1f4b0f2c69538",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "286b1f98-9188-4bed-97c0-65af25cec195",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.5346051619618386,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5407175889464235,
"example_count": 60... |
3e933083-bf0b-4bb9-bc13-3f757c93b4d3 | easy | sapient-sapiens | 2026-08-26 11:18:53.990963+00:00 | succeeded | 58a435efc3068dacbc9a20c91863f159d56eebe9fa45fc4a0fe6ccc7d1856ab3 | 20,313 | null | """Standalone nested inner/outer recurrence candidate.
Representation is categorical only: ``E_token + E_group + E_leftrel``. No
field is decoded into a number, no field value sets depth, halting, or any
mixing weight, and the only supervised target is the final answer.
Structure
---------
One prompt encoder produce... | {
"id": "3e933083-bf0b-4bb9-bc13-3f757c93b4d3",
"created_at": "2026-08-26 11:18:53.990963+00:00",
"db_md5": "3e32d6c9a71856cebe2ad7affee8cf3a",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "35955572-713d-411c-9edf-fd5b294ec1f4",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 4.95712271899729,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010833333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.855609611545443,
"example_count": 600,
... |
3e96a329-159b-4157-9744-1353077ddf6a | easy | Amr-Hegazy1 | 2026-08-03 22:07:13.778568+00:00 | succeeded | 23d3eac6e8359ca0592a721183b6c2cbbc90fd81f51b7076ae656b35a8600120 | 17,919 | null | """RecurrentGPT: prelude / gated shared recurrent core / coda.
Port of the recurrentGPT architecture to the One Layer Deeper contract.
The model is a depth-recurrent transformer:
* a fixed *prelude* stack embeds the prompt and produces ``head_output``;
* a *shared core* stack is applied ``r`` times. Every step ... | {
"id": "3e96a329-159b-4157-9744-1353077ddf6a",
"created_at": "2026-08-03 22:07:13.778568+00:00",
"db_md5": "225a7eaced136a47a69a5ea3a332cbca",
"submitter": "Amr Hegazy",
"github_login": "Amr-Hegazy1",
"run_id": "e03769ad-babf-4c38-b805-2993633470f6",
"tier": "easy",
"dataset_id": "e2",
"status": "suc... | {
"score": {
"mean_loss": 8.86708688735962,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.020833334419876337
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.76190185546875,
"example_count": 300,
... |
3ea12708-b3e0-4d40-bce2-c1c55d4f2348 | easy | Ranoobaba | 2026-08-16 06:10:03.712992+00:00 | failed | eeb85c7230cd02cb5a038bf442ae30be91b6a93ad8a0677caa2617879bb1ad70 | 3,798 | 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": "3ea12708-b3e0-4d40-bce2-c1c55d4f2348",
"created_at": "2026-08-16 06:10:03.712992+00:00",
"db_md5": "a8a86572d9aa789279d413bdc47401e1",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "7971e02b-a0d6-4fe4-a288-b2fe326ef18d",
"tier": "easy",
"dataset_id": "e1",
"status": "... | null
|
3ea63e99-95a7-4e1a-979a-9ad108710f82 | easy | erdavis0 | 2026-08-24 12:35:10.228308+00:00 | succeeded | 783fcd106ba99d648ca87a302210ea3f53034ff158d9adc82685d3eadb750a03 | 28,687 | null | """Reversible augmented-state transducer over aligned learned digit tapes.
Public delimiter syntax only routes opaque token categories into distinct,
right-aligned N, X, and T tapes. Learned local cells compile separate
interaction, carry, and modulus-conditioned workspaces. N is isolated from
interaction and carry ... | {
"id": "3ea63e99-95a7-4e1a-979a-9ad108710f82",
"created_at": "2026-08-24 12:35:10.228308+00:00",
"db_md5": "7b86bec6cb9246367a8281ab71caa6fd",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "1d562a7a-95b0-46c5-b350-68871055c1fc",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",... | {
"score": {
"mean_loss": 3.7256204896007787,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.710523873701224,
"example_count": 600... |
3ea941e1-d2cd-4074-9da1-357329abcb2f | easy | yashkant | 2026-08-13 01:12:13.297066+00:00 | succeeded | 19ee10a5002629ec699f190973def02290c904dec7606b44a170117e56d7d427 | 31,758 | null | """Round1701 Dykstra width16/hidden32 architecture pair on the D416 MemGLU carrier.
The fixed Round1141 carrier first produces its ordinary pooled representation.
A compact primal state and two Dykstra correction memories are initialized
from that immutable representation. Every parsed public-T cycle applies two
cond... | {
"id": "3ea941e1-d2cd-4074-9da1-357329abcb2f",
"created_at": "2026-08-13 01:12:13.297066+00:00",
"db_md5": "dc5f85ec308eaf0effdca1e21b1f0873",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "abdd6c94-9f97-444d-9445-0044048e21c9",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 3.9028817415237427,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.27145834639668465
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.059340000152588,
"example_count": 300,... |
3ead3cca-5ce0-42d4-9719-48da52b90f87 | easy | yashkant | 2026-08-14 16:06:30.502021+00:00 | succeeded | 64fffad84aa07dd47f8d07a02db18566bda81c1065b7bf297ff06b761199e20f | 34,866 | null | """Round1702 three-projections-per-T Dykstra recursion on the D416 carrier.
Each parsed public-T cycle applies an A-B-A sequence through one shared learned
smooth projection, reinjecting the immutable anchor on every call and updating
the corresponding Dykstra correction memory after each projection. All active
cycle... | {
"id": "3ead3cca-5ce0-42d4-9719-48da52b90f87",
"created_at": "2026-08-14 16:06:30.502021+00:00",
"db_md5": "91df24cf0fd3e4cf4ae643bb6e451b95",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "0acd89d8-8d60-4066-b055-ddf2f774d5fb",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 4.77331018447876,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.1681250035762787
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.343387126922607,
"example_count": 300,
... |
3eb65314-2912-4d64-b7d9-795664f0ef1d | easy | adambrzosko | 2026-08-30 15:45:26.133382+00:00 | succeeded | 3139f3baafa363bafeab6cdfa8f313ca1d7626cfa654cc0eb26a83088d2c32cb | 13,475 | null | """One Layer Deeper — digit-slot recurrent machine ("dm").
The prompt is [N] digits(N) [X] digits(x) [T] digits(T) and the label is the
decimal expansion of x^(2^T) mod N, tail-aligned: the last prompt position
carries the units digit, the one before it the tens digit, and so on.
Architecture. The model parses the ... | {
"id": "3eb65314-2912-4d64-b7d9-795664f0ef1d",
"created_at": "2026-08-30 15:45:26.133382+00:00",
"db_md5": "3f966c8760e28a29255d92bfebe2b403",
"submitter": "Adam Brzosko",
"github_login": "adambrzosko",
"run_id": "1568d222-9c0a-4f9d-b002-d4460d402109",
"tier": "easy",
"dataset_id": "e3",
"status": "s... | {
"score": {
"mean_loss": 3.0754518634592527,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003125000004656613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.0493525810787343,
"example_count": 80... |
3ec04e1a-888f-4e00-9be6-4633341aca30 | easy | khushidahi | 2026-08-02 22:40:58.536427+00:00 | succeeded | 8dec8a67f976080184aadd9931e3e881847477356df1ff4f7800cae8b023a31d | 8,515 | null | """Prelude -> recalled shared core -> coda model for One Layer Deeper.
R2 architecture reset:
- use a real one-block prelude and one-block coda;
- apply one shared recurrent Transformer core four times;
- inject immutable prelude context at every recurrent update;
- use sandwich RMS normalization for stable repeated a... | {
"id": "3ec04e1a-888f-4e00-9be6-4633341aca30",
"created_at": "2026-08-02 22:40:58.536427+00:00",
"db_md5": "0fcc526cd2de4c1f015033de6ab5067a",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "cea292e9-a157-408f-b9ac-6bc06e1debc1",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.2377418279647827,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014375000260770321
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2381319999694824,
"example_count": 30... |
3ec4c792-691e-43e3-a12c-888242d6379b | easy | CodeReclaimers | 2026-08-19 17:16:20.516224+00:00 | succeeded | ecf4a13e5aab1f4b9b4523a3a195b00ff9562444f63f454f4e016cdd6d616d13 | 34,943 | null | """v39_keyed: keyed-modulus CRT learner — Run A of T1_CERT_PLAN_20260819.
HYPOTHESIS A (alias modulus): h1's displayed N is a stable per-identity
alias phi(M) of a hidden modulus M <= 2^20 from enumerated factor pairs
(<=10-bit primes), while x and y are displayed plainly and the step closes
per prime channel (squarin... | {
"id": "3ec4c792-691e-43e3-a12c-888242d6379b",
"created_at": "2026-08-19 17:16:20.516224+00:00",
"db_md5": "0a0523d23903562a1b9c3c00e1a7b60a",
"submitter": "CodeReclaimers",
"github_login": "CodeReclaimers",
"run_id": "349d9917-4ce8-4874-991c-727f8a5f3964",
"tier": "easy",
"dataset_id": "e2",
"status... | {
"score": {
"mean_loss": 1.13506415967465,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.5
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2701283193493,
"example_count": 300,
"exact_ac... |
3ed12a11-5951-431b-ab06-cfd0ded95d5c | easy | viridale | 2026-08-02 13:23:24.233293+00:00 | succeeded | a8037d884490992ce127bb10ac49414bb2ae819b2f7f347f642b56e9eb9f4c6f | 43,467 | null | """RNG-stable streamed four-monomial closure plus open table.
hypothesis: v100's 4.05M added random draws shift the otherwise unchanged
open table to an initialization that misses its registered step-50 exactness
line. Draw the v099 head first, sample the new tail without consuming the
downstream global R... | {
"id": "3ed12a11-5951-431b-ab06-cfd0ded95d5c",
"created_at": "2026-08-02 13:23:24.233293+00:00",
"db_md5": "20deaed0b16c36a67233d348a3ebbd75",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "dee45ff6-5fdc-47a1-8afc-9a0eda092cea",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 0.003014546214626436,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.00301454615599774,
"example_count": 800,
"... |
3ed4b741-cd79-42d0-9c46-e149ab6ba9b0 | easy | erdavis0 | 2026-08-27 02:41:57.887716+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": "3ed4b741-cd79-42d0-9c46-e149ab6ba9b0",
"created_at": "2026-08-27 02:41:57.887716+00:00",
"db_md5": "86ddb36d905f79e81813767c4d60fc8f",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "4282e6ab-9d01-4455-ad9a-f39b10922f37",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",... | {
"score": {
"mean_loss": 3.514651298522949,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01416666666045785
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.522459030151367,
"example_count": 300,
... |
3ee451a7-8f68-4886-a0a6-946ce3e488f3 | easy | jordanrubin | 2026-08-06 07:43:09.201072+00:00 | succeeded | c83dc25c809f8cc2affd3d72265804370c3da6459badad9917993017f7f1e9ff | 16,244 | null | """Gauge-free pair-relational categorical looped Transformer.
This is a learned transition model, not an arithmetic program. The fixed
parts expose only the public prompt grammar and six representation priors:
right-origin decimal slots, a shared ordered digit geometry, unordered pair
tokens, immutable N recall, a so... | {
"id": "3ee451a7-8f68-4886-a0a6-946ce3e488f3",
"created_at": "2026-08-06 07:43:09.201072+00:00",
"db_md5": "b3d05b68de6786a4d3e2a080568872ba",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "9dd75fc0-c1c2-4b18-84d6-581e68ba584f",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 1.8865436911582947,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8524534702301025,
"example_count": 100,
"exa... |
3ee5188a-d3b0-4367-965f-76f63e654e3a | easy | benjaminW2025 | 2026-08-17 06:43:06.149177+00:00 | succeeded | 1cd04f8681e087726b62652eb39873c25d294ae6dca08d8766dd9856187203c2 | 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 = 256
NUM_HEADS = 4
... | {
"id": "3ee5188a-d3b0-4367-965f-76f63e654e3a",
"created_at": "2026-08-17 06:43:06.149177+00:00",
"db_md5": "9aa703cefc29b7753f77e8ead1da82fa",
"submitter": "benjawesome",
"github_login": "benjaminW2025",
"run_id": "bf9c2047-c4e3-4288-9f61-6ce2b7577744",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 3.776806950569153,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03833333496004343
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.122014999389648,
"example_count": 100,
... |
3eeded7f-342c-42a9-8732-a8c875ec470f | easy | Ranoobaba | 2026-08-19 06:59:08.779003+00:00 | succeeded | efbaea4ffbed20e5a809b52c5b3fdf58becb8bd4e553babb1a7569146f4264a9 | 5,760 | 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": "3eeded7f-342c-42a9-8732-a8c875ec470f",
"created_at": "2026-08-19 06:59:08.779003+00:00",
"db_md5": "5b03ef911c7c18c1737785ce8e09bfc8",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "c7c9ba8b-faed-486b-9a0e-e30dd731ba4d",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 6.2156812746328605,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03999999942878882
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.309569358825684,
"example_count": 100,... |
3ef5f395-8179-499a-b2c0-caf211e82285 | easy | erdavis0 | 2026-08-26 03:56:30.910415+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": "3ef5f395-8179-499a-b2c0-caf211e82285",
"created_at": "2026-08-26 03:56:30.910415+00:00",
"db_md5": "14ff631f83062f36483f24b1bd94c4c9",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "fa09fcf2-5af5-4e40-ad0a-285105c00e1c",
"tier": "easy",
"dataset_id": "e4",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.190647333184657,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004479166666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1969874759307237,
"example_count": 120... |
3ef66a1d-b50c-4b0c-8a62-0408fc26ca52 | easy | erdavis0 | 2026-08-24 16:38:04.889889+00:00 | succeeded | 8d18858c2c6e8c49faaa8fec6a8f5652d4aeade24c70fda3e7abf76d8b4bb69e | 26,473 | null | """Target-clamped primal-dual energy tape.
Public delimiter syntax only routes opaque token identities into separate,
right-aligned N, X, and T tapes. Learned local cells form distinct
interaction, carry, and modulus-conditioned workspaces. Two learned local
energy views then negotiate over the actual categorical ou... | {
"id": "3ef66a1d-b50c-4b0c-8a62-0408fc26ca52",
"created_at": "2026-08-24 16:38:04.889889+00:00",
"db_md5": "94c3763175db222103a68fb5cc760517",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "45c03926-becd-4a55-842e-2f48deb3c2d9",
"tier": "easy",
"dataset_id": "e10",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.242697238922119,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.10377778299152851
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.036086082458496,
"example_count": 125,
... |
3efd9037-1257-400f-9292-1c7de8142d4e | easy | yashkant | 2026-08-16 00:39:51.304817+00:00 | succeeded | b19290a9507fde506b98d9d490bb250711935570136da61e1c79e6f4888cd257 | 38,959 | null | """Round1818 parent-preserving temporal Anderson-Dykstra mixing.
The complete Round1792 AdamW5/EMA.995 W8/D2 carrier remains byte-identical at
a matched seed. Initialization of the added temporal mixer is isolated from
the inherited CPU RNG stream, and one exact-zero scalar gate keeps the scored
path equal to the par... | {
"id": "3efd9037-1257-400f-9292-1c7de8142d4e",
"created_at": "2026-08-16 00:39:51.304817+00:00",
"db_md5": "b2509923b46a48fc2d1319299bcbde3c",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "06c77676-0f12-47d6-a3df-b2c162e4c340",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.152417158581633,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666679084302
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.155923772820443,
"example_count": 600,... |
3f0aeefc-6219-4df1-9d1c-7300414f2ddc | easy | DDanlov | 2026-08-17 11:34:44.929053+00:00 | succeeded | 96078a10c183709b67b26b99de9688aecf696889e2d66d12f9e4aeed1044e867 | 29,521 | null | """
Scalable Recurrent Deliberation Transformer (ScalableDEPHNet)
============================================================
Architecture:
- Pure No-Norm: Single unit RMS normalization strictly isolated to the input token embedding layer.
Zero LayerNorm / RMSNorm inside recurrent deliberation blocks.
- Pure No-Skip... | {
"id": "3f0aeefc-6219-4df1-9d1c-7300414f2ddc",
"created_at": "2026-08-17 11:34:44.929053+00:00",
"db_md5": "23dcfac41bbf7793fa60a82dc9216c04",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "59cb46cb-90da-4f70-9c5c-1d6bf616bf86",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.184458113073411,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333307504654
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.202906370162964,
"example_count": 100,
... |
3f245484-a228-4d6b-a2ca-fe3ebbd7927e | easy | DDanlov | 2026-08-09 00:07:27.280890+00:00 | failed | 309a0a363f3fd87cf9e10f1ca018ea40f3c87102f468e5a7f447ff9ac09cc7fc | 15,981 | 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": "3f245484-a228-4d6b-a2ca-fe3ebbd7927e",
"created_at": "2026-08-09 00:07:27.280890+00:00",
"db_md5": "0315d4b5477360d9ae774ecbb9e4a9d4",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "d872944c-7e8b-4c17-814d-a4fe72686b59",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
3f24cdc3-329e-4ffe-9167-ebabe00b4fae | easy | shirvani-jr | 2026-08-22 13:16:48.216179+00:00 | succeeded | 1398bd08847c5eaab4959d387d99f9b6d5d587478909c52a6d433aa99a043c89 | 23,140 | null | """Looped exact-arithmetic digit machine.
The recurrent state is the decimal digit string of the current residue. One
loop application performs one modular squaring, so depth extrapolates in T by
construction. Every functional primitive — digit-by-digit product, full
adder, full subtractor, digit comparator, and com... | {
"id": "3f24cdc3-329e-4ffe-9167-ebabe00b4fae",
"created_at": "2026-08-22 13:16:48.216179+00:00",
"db_md5": "a1310d423bb5360fa56d444443de24fa",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "d6fc8d0b-080b-410b-8a6c-5d8df614d577",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded"... | {
"score": {
"mean_loss": 0.022861258126795292,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.022804344072937965,
"example_count": 300,
... |
3f2597be-4a8e-438d-957b-d938a362cf53 | easy | TheYkk | 2026-08-04 13:49:55.666225+00:00 | succeeded | 0565fd58f08c5b5028987e91ed771ad1cdc9ed337015a60a024f2a3f858a534c | 24,237 | null | """Weight-tied positional loop with the cross-position product as a primitive.
The change from the previous submission
---------------------------------------
The previous submission relied on attention followed by a gated MLP to express
a cross-position product: attention gathers position ``j`` into position ``i``,
a... | {
"id": "3f2597be-4a8e-438d-957b-d938a362cf53",
"created_at": "2026-08-04 13:49:55.666225+00:00",
"db_md5": "b6645eb3367355bb31aa5ee1a5ca0da3",
"submitter": "Kaan Karakaya",
"github_login": "TheYkk",
"run_id": "b6c39219-dd1b-4905-8bbb-8c04bc0651f5",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 13.522187232971191,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03166666720062494
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 12.971429824829102,
"example_count": 100... |
3f2fc392-5eb1-42e4-8dcc-73c6b6163311 | easy | 0Chris5R | 2026-08-07 09:01:44.901045+00:00 | succeeded | ac7f3bbb6cc7b7595716b6de88844e94c6e3b92dd1de90e257c66efbe7428137 | 30,189 | 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": "3f2fc392-5eb1-42e4-8dcc-73c6b6163311",
"created_at": "2026-08-07 09:01:44.901045+00:00",
"db_md5": "d75aaf0d05cb44735039a9335a14556e",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "bf885491-910b-4af2-8724-a5deece3780b",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.32781360225885,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012500000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.184136980852204,
"example_count": 600,
... |
3f36a335-db83-4b8f-8880-e7b87399f5e5 | easy | jamaliki | 2026-08-04 13:00:41.248401+00:00 | succeeded | a83731893204f98d55d91e3be5c753c3795defbf6f78b5c184bc7bc8b313a859 | 13,944 | null | """One Layer Deeper submission: clocked latent loop with FSQ restoration.
Design: an encoder reads the input once into a memory. A weight-tied core is
applied m times per macro-step over [register | scratch] slots, cross-attending
to the memory (input re-injection). At each clock edge, registers are snapped
through fi... | {
"id": "3f36a335-db83-4b8f-8880-e7b87399f5e5",
"created_at": "2026-08-04 13:00:41.248401+00:00",
"db_md5": "0ac28daa89579eefd1880c71486cc2f6",
"submitter": "Kiarash Jamali",
"github_login": "jamaliki",
"run_id": "14c283df-2a34-4aec-8968-aca41c986090",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 3.4970362186431885,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.023333333432674408
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.906998157501221,
"example_count": 100... |
3f3c2b6e-1282-486b-b74f-7f6b85e40751 | easy | JackCai1206 | 2026-08-13 04:03:08.426243+00:00 | succeeded | 888fa58db3e0499b619d8effdb8dff786f9a279d130ee51ed002a37d5d121f55 | 7,689 | null | """root — transformer++ iterated encoder, the tree's baseline node.
Principal directives baked in:
- STANDARD token embedding in, STANDARD linear unembedding over the vocab
at every position (loss masked by the evaluator's valid_mask). No numeric/
abacus/base-100 encoders here — those are CHILD nodes measure... | {
"id": "3f3c2b6e-1282-486b-b74f-7f6b85e40751",
"created_at": "2026-08-13 04:03:08.426243+00:00",
"db_md5": "89e2b0aace924998192d9295e1e13baa",
"submitter": "Ziyang Cai",
"github_login": "JackCai1206",
"run_id": "643788c3-b922-47ce-bc06-0d9974656e34",
"tier": "easy",
"dataset_id": "e1",
"status": "suc... | {
"score": {
"mean_loss": 4.997483015060425,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06333333253860474
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.231667518615723,
"example_count": 100,
... |
3f4d9a97-1f59-4490-a552-d35efb83000d | easy | karanganesan | 2026-08-05 10:10:49.692357+00:00 | succeeded | 0528fb86f84b45b8b93844d1231daf755bd038a775f03703d15f98f064a8cedc | 22,082 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "3f4d9a97-1f59-4490-a552-d35efb83000d",
"created_at": "2026-08-05 10:10:49.692357+00:00",
"db_md5": "bff81b5e7aa5e0bfbdbac6e724a3088e",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "4457d49b-183f-42ce-afb3-50de8cbe7617",
"tier": "easy",
"dataset_id": "e3",
"status": ... | {
"score": {
"mean_loss": 9.463072299957275,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011249999748542905
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.971543312072754,
"example_count": 800,... |
3f4e16e6-96ce-40b6-bf44-76e035bb08e8 | easy | khushidahi | 2026-08-03 02:58:38.313502+00:00 | succeeded | 9e0844229c8f67446e5b4e83025f9eb246e87f4a21b9d112a4b7a5097fe078f5 | 14,368 | 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": "3f4e16e6-96ce-40b6-bf44-76e035bb08e8",
"created_at": "2026-08-03 02:58:38.313502+00:00",
"db_md5": "85f33bf702f53bda659ca6d70a239922",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "cba4838a-233f-4c58-a39a-917faf96f246",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.2046669721603394,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007499999832361937
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1993634700775146,
"example_count": 60... |
3f4ea39c-29ca-4eda-8c9d-95d7605deab8 | easy | sankalp1999 | 2026-08-04 17:38:53.772235+00:00 | succeeded | cc2b7a3fc9b3e1f35c38d922c15aaf311ef03eec6f095ef149d043112c11b25e | 16,407 | null | """p3_neural: the neural branch of p2_program, alone and rule-clean.
Why this file exists
--------------------
p2_program scored 100.0% at T=64 by enumerating ~27k candidate transition maps
and evaluating them in exact int64 modular arithmetic inside `forward`. That
violates README rules 7 (no hard-coded algorithm in... | {
"id": "3f4ea39c-29ca-4eda-8c9d-95d7605deab8",
"created_at": "2026-08-04 17:38:53.772235+00:00",
"db_md5": "124bd95e2de2310af765cec37342ae94",
"submitter": "Sankalp ",
"github_login": "sankalp1999",
"run_id": "dbdb8d38-c758-4e25-8198-148d755407c8",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 2.3559730052948,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006249999860301614
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3943614959716797,
"example_count": 800,
... |
3f523aa2-28cf-48e4-83be-a61a017cc22b | easy | lpbb | 2026-08-19 13:06:07.336360+00:00 | succeeded | 87167bf85e9f82b364ffe6b08b38ad598dc88cb3bcda00608a6eb224ec210c43 | 9,886 | null | """S0 + RoPE + abacus embeddings + dropout, with a looped block: instead of
`num_layers` independent layers, a block of `num_layers` unique layers is run
`num_repeats` times with shared weights (e.g. 12 layers x 4 repeats = depth 48
but only 12 layers' worth of parameters). Weight tying across repeats forces
the block ... | {
"id": "3f523aa2-28cf-48e4-83be-a61a017cc22b",
"created_at": "2026-08-19 13:06:07.336360+00:00",
"db_md5": "b4b1711420ffcf1460247a674fb67679",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "87b1b28c-1bf3-4294-b03b-22b30e8f1576",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 3.1810911893844604,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.038333332166075706
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5706303119659424,
"example_count": 10... |
3f601226-715e-4c70-8c2e-fce00afcabd6 | easy | shreyash-chonkie | 2026-09-01 02:46:44.196876+00:00 | succeeded | fda5fcf858ed0df5105274a3b250320f40754a9bc29f869fc454c995a9c0f96b | 21,413 | null | """Learned token scanner feeding a random recurrent path reservoir."""
from __future__ import annotations
from itertools import permutations
import torch
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_s... | {
"id": "3f601226-715e-4c70-8c2e-fce00afcabd6",
"created_at": "2026-09-01 02:46:44.196876+00:00",
"db_md5": "fd5c9a383e4bdcea3df6a65c9fa689ba",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "2a30e65f-316c-4c43-bc2e-c8ca20aff746",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 24.70537567138672,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.024999999441206455
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 25.362804412841797,
"example_count": 100... |
3f69e34b-e165-41ac-92a4-1a9da51cd1c3 | easy | ayushnangia | 2026-08-11 13:35:53.699950+00:00 | succeeded | d167f3408f9f92d40e0f84849e465859b40fd4d750ed60c02b214f43ad8700ee | 39,843 | null | """rFA1 (FUNCTION-AGNOSTIC CAPACITY) — organizer intel 2026-08-11: hidden f
is "almost exactly the same, a very small algorithmic trick". Audit
(docs/hidden_inference.md) says the chassis's squaring priors are per-loop
compute budget and reduce provisioning. rFA1 relaxes them: SQUARE_SWEEPS
2->3, REDUCE_ROUNDS 4->6, EX... | {
"id": "3f69e34b-e165-41ac-92a4-1a9da51cd1c3",
"created_at": "2026-08-11 13:35:53.699950+00:00",
"db_md5": "07af170cd444b5747abd41e7f6e1b221",
"submitter": "Ayush Nangia",
"github_login": "ayushnangia",
"run_id": "b6c37a9f-514a-476a-b482-b59ad4cc163c",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.3413660526275635,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0054166666232049465
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.349837064743042,
"example_count": 60... |
3f6b60e7-e1d2-45f6-9f74-ac5e34e684e1 | easy | heathsanchez | 2026-08-17 03:25:01.482311+00:00 | succeeded | 4934893c986f3aecbb5029a7047cf9add53d59101cb2acba52eda2f5fd61ba76 | 3,178 | null | """E0002: exact baseline training, reuse the same block 4 times only in eval mode."""
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
EVAL... | {
"id": "3f6b60e7-e1d2-45f6-9f74-ac5e34e684e1",
"created_at": "2026-08-17 03:25:01.482311+00:00",
"db_md5": "645346bf92d0d3f61793a0e5142b5812",
"submitter": "Heath Sanchez",
"github_login": "heathsanchez",
"run_id": "38831dab-cf32-44f9-902e-385583e15a6d",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.3630857467651367,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0533333346247673
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4774155616760254,
"example_count": 100,... |
3f70fe8a-d0f6-464e-a4e1-cc415c2d1bbc | easy | ehonig | 2026-08-20 09:53:43.523774+00:00 | succeeded | fa7f3382fea16105a0527eb2d8014eb5b59c45c288ae44de41062d6e89dad21c | 24,717 | null | """A plain Transformer for repeated modular squaring, and nothing else yet.
This is a deliberate restart. The previous submission accumulated a field
parser, a carry scan, a learned reciprocal, periodic features, a digit
bottleneck, a depth selector and two cell types, most of them tuned around a
memorisation table th... | {
"id": "3f70fe8a-d0f6-464e-a4e1-cc415c2d1bbc",
"created_at": "2026-08-20 09:53:43.523774+00:00",
"db_md5": "9d411a2759da4ff3a26b1218ebd741ef",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "70704709-f428-4b0a-ae77-ec560ccd3951",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 2.217715620994568,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018124999478459358
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2164676189422607,
"example_count": 800... |
3f81358f-8a6b-4a0f-81ef-0438f9a2776e | easy | jordanrubin | 2026-08-07 16:56:59.581161+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": "3f81358f-8a6b-4a0f-81ef-0438f9a2776e",
"created_at": "2026-08-07 16:56:59.581161+00:00",
"db_md5": "671c15c2db39c821999bae96a7defa51",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "4f8d3166-e02e-4bb8-8d11-50b50e84edb5",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 5.12820373037631,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007499999832361937
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.171343184773697,
"example_count": 600,
... |
3f87172d-2589-4ced-9286-89d4258e1e7c | easy | sankalp1999 | 2026-08-04 11:50:04.041734+00:00 | succeeded | c47bba6372e39bf9d98eec823f895cf192c7905498ff29055297d65d170404ec | 7,643 | null | """Linear latent operator: squaring becomes a matrix, depth becomes its power.
The inductive bias comes from the group structure of the task. For modulus
N = p*q the units form Z_{p-1} x Z_{q-1} (for E1, 323 = 17*19 -> Z_16 x Z_18).
Writing x = g^e, one squaring step is e -> 2e. In a Fourier basis over that
group, the... | {
"id": "3f87172d-2589-4ced-9286-89d4258e1e7c",
"created_at": "2026-08-04 11:50:04.041734+00:00",
"db_md5": "5bb88420aed1afdda672ce6300077d37",
"submitter": "Sankalp ",
"github_login": "sankalp1999",
"run_id": "1fefd2bc-ab7a-4829-8acf-6a02aeaf1bd6",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 2.213739037513733,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014999999664723873
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3776350021362305,
"example_count": 100... |
3f8bb694-5492-47ee-a6af-fbd504685cd2 | easy | repst | 2026-08-09 21:42:54.813529+00:00 | succeeded | 557a96dc22877ddf1c1de2ebaf796a02bdf6b45968e841fea72cfc628b8bbf84 | 23,177 | null | """Strict forced-T operator-learning diagnostic for the public prompt contract."""
from __future__ import annotations
from typing import NamedTuple
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,... | {
"id": "3f8bb694-5492-47ee-a6af-fbd504685cd2",
"created_at": "2026-08-09 21:42:54.813529+00:00",
"db_md5": "3b3cded04942e8d1e0c1d123a2ab1aa4",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "971ae9cb-e00f-48eb-8f84-06a4b9e7fa54",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 6.828208923339844,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07333333790302277
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.588646411895752,
"example_count": 100,
... |
3f8cb76f-3213-48e7-a345-c8c651fd4479 | easy | benjaminW2025 | 2026-08-17 07:33:43.956621+00:00 | succeeded | 758b5a34beb9e9e499901404fd981b28077ed34fb527de28b3a7862c8289c13e | 3,665 | 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 = 256
NUM_HEADS = 4
... | {
"id": "3f8cb76f-3213-48e7-a345-c8c651fd4479",
"created_at": "2026-08-17 07:33:43.956621+00:00",
"db_md5": "20420785aac4bab89c77e00e9884eaff",
"submitter": "benjawesome",
"github_login": "benjaminW2025",
"run_id": "767d0e67-4581-4508-a4ff-afc66603e071",
"tier": "easy",
"dataset_id": "e5",
"status": "... | {
"score": {
"mean_loss": 2.399372952278413,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000024835269
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4509264075702615,
"example_count": 600... |
3f8ddda1-8867-4846-888b-d61919b8435b | easy | im-a-king-but-im-far-from-a-saint | 2026-08-10 15:27:37.861134+00:00 | succeeded | 7f0b68e22e19b808175ae97106794527cce91810c353b5d1b4278a9b5f40bcbb | 7,507 | null | """Weight-tied recurrent (Universal-Transformer-style) baseline.
Applies ONE shared transformer block NUM_LOOPS times instead of stacking
distinct layers. Re-injecting the original embedding at every iteration
keeps the recurrence anchored to the problem statement, which is what lets
this kind of network extrapol... | {
"id": "3f8ddda1-8867-4846-888b-d61919b8435b",
"created_at": "2026-08-10 15:27:37.861134+00:00",
"db_md5": "066266388acf70f1c1af3b93aa608a35",
"submitter": "im-a-king-but-im-far-from-a-saint",
"github_login": "im-a-king-but-im-far-from-a-saint",
"run_id": "5fbef117-ba15-46b8-9524-f445e8fae572",
"tier": "... | {
"score": {
"mean_loss": 2.2037822008132935,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.279421329498291,
"example_count": 100... |
3f8e494c-56aa-4aa1-a9b3-27259154af26 | easy | jordanrubin | 2026-08-13 17:59:35.597352+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": "3f8e494c-56aa-4aa1-a9b3-27259154af26",
"created_at": "2026-08-13 17:59:35.597352+00:00",
"db_md5": "3f710550312ea469d13b008d83980242",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "cb354486-4625-47a2-8bfd-90c7c99e98c4",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.8571816068563294,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02666666576017936
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.253236819008541,
"example_count": 600,... |
3f936e8a-ad41-4d79-b1af-04f11a1b58b5 | easy | DDanlov | 2026-08-12 22:54:50.202411+00:00 | failed | aa886e6293b4f8f0a992d1706241948ae9b4522e7fb6e34352acd0f2d272397d | 28,176 | null | from __future__ import annotations
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.checkpoint as cp
from torch.optim.optimizer import Optimizer
try:
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
... | {
"id": "3f936e8a-ad41-4d79-b1af-04f11a1b58b5",
"created_at": "2026-08-12 22:54:50.202411+00:00",
"db_md5": "b186f6143b43252b050da5a73a338247",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "e5c88912-525e-443a-b28f-ac699f1eef51",
"tier": "easy",
"dataset_id": "e2",
"status": "failed",
... | null
|
3f949788-f011-43b3-a18b-6dc66aca539e | easy | DDanlov | 2026-08-22 23:03:49.088394+00:00 | succeeded | 614fe4a738de99d9ef759ededf8f74b72b991eac1d264398363fafb17ca50c88 | 27,059 | null | """
Unified Hybrid Masked Deliberation Model (75.8M Parameters) - Variant 2 (Pure Skipless, No Cross-Attention)
Architecture:
- Scale: dim=1536, num_heads=24, d_ff=6144 (Exactly 75,826,176 parameters)
- Hybrid Prefix-Bidirectional + Causal Target Attention Mask
- Token Layout: [Prompt (L_p), 4 Pause Registers, Target T... | {
"id": "3f949788-f011-43b3-a18b-6dc66aca539e",
"created_at": "2026-08-22 23:03:49.088394+00:00",
"db_md5": "d00dc8acbeddc273f1ff00277d962b46",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "d4763cea-90fe-4709-b62e-42fa4e3aa205",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 140.69215925030818,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0050000000248352685
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 139.7448552563586,
"example_count": 60... |
3f9e67d9-3272-49d4-819a-54b095e6eefe | easy | viridale | 2026-08-03 22:41:20.278055+00:00 | succeeded | 2394f4270c2c90255fa93d69719c0f052fdcbb973bb18204abf1132dd6ef4f80 | 147,614 | null | """Mismatch-triggered maximal five-instruction recurrent-program tail.
hypothesis: the hidden recurrence may be a short exact program of length five;
this is the deepest complete 30-primitive word language below the ceiling.
axis: arch
target: revised h1 Max T at least1 while preserving every v149 learned value.
e... | {
"id": "3f9e67d9-3272-49d4-819a-54b095e6eefe",
"created_at": "2026-08-03 22:41:20.278055+00:00",
"db_md5": "5f506dfe69578f373a14e9f7cd430d78",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "a682dc13-57b2-4c3b-b11b-221be04cfb19",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 0.0030203700887918954,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0030203700470348266,
"example_count": 800,
... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.