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 |
|---|---|---|---|---|---|---|---|---|---|---|
5116edaf-be25-455b-9e54-2ed542a350d7 | easy | Cree0618 | 2026-08-29 06:37:27.456216+00:00 | failed | dce016ec42923bce04393ea95cf9a6fab8fd6c837e4436bec6f535327b922067 | 4,519 | null | """E008: E002 tied recurrence trained with StableMax cross-entropy.
Single-variable ablation: architecture and AdamW match E002; only the training
loss changes. StableMax uses a logarithmic transform before normalization,
reducing the incentive for already-correct logits to grow without bound.
Evaluation is unchanged ... | {
"id": "5116edaf-be25-455b-9e54-2ed542a350d7",
"created_at": "2026-08-29 06:37:27.456216+00:00",
"db_md5": "ec32db9d136751184cee1968edc3c011",
"submitter": "Cree0618",
"github_login": "Cree0618",
"run_id": "077801dd-f13c-4dc1-b7f0-00bc3d5408a4",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",... | null
|
5125bd47-f451-4226-81c6-357b5985357b | easy | oupadhyay | 2026-08-12 11:37:40.526813+00:00 | succeeded | f398b0ac128d3c4b265a6148aaac4f0de398294007f22885a83813887b7bc97d | 4,302 | null | """Width-shared three-tap place-mixing SGU for Easy E5."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,assert_model_state
W,D,H=4,112,224
class C:
def __init__(s,vocab_size,max_seq_len): s.... | {
"id": "5125bd47-f451-4226-81c6-357b5985357b",
"created_at": "2026-08-12 11:37:40.526813+00:00",
"db_md5": "0c8db6e6705fb099e085b5f2c8c3dad4",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "02e96713-db70-4ece-bf9e-9b999bf2e844",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.305236756629435,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000037252904
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.207192139775229,
"example_count": 600,... |
51282226-3717-46c4-82bf-1e055ab26a0a | easy | khushidahi | 2026-08-08 00:53:32.385675+00:00 | succeeded | 50ee2a5503edcc896870f60ad8863f23b6d570584f3fdbfab1faea928af8fe1b | 13,208 | null | """R42 Fourier-number representation experiment.
All input-dependent numeric parsing stays inside differentiable GPU tensor
operations. This is a learned-model representation, not a task solver: it does
not compute modular squaring or create intermediate labels.
"""
from __future__ import annotations
import math
impor... | {
"id": "51282226-3717-46c4-82bf-1e055ab26a0a",
"created_at": "2026-08-08 00:53:32.385675+00:00",
"db_md5": "9503add48bdfbc355a342651da411555",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "cd072e7e-ebfe-475f-8789-4344d6ae8468",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.163787364959717,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333590999246
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1713123321533203,
"example_count": 600... |
51308900-9af9-49bc-a1c8-9e98bfa5f59e | easy | velocizapkar | 2026-08-06 22:00:00.208744+00:00 | succeeded | d733ec17e1c982e0d89919d7fb7fc2d1371e8b10fecd6d489fb0ae5b402cce38 | 53,088 | 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": "51308900-9af9-49bc-a1c8-9e98bfa5f59e",
"created_at": "2026-08-06 22:00:00.208744+00:00",
"db_md5": "bcbccc953ecbf6f0021ca322952ca604",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "74c34caa-2402-4f48-aea1-5489c0528a8e",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 2.372032046318054,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003124999930150807
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5951318740844727,
"example_count": 800... |
513690a1-83f7-4531-8809-020654ac1c62 | easy | yashkant | 2026-08-12 03:58:09.523465+00:00 | succeeded | ecfcd5742f0a55d077f2c4fa900ca1d74ca7eec6530261c730786b5375dcbd37 | 31,389 | null | """Round1702 one-projection-per-T Dykstra recursion on the D416 carrier.
Each parsed public-T cycle applies one shared learned smooth projection and
alternates the A/B set code and correction memory across cycles. The immutable
anchor is reinjected on every call; all active cycle outputs are averaged, and
terminal CE... | {
"id": "513690a1-83f7-4531-8809-020654ac1c62",
"created_at": "2026-08-12 03:58:09.523465+00:00",
"db_md5": "8eebb046fd592d64c0456e35f3a4ba12",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "e15ada96-5f37-4d82-aa44-7fb4d2954ef2",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 3.4032700061798096,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.15125000849366188
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.616693496704102,
"example_count": 300,... |
5138d796-f1ec-48c7-9fb7-0a129c8942e9 | easy | DDanlov | 2026-08-15 20:56:26.221700+00:00 | succeeded | 3eb517cb9667863720ce75d439fd67ebc36b79a26534e0b145dad12284313550 | 17,950 | null | """
Official Submission for One Layer Deeper Challenge
Architecture: Deep Equilibrium Port-Hamiltonian Deliberation Network (Model C v3 / DEPH-Net)
- Input: Multi-Scale Linear Harmonic Torus Frequency Encoder (omega_j = 1..64)
- Recurrent Core: Continuous Port-Hamiltonian Dynamical Flow dz/dt = (J - R) * grad_E
* J =... | {
"id": "5138d796-f1ec-48c7-9fb7-0a129c8942e9",
"created_at": "2026-08-15 20:56:26.221700+00:00",
"db_md5": "623970f9d216ffb45b2ef4f8507de941",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "be1f27f3-c812-4231-bf7b-371ca165542a",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.3084014681225336,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.028333333507180213
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.227163528328511,
"example_count": 100... |
51464764-dbf4-4091-9c72-72af54a419a4 | easy | shirvani-jr | 2026-08-29 02:52:05.487683+00:00 | succeeded | 459da493dd2908cad9a8c9872fbad420462944f11d66ff0b4106740ae7ffd3e9 | 47,739 | 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": "51464764-dbf4-4091-9c72-72af54a419a4",
"created_at": "2026-08-29 02:52:05.487683+00:00",
"db_md5": "bd4bfcc15bf65f683c4a1d57d23e89cb",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "2ef12cf5-521f-409d-9e06-4ca275ea88d1",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.2034553587436676,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6937500238418579
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.9067326188087463,
"example_count": 300,... |
5147c318-2a11-4204-8a35-1ba861661f67 | easy | ehonig | 2026-08-19 16:49:53.812660+00:00 | succeeded | 84807a9b3fbdd1623e837d3c104c3be3068d7103ca97f7e3e356c67dbc4d2b64 | 16,887 | 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": "5147c318-2a11-4204-8a35-1ba861661f67",
"created_at": "2026-08-19 16:49:53.812660+00:00",
"db_md5": "d64131975e4b45cab4c8754d2100fe0b",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "a55c1733-6d40-4589-991c-3ddd2cbc45a2",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 4.811290264129639,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666772216558
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.654131889343262,
"example_count": 300,... |
5148a03d-c869-4ceb-9d9d-5bb833c9a31a | easy | khushidahi | 2026-08-07 04:32:00.590364+00:00 | succeeded | 8dc8a5bc1926d5ce6f0c29a81a4e990560cc625fd26f3c12205a1ef863d1eef7 | 16,969 | 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": "5148a03d-c869-4ceb-9d9d-5bb833c9a31a",
"created_at": "2026-08-07 04:32:00.590364+00:00",
"db_md5": "69f925fdf1d07d9fcf968d68901d4933",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "ffb8b9da-21a4-4400-a060-029efbe3d71c",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1970218420028687,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01000000024214387
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.196169376373291,
"example_count": 600,... |
514a7f4b-6296-4ae3-85aa-5abdde229f15 | easy | KaustubhKumar05 | 2026-08-29 04:49:10.637759+00:00 | succeeded | 89285038b28e448631ad6034fb5a7007c1cf5db04998c8d5583268facf41ab02 | 12,682 | null | """op2_t1w -- stock cell_op2 + T=1 loss upweighting (the scored rung).
Forked from cell_op2. ORIGINAL DOC BELOW.
cell_op2 -- ONE shared squaring operator with internal depth, applied T times.
WHY THIS SHAPE (verified on the real data, not assumed):
answer(T=t) == x squared t times mod N 20000 / 20000 e... | {
"id": "514a7f4b-6296-4ae3-85aa-5abdde229f15",
"created_at": "2026-08-29 04:49:10.637759+00:00",
"db_md5": "de146efc427260a28a698b29f1f546c3",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "8eef0b7e-f681-4eed-a054-87550fa2d503",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 2.8812400060505867,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000024835269
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.962239108545363,
"example_count": 600... |
5152a091-1b7a-4c48-8770-d5e5524def15 | easy | liam-gb | 2026-08-13 01:55:26.026970+00:00 | succeeded | 5ae4da782d01d42a5e2630f18f3c7dc668f4f397431ff7645051fa3182e49267 | 16,651 | null | """rns_pm_gather_ct_widet_drank: dense loop RETAINED, plus a parallel
residue track supplying the CRT rank.
Base: widet with lookup tables at lr 3e-2.
The wide readout is what makes long answers reachable; the trim is what pays
for it, by not running loop iterations whose update is multiplied by zero.
Base docstring... | {
"id": "5152a091-1b7a-4c48-8770-d5e5524def15",
"created_at": "2026-08-13 01:55:26.026970+00:00",
"db_md5": "1c479f1e07ec8e176359058d7ba96f62",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "0b61e827-489f-419e-bfd4-6aff4da6f3c8",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 9.14290165901184,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.10749999713152647
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 10.606706619262695,
"example_count": 800,
... |
51542a0f-2ef8-412b-b0b4-43382902c605 | easy | mnida | 2026-08-23 22:19:59.182245+00:00 | succeeded | d88ba702daf7a1be2b224e56215f45570d883b5ec2ab64d5afbe346634a3483a | 9,106 | null | """Wide fused recurrence with Muon on the shared transition matrix."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
asser... | {
"id": "51542a0f-2ef8-412b-b0b4-43382902c605",
"created_at": "2026-08-23 22:19:59.182245+00:00",
"db_md5": "fe47acadd8b35a256d0bbee5faf1024e",
"submitter": "mnida",
"github_login": "mnida",
"run_id": "fb540944-5c0e-43fc-9a06-1d416a34af8b",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 9.247302473017612,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0075
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.877849151200778,
"example_count": 600,
"ex... |
5154676e-3889-4b94-9ae8-6c194db92cad | easy | shirvani-jr | 2026-08-23 20:02:05.121112+00:00 | failed | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "5154676e-3889-4b94-9ae8-6c194db92cad",
"created_at": "2026-08-23 20:02:05.121112+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "d56a8a0a-083c-4d68-ac31-27b0718ad946",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
51565519-a5d3-4c5d-9912-bb1877e0f098 | easy | oupadhyay | 2026-08-29 18:57:58.055758+00:00 | succeeded | 927e29bdb8bf02bf526ee0411b7fceb6218d6d87b4aacacd4810822f0e32c17d | 4,316 | null | """Conditional structured-loss screen: hard25."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,TokenLossBatch,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,max_... | {
"id": "51565519-a5d3-4c5d-9912-bb1877e0f098",
"created_at": "2026-08-29 18:57:58.055758+00:00",
"db_md5": "35fea2a9bbee43d0bda9b76800373a66",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "736e6f90-70cd-4870-95ed-1cf60f9b6014",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 1.9799326062202454,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.043468470685184
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9565449953079224,
"example_count": 60,
... |
515ac2ae-50ef-4bf2-a356-d9d280541e02 | easy | ehonig | 2026-08-19 16:53:34.811336+00:00 | succeeded | 84807a9b3fbdd1623e837d3c104c3be3068d7103ca97f7e3e356c67dbc4d2b64 | 16,887 | 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": "515ac2ae-50ef-4bf2-a356-d9d280541e02",
"created_at": "2026-08-19 16:53:34.811336+00:00",
"db_md5": "d64131975e4b45cab4c8754d2100fe0b",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "e6babf3b-1ba1-4ef8-9d1d-3bfaa9444828",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.2212726806399523,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666610787312
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2434732913970947,
"example_count": 60... |
5165ad30-5319-4cf1-ac6c-4e16e84534fc | easy | amirmeisami | 2026-08-17 23:27:08.189338+00:00 | succeeded | 1f0bcbc0fc4080a6124372a41143122aa9ab3e8548e40de941e218393c326ecd | 13,666 | null | """T023 P4: D with compositional N-conditioned low-rank adapters."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, TokenLossBatch, assert_model_state
D_MODEL, POLY_WI... | {
"id": "5165ad30-5319-4cf1-ac6c-4e16e84534fc",
"created_at": "2026-08-17 23:27:08.189338+00:00",
"db_md5": "e1b946b154ef216e6ff886bf4c4543fb",
"submitter": "Amir Meisami",
"github_login": "amirmeisami",
"run_id": "f969f688-f590-4bd0-b2a6-18892e9b0d40",
"tier": "easy",
"dataset_id": "e3",
"status": "s... | {
"score": {
"mean_loss": 4.954162895099961,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0075
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.967719115175818,
"example_count": 800,
"ex... |
516b45d9-7687-4c2a-98e3-a44865bdfe68 | easy | DDanlov | 2026-08-17 18:56:38.306989+00:00 | succeeded | 03958423b1cafed9a40ea9c6759f891f45096810e488e6a1a1e5aceef745afc5 | 23,439 | null | import math
from typing import Optional
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark import Submission, assert_model_state, OptimizerBundle
except ImportError:
try:
from client import Submission, assert_model_state, OptimizerBu... | {
"id": "516b45d9-7687-4c2a-98e3-a44865bdfe68",
"created_at": "2026-08-17 18:56:38.306989+00:00",
"db_md5": "e99e4375f071c551a5c7e1087a4a0ff4",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "c95c46a7-ee27-4be2-8082-417f99012047",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.1580898767761423,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.033333333730697634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1532258934049464,
"example_count": 10... |
517bf721-79c2-448c-a1a9-03405ead8bc3 | easy | liam-gb | 2026-08-17 11:32:26.994166+00:00 | succeeded | 936ea05de67f15515fc3e81d838b067bcbeb44d73d467fc5f6b2156fc87ef97a | 15,810 | null | """rulemix: the per-prime step is a LEARNED MIXTURE OVER FIXED EXPONENT MAPS.
The identification-transfer body. Its predecessor lineage has per-class
freedom in the transition (pointwise pushforwards can be fit row by row), so
nothing forces uncovered residue classes to obey the rule identified on
covered ones. Here t... | {
"id": "517bf721-79c2-448c-a1a9-03405ead8bc3",
"created_at": "2026-08-17 11:32:26.994166+00:00",
"db_md5": "be7d4a810b4a0f98b680749deda31df0",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "8ee78d67-7f45-4a36-840a-90a3060dd649",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.424546718597412,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4750000014901161
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.690312147140503,
"example_count": 100,
... |
518169e0-d387-4f76-a93d-ee6619721e29 | easy | DDanlov | 2026-08-10 08:22:21.180966+00:00 | failed | 34d0b3d271411a11f23f01c44a58786a0b44ce76c62edd5041cdc490f5efc88c | 16,397 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class RohanShampoo(Optimizer):
"""Self-contained RohanShampoo optimizer with eigenvalue ma... | {
"id": "518169e0-d387-4f76-a93d-ee6619721e29",
"created_at": "2026-08-10 08:22:21.180966+00:00",
"db_md5": "1d0642f26e76ce078a6d206fddc3825a",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "a04ba2d8-277d-49a6-9900-8834767f2dc8",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
519060a3-182a-4fe6-85d6-3420b34635fd | easy | MarcoF1 | 2026-08-13 06:33:14.842603+00:00 | succeeded | d4139cff3f7d7d582408e261f81405d506195be9a670998b6a6f1c0fd01df6bd | 5,886 | null | """Weight-tied looped transformer; latent depth scales with the prompt's T.
Hypothesis treatment arm: one tied block learns "one modular-squaring step" and
is re-applied 2T+2 times, where T is parsed from the prompt's digit tokens.
Input re-injection and per-iteration RMSNorm keep the state bounded so the
recurrence c... | {
"id": "519060a3-182a-4fe6-85d6-3420b34635fd",
"created_at": "2026-08-13 06:33:14.842603+00:00",
"db_md5": "174ef6cbc2b969ef234ceae8c3b7a915",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "a678bd55-d400-4a24-9964-a294c5a83fc2",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.1196996339130543,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03666666721304258
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.172919759109839,
"example_count": 100,... |
51985bf2-e008-418c-9ea9-691b066c77e0 | easy | alirezashirvani-jr | 2026-08-29 09:44:58.229679+00:00 | succeeded | 2b0223ff035c5c9c551319a2bfe8d9af881764d80db781b124aabaee8d426ac1 | 32,601 | null | from __future__ import annotations
import math
import time
from contextlib import nullcontext
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
MODEL_... | {
"id": "51985bf2-e008-418c-9ea9-691b066c77e0",
"created_at": "2026-08-29 09:44:58.229679+00:00",
"db_md5": "050524d36ff9ab09e87d1ff83923e1e0",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "f08047da-f663-4c49-879c-e2893034ab04",
"tier": "easy",
"dataset_id": "e7",
... | {
"score": {
"mean_loss": 1.903087079524994,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6761141121387482
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.354258418083191,
"example_count": 85,
... |
519ccba7-6aaf-49b0-9063-b4f6fbd8ea6d | easy | jagilley | 2026-08-22 19:10:01.920469+00:00 | succeeded | 104519ed9c6f9656e69235393773cb601d78c8a0a7e71456182e6d3ab665fbdf | 27,949 | null | """One Layer Deeper submission — a tied operator applied `T` times, with an optional
label-free closure loss, ported from `experiments/one_layer_deeper/ballistic_depth/`.
Two arms, selected by `ARM` below. They share **every parameter of the model**; only the
loss differs, which is how the research node runs its arms ... | {
"id": "519ccba7-6aaf-49b0-9063-b4f6fbd8ea6d",
"created_at": "2026-08-22 19:10:01.920469+00:00",
"db_md5": "e0c31ad352e9dcb73694c22eed607cfc",
"submitter": "Jasper",
"github_login": "jagilley",
"run_id": "7a9deb77-0217-4990-b0fb-90222d023d61",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.6176296655296802,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.775365762111852,
"example_count": 600... |
51a23b63-59f7-4283-b78f-79d1cbd22047 | easy | khushidahi | 2026-08-17 16:45:09.105549+00:00 | succeeded | 8c424cd4cc81be4dd096606f0a7b6cff5c12c911df71676088134d201bb2a697 | 29,488 | null | """R216: learned carry-then-reduce categorical machine.
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 reduction.
The t... | {
"id": "51a23b63-59f7-4283-b78f-79d1cbd22047",
"created_at": "2026-08-17 16:45:09.105549+00:00",
"db_md5": "7738543b5732fb508dd4c90dce36f39f",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "eaa26ad7-5274-4c5f-a668-f8322f21ae52",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 0.7945809960365295,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.846846878528595
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.4164184331893921,
"example_count": 60,
... |
51a74280-8f1f-4a28-9fbd-d2e8515fe9cc | easy | lpbb | 2026-08-18 10:14:01.242158+00:00 | succeeded | 08624877ad1e14f7dd58095ec1546741133065f0953424db474c99675ba8ae6b | 6,792 | null | """S0 + abacus embeddings: same 12 untied layers as s0, plus a per-digit
significance tag (units=1, tens=2, ...) embedded additively, with random
per-row offsets at train time so tag rows beyond the trained digit counts
still receive gradient (McLeish et al. 2024)."""
from __future__ import annotations
import torch
i... | {
"id": "51a74280-8f1f-4a28-9fbd-d2e8515fe9cc",
"created_at": "2026-08-18 10:14:01.242158+00:00",
"db_md5": "1261a09daff5e186c8b81298638b0718",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "2a9bd33b-36e1-455d-864e-1b3c506d0d6c",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 3.962056040763855,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04999999888241291
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.035905361175537,
"example_count": 100,
... |
51ac5e7a-59d6-4e40-9973-5e6bfa73a695 | easy | DDanlov | 2026-08-18 09:01:28.802618+00:00 | succeeded | 2daa3dabbf47e1f004651c2e7895944312e775b363dd9a9f753435d4e887457f | 23,511 | null | """
Official Submission for One Layer Deeper Challenge
Architecture: 4-Layer Recurrent Core Pure No-Norm No-Skip DKS Reasoning Model:
- 4 Distinct Pure No-Norm No-Skip Transformer Layers in the Recurrent Core: L1 -> L2 -> L3 -> L4
- Unrolled T times for serial modular squaring
- dim=512, 16 heads, d_ff=1024, RohanShamp... | {
"id": "51ac5e7a-59d6-4e40-9973-5e6bfa73a695",
"created_at": "2026-08-18 09:01:28.802618+00:00",
"db_md5": "a45a27d6e56da3f75f753d1345b5cf43",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "ae65225b-4bbe-4def-abff-80938b99e533",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.42655863643367,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.050000000447034836
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3603463857921203,
"example_count": 100,... |
51acff9d-9949-4f5e-b3f9-0ceb87035845 | easy | sapient-sapiens | 2026-08-20 15:11:13.313969+00:00 | succeeded | 0c0a3fe3363bbd71371446fb3add6f96fd2008b94ad7b0f2d923a3152bac4110 | 38,861 | null | """Closed learned spectral interface around a categorical recurrent operator.
The recurrent state is a bank of right-aligned categorical digit slots. Each
application uses a shared, shallow N-to-X conditioner to create learned phase
coordinates, injects paired sin/cos features as a small residual, applies the
ordinar... | {
"id": "51acff9d-9949-4f5e-b3f9-0ceb87035845",
"created_at": "2026-08-20 15:11:13.313969+00:00",
"db_md5": "8c774f8d3eaf9e85152ab098a32d6710",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "d3bac1c2-e6b3-4650-bfaa-9080cec952c0",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1414637565107117,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0070833333457509685
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.139398120443917,
"example_count": 60... |
51adb1eb-4855-49f2-a730-5b851aa42769 | easy | himalalps | 2026-08-06 12:34:56.126826+00:00 | succeeded | e99d4a2096ba14060657659fb8ab58f44b5460967f0beb5015c5d7921f54568c | 54,914 | 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": "51adb1eb-4855-49f2-a730-5b851aa42769",
"created_at": "2026-08-06 12:34:56.126826+00:00",
"db_md5": "5920650778735ac4e3cff877fade1373",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "2be35ad6-9a66-45d8-815c-50cbd2fe0239",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 12.454481826574952,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010416666623204946
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 13.27503776550293,
"example_count": 600... |
51b671a2-ffb3-408c-b6b7-494a862f8c25 | easy | alirezashirvani-jr | 2026-09-01 00:00:41.736802+00:00 | succeeded | a33e1f5ed935ebefd724123a83fa832a540218a7f87618a10b665e79ead7769d | 20,681 | null | from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
PAD = 0
MODULUS = 2
VALUE = 3
DEPTH = 4
ANSWER = 5
EO... | {
"id": "51b671a2-ffb3-408c-b6b7-494a862f8c25",
"created_at": "2026-09-01 00:00:41.736802+00:00",
"db_md5": "711d90661d2f24e3ada92579cc60544c",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "311038a1-564b-4bf9-acc4-b8a30297b14c",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 3.518680348689093,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02375000004967054
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5733610396962527,
"example_count": 600,... |
51bbba55-582b-499b-9d88-6fcb3f6387f1 | easy | khushidahi | 2026-08-17 15:57:18.489439+00:00 | succeeded | 2b43cd58227e1489f11de66f7a32bc7d3d356dc5a95e6794442b9cd0bbd80b9c | 23,980 | null | """R207: batch-512, one-shared-visit-per-digit serial Transformer.
This is the interaction arm of a preregistered 2x2 factorial. Relative to
R201 it combines the independently tested 512-example batch and one shared
block visit per supplied modulus digit. No representation, parameter, loss,
annealing, optimizer, or ... | {
"id": "51bbba55-582b-499b-9d88-6fcb3f6387f1",
"created_at": "2026-08-17 15:57:18.489439+00:00",
"db_md5": "6d0c47be8deb519972e04233eeff23c2",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "81208bc0-a654-44f0-940b-e44002a73762",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 2.51001513004303,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033783784601837397
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.571774959564209,
"example_count": 60,
... |
51bca395-959d-4b3f-bde6-151f731a247f | easy | oupadhyay | 2026-08-24 00:16:40.232093+00:00 | succeeded | af0c1046b7d9348d5dbd46185276c12789eae9be7a86e47c8d26a3dab572a79d | 4,319 | null | """Conditional structured-loss screen: hard50."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,TokenLossBatch,assert_model_state
W,D,H=4,128,64
class C:
def __init__(s,vocab_size,max... | {
"id": "51bca395-959d-4b3f-bde6-151f731a247f",
"created_at": "2026-08-24 00:16:40.232093+00:00",
"db_md5": "de06dca064b0ab0ef142cec4577c8cc5",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "6a6beebc-7548-4163-a0ad-69ac42e20ef4",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.977686027503246,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010416666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.626277244679062,
"example_count": 600,... |
51bfd052-2199-4b9b-a275-d9e3a81ae32a | easy | shreyash-chonkie | 2026-08-24 23:55:29.982918+00:00 | succeeded | d64bbafa4a82d5d78982aec75998bcfec3e3d416bb3f8c44ff63efcf77aa93f5 | 15,459 | null | """Recurrent attention with improvement-weighted pool feedback."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_mo... | {
"id": "51bfd052-2199-4b9b-a275-d9e3a81ae32a",
"created_at": "2026-08-24 23:55:29.982918+00:00",
"db_md5": "1253f45c068549baf3c3f4905f581fdf",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "18a30eb9-efcd-402a-9dc7-c071af4defab",
"tier": "easy",
"dataset_id": "e7",
"status": "... | {
"score": {
"mean_loss": 3.47642320394516,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06390374526381493
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.04922342300415,
"example_count": 85,
... |
51c88f83-dcde-40c3-9004-86d066ca9173 | easy | AdityaVG13 | 2026-08-13 05:07:01.009508+00:00 | succeeded | fcee10e53df29724eda3472b6737a39939d12e7ce7a01fb610a66dae8f65118e | 103,240 | null | """AEGIS V6 aegis_primal384: Adaptive Epistemic Guarded Interval Synthesis.
One Layer Deeper submission candidate. All task transitions remain learned
end-to-end. The deterministic shell only identifies marker-delimited roles,
constructs a finite semantic tape, schedules learned interval programs, and
applies a trival... | {
"id": "51c88f83-dcde-40c3-9004-86d066ca9173",
"created_at": "2026-08-13 05:07:01.009508+00:00",
"db_md5": "b89d721b45c3bb4c524efca4d399d905",
"submitter": "AdityaG",
"github_login": "AdityaVG13",
"run_id": "7ad46be6-d97f-45cf-83ec-e537fb75005a",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 2.17783784866333,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6466666758060455
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.8514373302459717,
"example_count": 100,
... |
51cf52b7-53a9-4adf-a3c8-1b6e9bb68291 | easy | briheet | 2026-08-05 12:43:18.186765+00:00 | succeeded | f7cadd0006d031f819ecb6479e2c498ae5ca88c8481f282d543897059cf372bb | 5,752 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 96
NUM_HEADS = 4
LI... | {
"id": "51cf52b7-53a9-4adf-a3c8-1b6e9bb68291",
"created_at": "2026-08-05 12:43:18.186765+00:00",
"db_md5": "e2d9a227ffff5532e925f9d4df6db3b4",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "3f2c5266-34fb-4ee1-b11e-397e2152b58f",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 4.8272868394851685,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.016666666604578495
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.858090400695801,
"example_count": 100... |
51d04e71-0de9-47dd-85bd-403eb8e99651 | easy | yashkant | 2026-08-19 02:06:06.235797+00:00 | succeeded | 4bf48738e795a7ee1545a7695c8f22652e00337f245c69b97ce43d83ecaf31c1 | 46,509 | null | """Round2202 factorizes one tiny recurrent correction into residue and carry.
The measured correction-reset official carrier stays intact. A pooled
four-value/four-carry state runs one shared transition twice per parsed-T
cycle and reinjects its immutable input halves on every call. Carry is
learned from signed resi... | {
"id": "51d04e71-0de9-47dd-85bd-403eb8e99651",
"created_at": "2026-08-19 02:06:06.235797+00:00",
"db_md5": "a1a52e211c1ef193c4ff2934dfdc88e1",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "efff9437-bb38-431f-9b02-668b0a06f798",
"tier": "easy",
"dataset_id": "e4",
"status": "succeed... | {
"score": {
"mean_loss": 2.0698853208929586,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0104166666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.07255501460839,
"example_count": 1200,
... |
51d33863-62bf-46f6-8994-6a8ea3309be1 | easy | gauravmishra | 2026-08-09 04:34:18.088623+00:00 | succeeded | 606e9546cab04197af68be717bb1565d0fb786ec35997772861a0cdd519e8462 | 5,040 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
D_MODEL = ... | {
"id": "51d33863-62bf-46f6-8994-6a8ea3309be1",
"created_at": "2026-08-09 04:34:18.088623+00:00",
"db_md5": "6b0aa018f1a89db4b9a193d6e9fe8584",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "767c0a52-ee00-4628-9fad-95373e1784a7",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.4405653194650725,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0054166666790843015
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.488856858736731,
"example_count": 60... |
51dbb27a-0aab-4c83-be3d-f81325df4729 | easy | sapient-sapiens | 2026-08-15 08:16:37.342940+00:00 | succeeded | 0a3d0d0c3f9126ecc23c14b7cfa2d3cbbf2b7d83fde4082aa153a5e8db313f14 | 10,993 | null | """Narrow/deep looped Transformer with optional latent workspace tokens.
Research candidate for sampled-modulus fixed-T=2. The encoder builds immutable
prompt memory. A stack of learned operator blocks updates a mutable latent
stream, and the whole stack is reused for each task recurrence. Optional blank
workspace ... | {
"id": "51dbb27a-0aab-4c83-be3d-f81325df4729",
"created_at": "2026-08-15 08:16:37.342940+00:00",
"db_md5": "69883ebcb5ec3d0fc3ae9bc2dc8c7f5b",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "229a1a46-ec9e-4ba9-94cc-d394645faea0",
"tier": "easy",
"dataset_id": "e4",
"status": "su... | {
"score": {
"mean_loss": 2.064789179194178,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005520833333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0678909567380104,
"example_count": 120... |
51dc02fc-0602-4bf5-887f-635710246997 | easy | TZZheng | 2026-08-06 17:42:40.566361+00:00 | succeeded | 9dcc13448f3f059d4804332205f458917908d8596428cee66b77c7b7b1dd76a3 | 7,304 | null | """Self-contained standard Transformer scaling-sweep submission."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
VARIANT = {'block... | {
"id": "51dc02fc-0602-4bf5-887f-635710246997",
"created_at": "2026-08-06 17:42:40.566361+00:00",
"db_md5": "babdc3c69a8a5aea5cff338c5c9190e2",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "a9e2b129-b021-4758-a6f2-86754dd01468",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.637149214744568,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05333333043381572
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.286515235900879,
"example_count": 100,
... |
51e46f7b-6411-47e3-b0d2-7b3ff5a6e288 | easy | DDanlov | 2026-08-17 12:03:19.012090+00:00 | succeeded | 5b4c9d9f506696c649f39a5b79a8678c6f68ebb2d0634f7f47bd27e0fe36e626 | 31,564 | 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": "51e46f7b-6411-47e3-b0d2-7b3ff5a6e288",
"created_at": "2026-08-17 12:03:19.012090+00:00",
"db_md5": "c40981f41bc7e76677edae4b27f3a122",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "c5b4a8c3-0ba4-42b0-bf9d-33bb71b14a09",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.149700341527896,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05166666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.363011598587036,
"example_count": 100,
... |
51e6be4b-a264-41a4-8089-27f8666e0355 | easy | AFeinfield | 2026-08-13 00:06:42.018548+00:00 | succeeded | 9d2b42dcce7b8da4f32c3dc80dcdcc01c76e55a0b5979888e7fdca149184cf1b | 6,581 | null | """Looped (weight-tied) bidirectional Transformer, d=256, 8 iterations.
012: 011 + weight decay 1.0 — grokking push now that reuse unlocked memorization
to amortize the evaluator's ~200ms/iteration overhead into 8x more steps.
prior. Same parameter count order as a 1-block model; depth comes from reuse.
Warmup + cosin... | {
"id": "51e6be4b-a264-41a4-8089-27f8666e0355",
"created_at": "2026-08-13 00:06:42.018548+00:00",
"db_md5": "325e109268fa99d365b41d4a7e0378d0",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "1f9d7186-76e4-41b8-ab48-d3e6bbed65a2",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 7.398353576660156,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.776507377624512,
"example_count": 100,
"exact... |
51ec696c-c662-4cff-952e-2f0850c0ed76 | easy | chuk-uzowihe | 2026-08-12 17:16:24.209135+00:00 | succeeded | 0db6ebecee0cce5916759bd6e4cc8729d19c16dd233d5dce90219ab970e30120 | 22,601 | null | """v10: dense fixed-loop recurrence with deep supervision.
Lineage: FPRM structure (alpha/beta-tied residual scaling, pre-rms_norm,
input injection every iteration) + the v8 2D digit lattice / RoPE task
encoding. Three deliberate departures from the v9 MoEUT line:
1. **Full BPTT.** The loop runs a fixed N_ITER iterat... | {
"id": "51ec696c-c662-4cff-952e-2f0850c0ed76",
"created_at": "2026-08-12 17:16:24.209135+00:00",
"db_md5": "fad72fcd4c5f0c42cc77b784fa7142ee",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "e6da2e8b-9cf0-4160-bcad-2102123777cb",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 9.990817546844482,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.341325759887695,
"example_count": 100,
"exact... |
51f3588d-03d2-4aad-9dd5-d11d07144bc6 | easy | im-a-king-but-im-far-from-a-saint | 2026-08-12 10:26:06.235143+00:00 | succeeded | 8c567285bede8dbd9ddea815dcd72915489b360745d581e54e37d772d9f6fb9a | 8,393 | 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 extrapolat... | {
"id": "51f3588d-03d2-4aad-9dd5-d11d07144bc6",
"created_at": "2026-08-12 10:26:06.235143+00:00",
"db_md5": "f5f1dca9fe28a4dc2576cec32eae275f",
"submitter": "im-a-king-but-im-far-from-a-saint",
"github_login": "im-a-king-but-im-far-from-a-saint",
"run_id": "36e88b6c-8c4b-47db-af02-bbcfe991062b",
"tier": "... | {
"score": {
"mean_loss": 2.4417275535445038,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000037252904
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.399186026355076,
"example_count": 600... |
51f8637a-64ac-4438-a664-538c795db9f4 | easy | anirudh-chakravarthy | 2026-08-12 01:42:41.557998+00:00 | succeeded | 88cedd182366010fddb93569c44d365afda1cc299ef5825ce605e7267028a892 | 14,649 | null | """A learned state transition applied once per requested time step."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
asser... | {
"id": "51f8637a-64ac-4438-a664-538c795db9f4",
"created_at": "2026-08-12 01:42:41.557998+00:00",
"db_md5": "595528a92fc59380659ee84f88e33146",
"submitter": "Anirudh S Chakravarthy",
"github_login": "anirudh-chakravarthy",
"run_id": "3a436dc2-69b7-4aee-a684-361a8b86993a",
"tier": "easy",
"dataset_id": "... | {
"score": {
"mean_loss": 3.0133921726519235,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8746443992238406,
"example_count": 60... |
51fce049-d34e-4385-a36f-48f1e67ab454 | easy | ipsa-loquitur | 2026-08-26 16:24:19.776891+00:00 | failed | 2d9f4f3331ca419636cb39310881dafc828cb2a064f881d8fef0e77a7487ffa4 | 15,929 | null | """Standalone NetF-V1 submission candidate for One Layer Deeper.
The learned rule is shared across bit positions, modular-square iterations,
moduli, and tape widths. During training, exact double-and-add states provide
dense one-step supervision. During evaluation, the model receives no state
corrections and rolls out... | {
"id": "51fce049-d34e-4385-a36f-48f1e67ab454",
"created_at": "2026-08-26 16:24:19.776891+00:00",
"db_md5": "faf08fffbdb0cf932240bbeff7141be0",
"submitter": "ipsa-loquitur",
"github_login": "ipsa-loquitur",
"run_id": "2385ba69-26d0-41df-b1ba-29aed05adef4",
"tier": "easy",
"dataset_id": "e1",
"status":... | null
|
5205354c-c8f2-41a3-8823-4ebc67134dbc | easy | gauravmishra | 2026-08-09 18:51:56.221840+00:00 | succeeded | d7993f55d1c880e09222ac6181e0202c5148955f9f85a35c895d061e8f7080a2 | 6,078 | null | """Compound-scaled Transformer with distinct blocks and PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_st... | {
"id": "5205354c-c8f2-41a3-8823-4ebc67134dbc",
"created_at": "2026-08-09 18:51:56.221840+00:00",
"db_md5": "fee088af557905af4b4c8510258e4892",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "4188f23a-2987-4d45-8b1f-85cf59a4e1ad",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.4341536001024444,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00625
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5273795940416277,
"example_count": 600,
... |
521f193d-792d-4183-990a-15c221d44874 | easy | alirezashirvani-jr | 2026-08-31 07:28:23.494667+00:00 | succeeded | c252ce15d535434ae9695254770a32b96e97a7120717f3216a16792d4ff23118 | 34,903 | null | from __future__ import annotations
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
TRACK_FLOOR = 3
TRACK_CEILING = 96
STAGE_B... | {
"id": "521f193d-792d-4183-990a-15c221d44874",
"created_at": "2026-08-31 07:28:23.494667+00:00",
"db_md5": "b9699e2f9827d45ad6df7d21eed39459",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "52f80d2e-0655-477e-906b-aac2204a3ff3",
"tier": "easy",
"dataset_id": "e1",
... | {
"score": {
"mean_loss": 0.2234562337398529,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9633333384990692
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.2714194655418396,
"example_count": 100,... |
5222b2b5-cd18-4f2c-a1c3-e28e354d3e35 | easy | 0Chris5R | 2026-08-02 15:32:18.538287+00:00 | failed | eb7b0b452c4c32c8fa8348e58f5d540aac0f19e62e0935e31ae0a7e7dcc34d94 | 44,689 | null | from dataclasses import asdict, dataclass
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
SUBMISSION_STARTED_AT = time.monotonic()
TRAINING_BUDGE... | {
"id": "5222b2b5-cd18-4f2c-a1c3-e28e354d3e35",
"created_at": "2026-08-02 15:32:18.538287+00:00",
"db_md5": "427cc82613eda3b3efb0daf04dfd4a35",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "2fda4e3c-66be-4d4d-bfd1-d2282823645e",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
522df38f-993c-4356-ac0c-77f3cb301e7a | easy | oupadhyay | 2026-08-29 17:25:57.172781+00:00 | succeeded | 7685f8865d8b205b1b0bdb24b3ee7e77bcf9f3b47438d3a399b5bc622cfb56cf | 3,988 | null | """Conditional optimizer screen: lr5."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_size,s.... | {
"id": "522df38f-993c-4356-ac0c-77f3cb301e7a",
"created_at": "2026-08-29 17:25:57.172781+00:00",
"db_md5": "15672fe33fcc64bebb17b51e7067abc1",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "30dfce75-7a2c-4113-b4b4-06aec8d54679",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 1.9974483251571655,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05855856090784073
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9449903964996338,
"example_count": 60,... |
522f1723-e2cf-4711-acff-13785a81fda0 | easy | liam-gb | 2026-08-13 02:18:33.099492+00:00 | succeeded | 5ae4da782d01d42a5e2630f18f3c7dc668f4f397431ff7645051fa3182e49267 | 16,651 | null | """rns_pm_gather_ct_widet_drank: dense loop RETAINED, plus a parallel
residue track supplying the CRT rank.
Base: widet with lookup tables at lr 3e-2.
The wide readout is what makes long answers reachable; the trim is what pays
for it, by not running loop iterations whose update is multiplied by zero.
Base docstring... | {
"id": "522f1723-e2cf-4711-acff-13785a81fda0",
"created_at": "2026-08-13 02:18:33.099492+00:00",
"db_md5": "1c479f1e07ec8e176359058d7ba96f62",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "80dbd61f-2cb8-4f48-bf7e-9260a0ba7e72",
"tier": "easy",
"dataset_id": "e4",
"status": "succeeded"... | {
"score": {
"mean_loss": 4.1129579267942775,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018437499962747097
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.080996543777839,
"example_count": 120... |
524c3210-893d-4029-9ce7-500f6756f200 | easy | lpbb | 2026-08-07 15:26:37.904934+00:00 | succeeded | 576fd19fd500d5ee1351b633b3a0cd5d8ad4f53b399c9717d12379bda8d68dd1 | 19,374 | null | """VARIANT R2a `tiny` — capacity-threshold axis.
Round-1 verdict (grokfast / swiglu / smooth, all H100): every basis and
optimizer finds memorization as the FIRST minimum; regularization and
confidence-taxing only delay or bound it, never redirect it. The one
force that makes memorizing 3200 rows impossible rather th... | {
"id": "524c3210-893d-4029-9ce7-500f6756f200",
"created_at": "2026-08-07 15:26:37.904934+00:00",
"db_md5": "89101f64b970b4c5946a4bc178274815",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "d3018461-2857-4354-8811-13b70d6fce55",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 17.382164734780964,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000004656613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 20.101452019017035,
"example_count": 80... |
524d79ea-e97f-423e-8bb3-433708cfe458 | easy | DDanlov | 2026-08-24 20:40:50.231890+00:00 | succeeded | 46f1d143bc24dee6f140a4816284cece236bfd64d98776e929b984526a66de2b | 49,568 | null | """
Unified Pure Skipless Prefix Deliberation Model + Transformer Readout Decoder
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 tokens in loop)
- Recurrent At... | {
"id": "524d79ea-e97f-423e-8bb3-433708cfe458",
"created_at": "2026-08-24 20:40:50.231890+00:00",
"db_md5": "f01bef0f2cdfc6f8812aee458fb7c8e0",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "f3e51e1f-952d-4b1f-994e-af4fdaafbede",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.675495072466645,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.711159852588123,
"example_count": 600,... |
5252bdf9-8080-464a-a719-1839686f06c5 | easy | adamsoliev | 2026-08-01 23:32:22.133332+00:00 | succeeded | a934393c6cbafb8e3bd8959983ca4886cb057bd9dbe26518f0ecd34fb67ea854 | 13,491 | null | """Range-independent digit-compositional refinement model."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
from torch import Tensor, nn
WIDTH = 256
... | {
"id": "5252bdf9-8080-464a-a719-1839686f06c5",
"created_at": "2026-08-01 23:32:22.133332+00:00",
"db_md5": "d5bff8a1123890edf0f05a800bd7904d",
"submitter": "Adam Soliev",
"github_login": "adamsoliev",
"run_id": "c4c71d98-dc75-4c91-930c-0600dc886112",
"tier": "easy",
"dataset_id": "e5",
"status": "suc... | {
"score": {
"mean_loss": 9.479588508605957,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01791666680946946
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.627457618713379,
"example_count": 600,
... |
52559fed-b90e-4442-8f83-abcdb2ba2d41 | easy | isaac0804 | 2026-08-07 11:31:59.539946+00:00 | succeeded | 9f65da167b4517a48589765cafa0163b79bb669ecd7c84a5e2e345a0ad96c5ba | 8,400 | null | """nGPT: Normalized Transformer with Representation Learning on the
Hypersphere (Loshchilov et al., arXiv:2410.01131), simplified port.
Motivation: a local diagnostic session on E3 (docs/findings.md) found that
the standard RMSNorm pre-LN baseline (`baseline_adamw_deep`) grows an
unbounded residual-stream norm as trai... | {
"id": "52559fed-b90e-4442-8f83-abcdb2ba2d41",
"created_at": "2026-08-07 11:31:59.539946+00:00",
"db_md5": "62d797e10c25ac9eef5272ca5002343d",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "70c18cea-dd75-49a8-9b67-b77cdb631f7b",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.870945692062378,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333656191826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.80770742893219,
"example_count": 100,
... |
525a1c4f-2a12-4e0f-961b-b2f970e0e138 | easy | DDanlov | 2026-08-26 00:36:03.311780+00:00 | failed | bfd38b233bd59544cdbc9c2c7678880f7a1b214837499d90997ee27ddccba577 | 35,466 | null | import math
import time
from typing import Optional, Tuple, Any, Dict, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
def log_debug(msg):
with open("debug_log.txt", "a") as f:
f.write(msg + "\n")
try:
from benchmark.api import ModelSpe... | {
"id": "525a1c4f-2a12-4e0f-961b-b2f970e0e138",
"created_at": "2026-08-26 00:36:03.311780+00:00",
"db_md5": "191cd2c134e6cf2c59ddb8e65e1f740d",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "31467059-783d-4be4-8b1e-de752ccbac17",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
52639290-254a-47cf-8a20-f5877c3508e2 | easy | karanganesan | 2026-08-08 03:15:52.492592+00:00 | succeeded | 78f2c917833663bce5afc4602118af38067b3a9d8fa3141bb33a1979295f8a0f | 27,203 | 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": "52639290-254a-47cf-8a20-f5877c3508e2",
"created_at": "2026-08-08 03:15:52.492592+00:00",
"db_md5": "c7ee3c4023ef1b1deada3757c3787e83",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "f6d15817-8e22-4c1e-b14f-e921e4ec5d81",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 3.5674434900283813,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007499999832361937
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.687122106552124,
"example_count": 600... |
526775c8-e1f7-4f14-9199-f8d74ab66458 | easy | sapient-sapiens | 2026-08-17 17:49:27.906956+00:00 | failed | daa561c059aa45550df039a17614a41f416672556d582aea67a21d8c7b93e4de | 23,207 | null | """Compute-matched v26 control: uncapped steps, MAX_CHUNKS=8, SAM kept.
Architecture changes relative to the hosted ACT max-4 source:
- selectable recurrence depths are 1 through 64;
- canonical cumulative-mass halting genuinely removes halted examples from
later recurrent applications;
- the T-field representation... | {
"id": "526775c8-e1f7-4f14-9199-f8d74ab66458",
"created_at": "2026-08-17 17:49:27.906956+00:00",
"db_md5": "fdd7ee09df2d8dc66c4dc30e9bfb33ef",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "47858b6f-f0e6-4443-ba1e-71cbeb27d30b",
"tier": "easy",
"dataset_id": "e6",
"status": "fa... | null
|
526a4219-4dc4-4410-b947-177deec2a325 | easy | jamaliki | 2026-08-04 17:31:18.470565+00:00 | succeeded | 65c8f21268d26b8100fe5c6a9562b915957d1431b8ce4423dd9237eb11851ed7 | 19,605 | 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": "526a4219-4dc4-4410-b947-177deec2a325",
"created_at": "2026-08-04 17:31:18.470565+00:00",
"db_md5": "de5b3651a54b3ab8942871c6622fdd1d",
"submitter": "Kiarash Jamali",
"github_login": "jamaliki",
"run_id": "c3c050b5-716a-43d1-a253-c229a77e15be",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 16.760818004608154,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.019999999552965164
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 14.439312934875488,
"example_count": 10... |
52715b70-102a-4f53-b1ca-21f95569dd37 | easy | DDanlov | 2026-08-12 20:32:18.193999+00:00 | failed | acd1086c2a12e5d5e429cf8370a4b9251fd7a63f720332c5fd0f0810a01de355 | 25,364 | null | from __future__ import annotations
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
try:
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
except ... | {
"id": "52715b70-102a-4f53-b1ca-21f95569dd37",
"created_at": "2026-08-12 20:32:18.193999+00:00",
"db_md5": "32af68b3294731833c4b04892e1c0788",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "c9736235-c8ce-4b58-ae54-e5a4e04304ad",
"tier": "easy",
"dataset_id": "e3",
"status": "failed",
... | null
|
52716871-c791-433e-9f48-ca0e9798bab6 | easy | DDanlov | 2026-08-23 21:55:08.004233+00:00 | succeeded | ee88ebf42a29560039e398c77269f9e96b747ad5768114be852e1f7c94e109e1 | 31,834 | null | """
Unified Hybrid Masked Deliberation Model (71.9M Parameters) - Tier 1 (Pure Skipless, No Cross-Attention)
Architecture:
- Scale: dim=2048, num_heads=32, d_ff=8192 (Exactly 71,889,920 parameters)
- Hybrid Prefix-Bidirectional + Causal Target Attention Mask
- Token Layout: [Prompt (L_p), 4 Pause Registers, Target Toke... | {
"id": "52716871-c791-433e-9f48-ca0e9798bab6",
"created_at": "2026-08-23 21:55:08.004233+00:00",
"db_md5": "37af542d9ae01078d60ad474651933ea",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "9015fa26-1642-420a-b13c-4e03f0ff629a",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.7187891795945927,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.727534557137254,
"example_count": 600,
"exac... |
52894c9c-3519-49ab-8c7a-0e8196d6a9c3 | easy | himalalps | 2026-08-07 01:50:21.310243+00:00 | succeeded | 824469f443d6c9544726877e974dd825803f877ff6ab7deb4db637dba7394185 | 67,143 | 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": "52894c9c-3519-49ab-8c7a-0e8196d6a9c3",
"created_at": "2026-08-07 01:50:21.310243+00:00",
"db_md5": "f84d2e0da8003c5fa6abd2b4c63fc11f",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "bee2e90a-1573-4e77-be83-73c0e5a71fdc",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.157029660603216,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000024835268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1596181392669678,
"example_count": 600... |
5289dda4-2fba-4329-a33d-622c466cc200 | easy | sapient-sapiens | 2026-08-09 15:16:21.881469+00:00 | succeeded | 585c3c8a6b66c902e2e67cdb001d76a2a315a1c3ff23dd1982c1ee8041062e9f | 14,829 | null | """Near-500M adaptive recurrence grown from the hosted-working baseline."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
... | {
"id": "5289dda4-2fba-4329-a33d-622c466cc200",
"created_at": "2026-08-09 15:16:21.881469+00:00",
"db_md5": "74d3c524f2af8cdbcf95da8d165be1e7",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "38a5c768-8f2e-4e01-9542-4c8fbd2002e7",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 2.2518350531698212,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.016666666666666666
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.354770547863263,
"example_count": 100... |
528b78e0-dabe-45d8-8134-c6a6349aa4f1 | easy | yashkant | 2026-08-10 19:13:20.096419+00:00 | succeeded | 3d92d2def7d1f6ee17e42414e37fa57bc7f4ef531df5aff545cba01adee937c0 | 30,238 | null | """Round1289 CFNN AdamW LR4.5e-3, EMA0.997 arm."""
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 Tensor, nn
D_... | {
"id": "528b78e0-dabe-45d8-8134-c6a6349aa4f1",
"created_at": "2026-08-10 19:13:20.096419+00:00",
"db_md5": "36f3b0fb4750e7a64beded9d04dd0f82",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "c808761f-18f4-4726-b499-3ac591a1c825",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 3.690921425819397,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.2570833470672369
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.7368998527526855,
"example_count": 300,
... |
5294e4b3-b442-48c9-a31e-db2a278e2e26 | easy | sapient-sapiens | 2026-08-09 21:33:44.480725+00:00 | succeeded | 5bb4e61f0c3687019fbff8d90d6b56ec8467719a08ba00c1fe0e69800328728e | 15,567 | null | """T-independent categorical recurrence with learned latent-depth selection."""
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,
Toke... | {
"id": "5294e4b3-b442-48c9-a31e-db2a278e2e26",
"created_at": "2026-08-09 21:33:44.480725+00:00",
"db_md5": "d5518602a2c831b1a0df7d2053fd26b5",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "ba579a90-12d5-4a31-9e76-b3af827a4704",
"tier": "easy",
"dataset_id": "e3",
"status": "su... | {
"score": {
"mean_loss": 2.1815608254237784,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0025
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.175546143603616,
"example_count": 800,
"e... |
529d2526-592c-49fe-9d69-55bf16a24e4d | easy | jordanrubin | 2026-08-12 07:01:27.289494+00:00 | succeeded | 69cdc2478a82d00d4b61b28be83f03737586f93b6673851041ede3009176d1a1 | 24,918 | 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": "529d2526-592c-49fe-9d69-55bf16a24e4d",
"created_at": "2026-08-12 07:01:27.289494+00:00",
"db_md5": "badf552b7656da07ac1aa0b25f969db5",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "e975df59-f30f-4dbb-bb35-814e198df2d9",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.2130781164980453,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666545594732
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2148168593511453,
"example_count": 60... |
52a33a27-5ae9-41f7-88c7-9e74990bde0e | easy | sapient-sapiens | 2026-08-30 17:09:08.012337+00:00 | succeeded | 423303a863fa25d6807a6cb2615f3f5ba3782ef38c30c59478a361f279b18487 | 16,146 | null | """Eligible direct-Horner Transformer with parameter-free A-commit RMS matching.
Each outer square has four fixed Horner positions. S resets at each position,
then one tied call commits latent S before a second tied call commits A. The
full model therefore makes 8*4*2=64 calls. Each accepted A write is RMS-matched
... | {
"id": "52a33a27-5ae9-41f7-88c7-9e74990bde0e",
"created_at": "2026-08-30 17:09:08.012337+00:00",
"db_md5": "ce0f3fd5403342ba198b6dd07048726f",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "90e8e4f8-0549-4bb9-b86e-faa1e43eb095",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.173187685771992,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009583333345750969
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1663678911234765,
"example_count": 600... |
52ac2e06-2647-4a46-b8eb-7a1360875b69 | easy | pepinu | 2026-08-21 06:36:35.121594+00:00 | succeeded | 8f2604323b0935ddcc7d49ce89cf4a60dda7735abe7afeae832b0240f7769930 | 40,010 | null | """Squaring mod N as a learned permutation on CRT components.
WHY THIS SHAPE. (Z_N, x) is a commutative monoid, so every finite-dimensional
linear representation splits into multiplicative characters and squaring is
diagonal there -- the component basis is the only one in which this operation is
shallow. Two consequen... | {
"id": "52ac2e06-2647-4a46-b8eb-7a1360875b69",
"created_at": "2026-08-21 06:36:35.121594+00:00",
"db_md5": "67bdec213264108b780b80da455f8b04",
"submitter": "pepinu",
"github_login": "pepinu",
"run_id": "67977e05-6c89-476b-b100-6d493242ff4b",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 0.014076166320592165,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.014087258838117123,
"example_count": 100,
... |
52b57d06-d718-4b55-9daa-684c6f002769 | easy | ehonig | 2026-08-20 09:55:36.595633+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": "52b57d06-d718-4b55-9daa-684c6f002769",
"created_at": "2026-08-20 09:55:36.595633+00:00",
"db_md5": "9d411a2759da4ff3a26b1218ebd741ef",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "9d80aa44-2e83-405c-a002-d4e0cf10c9c9",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.2198887868440367,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003333333386108279
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2355477809906006,
"example_count": 60... |
52ba158b-872f-4bac-a04b-f39aa6bcabcf | easy | hamzah2304 | 2026-08-15 12:18:34.196202+00:00 | succeeded | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "52ba158b-872f-4bac-a04b-f39aa6bcabcf",
"created_at": "2026-08-15 12:18:34.196202+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "hamzah2304",
"github_login": "hamzah2304",
"run_id": "f37a3473-de9e-4c43-8d2e-90bdafde70f0",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.996520519256592,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02666666731238365
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.943193197250366,
"example_count": 100,
... |
52c8c99c-dcc9-45ec-89a3-a8d024f089c2 | easy | alexkreimer | 2026-08-30 01:46:12.183379+00:00 | succeeded | edf9c5fd94bc87db0997257b9d3dcc192e6271200ef5b98e01e958d4ad13cc14 | 67,434 | null | """Recursive squaring model: weight-tied cell iterated T times, reading N
through cross-attention to a fixed memory instead of additive injection.
Starting point for One Layer Deeper: one small block (self-attention over
the evolving state, cross-attention into a fixed memory, then an MLP) is
applied repeatedly to a s... | {
"id": "52c8c99c-dcc9-45ec-89a3-a8d024f089c2",
"created_at": "2026-08-30 01:46:12.183379+00:00",
"db_md5": "8c63ca196d21cdaf760c8448bb956b3b",
"submitter": "Alex Kreimer",
"github_login": "alexkreimer",
"run_id": "49632e89-b4cf-4d26-9db3-4aa4b33bbb42",
"tier": "easy",
"dataset_id": "e3",
"status": "s... | {
"score": {
"mean_loss": 2.3780651995329283,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.409850808618246,
"example_count": 800,
"... |
52cb87f3-2a92-4191-be59-3e494d5185e0 | easy | ddoan | 2026-08-14 13:14:48.332576+00:00 | succeeded | 482252bad7cf229c20d69464a58fe5dd91ace568341874fa253d370b2444185b | 14,074 | null | """Pinned place codec with bilinear routing and a learned carry scan."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
BatchReuseContext,
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatc... | {
"id": "52cb87f3-2a92-4191-be59-3e494d5185e0",
"created_at": "2026-08-14 13:14:48.332576+00:00",
"db_md5": "b1fe85bfcd6d252a249e77929dbb95e5",
"submitter": "Doug Doan",
"github_login": "ddoan",
"run_id": "2031d962-041b-440f-83a0-02532ebdfe79",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.0219513177871704,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.028524160385132,
"example_count": 100,
"exac... |
52ce8e8d-ddca-4911-9826-dfc2365a22f1 | easy | ipsa-loquitur | 2026-08-26 04:52:09.019502+00:00 | succeeded | 7600fb2d86ed5f9aaa1edd8433cae9fd6e7393f18e98e7a014f2d43283105320 | 15,929 | null | """Standalone NetF-V1 submission candidate for One Layer Deeper.
The learned rule is shared across bit positions, modular-square iterations,
moduli, and tape widths. During training, exact double-and-add states provide
dense one-step supervision. During evaluation, the model receives no state
corrections and rolls out... | {
"id": "52ce8e8d-ddca-4911-9826-dfc2365a22f1",
"created_at": "2026-08-26 04:52:09.019502+00:00",
"db_md5": "de9f782a107774fb5633dceeebf1fc39",
"submitter": "ipsa-loquitur",
"github_login": "ipsa-loquitur",
"run_id": "0304e98d-b3c8-44c7-8d27-3fdf3b55bd12",
"tier": "easy",
"dataset_id": "e4",
"status":... | {
"score": {
"mean_loss": 9.481244961828757,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.5309375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 11.784753409082995,
"example_count": 1200,
... |
52d2ce6f-749d-4cf5-b3e2-e5086dda1b37 | easy | jamaliki | 2026-08-04 10:49:16.281548+00:00 | succeeded | d4ed5a8e37c4275ecabbf8c3ddd1808a9873d1f39fc13d1c8b82a72e32db3d77 | 13,230 | 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": "52d2ce6f-749d-4cf5-b3e2-e5086dda1b37",
"created_at": "2026-08-04 10:49:16.281548+00:00",
"db_md5": "71fd0a9c7df392c12a9119834baa340e",
"submitter": "Kiarash Jamali",
"github_login": "jamaliki",
"run_id": "22f4dad6-6341-43b7-b683-f3583607946a",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 3.0822832584381104,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05833333171904087
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.163506507873535,
"example_count": 100,... |
52d96873-d82f-4492-bfa9-74ec49cb3ca2 | easy | briheet | 2026-08-07 03:55:07.130351+00:00 | succeeded | a91f8a33d01beb6c229e3e854a41f3cc72b6d48fc81f5040ed0431bc4343c74b | 5,951 | null | """Transformer encoder with a small, recurrent learned latent state."""
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 = 64
... | {
"id": "52d96873-d82f-4492-bfa9-74ec49cb3ca2",
"created_at": "2026-08-07 03:55:07.130351+00:00",
"db_md5": "2d2a9c6e4de09d81cfefcd317f724db0",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "a68b327d-4b95-401c-b355-89e0569f1946",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.625446915626526,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.029999999329447746
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6964242458343506,
"example_count": 100... |
52e2a764-acf1-4fbe-976d-8e9302cb1011 | easy | chad-atexpedient | 2026-08-16 23:07:22.408176+00:00 | succeeded | 67d300f61de5d5566afb2f94a9a4047eeeb0ccbda2d5a718cd8f3cb352060553 | 21,858 | null | """X76: X75 with a HIGHER LR floor (0.5*peak = 6e-4) for stronger seen-N.
X75 preserved bilateral OOD-N at 8401 Medium steps but its seen-N T1 (2/768) was
below X63's (4/768) because parking at the very low 3e-4 floor limits how much
the seen-N signal sharpens late. X76 raises the floor to 6e-4: enough late-stage
LR t... | {
"id": "52e2a764-acf1-4fbe-976d-8e9302cb1011",
"created_at": "2026-08-16 23:07:22.408176+00:00",
"db_md5": "1e59990f983c1ec8056786173f42d617",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "66e644f1-d347-47cf-93fd-180fd4d71550",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.2072661609393363,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2050083549567936,
"example_count": 60... |
52e583a1-6f74-458b-a17e-f018b1d77c5a | easy | sapient-sapiens | 2026-08-31 02:24:11.384256+00:00 | succeeded | bd7950f97fa52581165518baf3057a6693fc6fea24ff26dc887f0a418ef3d092 | 24,288 | null | """Eligible E5 parallel categorical reducer.
Each Horner step is factored into four calls to two shared Transformers:
C1 = Form_theta(A, phase=10A)
A1 = Reduce_theta(C1, N)
C2 = Form_theta(A1, X, d, phase=A+dX)
A2 = Reduce_theta(C2, N)
The same formation block is tied across its two phases, every Hor... | {
"id": "52e583a1-6f74-458b-a17e-f018b1d77c5a",
"created_at": "2026-08-31 02:24:11.384256+00:00",
"db_md5": "759919827af407c1708ca4149008c29b",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "30cdfde6-5566-4246-ad45-1dbdfe9e8ef2",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.157518660330375,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1574542971469897,
"example_count": 600... |
52eb7611-5b7c-4b62-aa02-e5222bda8d5e | easy | lpbb | 2026-08-12 08:39:46.230721+00:00 | succeeded | 4a407c432eed68e969a60e8ebfbc1dbb325c3a6316d77b122dbd86982cbeebb3 | 24,885 | null | """DIGIT-STATE DISCRETE FLOW (v2) — diffusion-style quantized recurrence, e1.
v1 (submissions/agents/diffusion) put the recurrent state in a learned
codebook. Flaw found on paper: nothing ties step(enc(x)) to enc(f(x)), so
the code space can fragment into per-chain private paths and the coverage
argument (98-99% of e... | {
"id": "52eb7611-5b7c-4b62-aa02-e5222bda8d5e",
"created_at": "2026-08-12 08:39:46.230721+00:00",
"db_md5": "90967c4e8b202046b9d848e1acfdbcc9",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "d69279ad-6819-4fff-af05-c52766bfb7f9",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 3.8733608995305575,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.857557711402659,
"example_count": 800,
"exa... |
52ee4ca0-b516-4484-ab9d-53cdd1110835 | easy | yashkant | 2026-08-17 00:47:38.711348+00:00 | succeeded | d29f9668ee99bafa7f0ec9c57909e7242e513104db06c3c407bf41343bdf0ef6 | 52,170 | null | """Round2152 SAM4 active-Dykstra antisymmetric twin expansion.
The complete SAM-every-four Round2098 parent remains active. One byte-identical
copy of its compact recurrent Dykstra correction is appended without advancing
the parent's RNG stream. If the inherited branch is A and the clone is B, the
attachment become... | {
"id": "52ee4ca0-b516-4484-ab9d-53cdd1110835",
"created_at": "2026-08-17 00:47:38.711348+00:00",
"db_md5": "4d3be06210b135bae6e6446869e50507",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "375220f0-1c4a-4f51-abc3-6311c278262a",
"tier": "easy",
"dataset_id": "e4",
"status": "succeed... | {
"score": {
"mean_loss": 2.0708681117155763,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014791666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0728513412827216,
"example_count": 12... |
52f46821-12d9-42ab-95f9-d7f2f340c262 | easy | viridale | 2026-08-25 01:23:19.823316+00:00 | succeeded | 26ef4f100c72d4b8b09828f9805687b5545e84b5024b09f2a26a023c5110e4c1 | 23,507 | null | """Constraint-posterior bilinear Transformer program.
hypothesis: H1 and no-T1 Mediums withhold primitive rows, so direct congruence
evidence is silent; marginalizing one learned bilinear program against several
supplied deep endpoints identifies the observable semigroup action without a
functional-root gr... | {
"id": "52f46821-12d9-42ab-95f9-d7f2f340c262",
"created_at": "2026-08-25 01:23:19.823316+00:00",
"db_md5": "a6fd6ad643166bbe2b02dc3d4d0f99c9",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "9138e816-f906-4f4b-a68b-1556d245a189",
"tier": "easy",
"dataset_id": "e6",
"status": "succe... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 60,
"exact_accuracy": 1.0,
"c... |
52fe0ea6-8aa4-4fb5-9927-924c4b6278a5 | easy | yashkant | 2026-08-19 01:19:43.218862+00:00 | succeeded | 821067c8c528511c0db7e33e2118be1a26584442d85858058244d0f901aa97a1 | 41,440 | null | """Round2190 removes cross-step Dykstra correction carry.
The retained official carrier, primal recurrence, two learned projections,
antisymmetric set codes, parsed-T allocation, optimizer, and EMA are fixed.
The candidate reuses the learned initial A/B correction offsets on every
cycle instead of carrying hand-writte... | {
"id": "52fe0ea6-8aa4-4fb5-9927-924c4b6278a5",
"created_at": "2026-08-19 01:19:43.218862+00:00",
"db_md5": "232687cd117a8d4468822e1ff2405cc0",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "bcd98152-5814-4d6d-8edd-32497a8a9c22",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 4.103299260139465,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04833333380520344
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.372521877288818,
"example_count": 100,
... |
52ffedf3-cbb7-45d3-bc00-203b6e22892f | easy | oupadhyay | 2026-08-08 11:19:42.580953+00:00 | succeeded | f7cd37d605a00a2cc099e1e3e3648c02abb6cbd2f813f77f869c6dd936f59c9b | 9,526 | null | """Tied equilibrium transition over categorical, LSD-relative digit places."""
from __future__ import annotations
import math
import torch
from torch import Tensor, nn
import torch.nn.functional as F
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
PAD, N_MARK, X_MARK, ... | {
"id": "52ffedf3-cbb7-45d3-bc00-203b6e22892f",
"created_at": "2026-08-08 11:19:42.580953+00:00",
"db_md5": "5922bac821bd636cac3079655cab6047",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "dada690a-9133-47af-8e3b-c754eb0634b7",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.208406872048794,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0020833333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1747301929200176,
"example_count": 60... |
5301f9e5-eb5b-418d-9617-69136bf4360d | easy | shreyash-chonkie | 2026-09-01 01:18:04.512018+00:00 | succeeded | 588d9a3597d30ba029f0914d3db3782ea91feb3b3dfafc2babf129666fba1769 | 9,876 | null | """Diagnostic Koopman recurrence with explicit numeric training wheels."""
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,
a... | {
"id": "5301f9e5-eb5b-418d-9617-69136bf4360d",
"created_at": "2026-09-01 01:18:04.512018+00:00",
"db_md5": "8d1903cf608f8323d3582e0ebb5bf61f",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "2ca85892-17a4-46c3-a8c5-f489913ecd40",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 824.626953125,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666716337204
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1152.6085205078125,
"example_count": 100,
... |
530ba541-84dc-4af9-9b1c-65841f426570 | easy | AFeinfield | 2026-08-27 23:23:56.233229+00:00 | succeeded | 44a2cf9c73e2c768dd1bc0c1462cc3283e45510c40297c34615f4e57d148e31a | 13,865 | null | """165: 158 raw features + 161 state clamp (valve + hygiene)
159/160 failed hosted with NaN: bounded features removed the loss-side
pressure valve on place magnitudes, so iterated squaring inflated the
state to inf and sin(inf)=NaN. Clamping the state after carry
normalization keeps every intermediate finite regardles... | {
"id": "530ba541-84dc-4af9-9b1c-65841f426570",
"created_at": "2026-08-27 23:23:56.233229+00:00",
"db_md5": "8580e086a610befdc9f30f53263bb498",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "06365f4e-9d42-43f8-8894-c31afc69d39c",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.2667788741040464,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013750000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.255574255246218,
"example_count": 600... |
530ea903-df03-43bc-b9fb-bffb3d6b0ce7 | easy | DESU-CLUB | 2026-08-24 10:39:30.601781+00:00 | succeeded | 3f646963ef3544414e8626decab1bfae501b81261ea007582b5bd7b151a01678 | 10,579 | null | """Huginn-style looped Transformer whose iteration count is read from the T token.
prelude -> [core applied T times, input re-injected each step] -> coda
Forked from experiments/_baselines/adamw_transformer.py; the Block, RMSNorm and
optimizer are unchanged so that the only varied axis is the depth mechanism.
"""
fr... | {
"id": "530ea903-df03-43bc-b9fb-bffb3d6b0ce7",
"created_at": "2026-08-24 10:39:30.601781+00:00",
"db_md5": "c8c826c67d1e8e59c4339d0306ea9f17",
"submitter": "Low Keng Hoong, Warren",
"github_login": "DESU-CLUB",
"run_id": "82a320b9-d900-44ac-a51d-d2c21cc5c4c0",
"tier": "easy",
"dataset_id": "e6",
"sta... | {
"score": {
"mean_loss": 8.165676273711739,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.10202703005760103
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.092413902282715,
"example_count": 60,
... |
53147ecb-9c4d-40e9-bd48-4fc1217eab4f | easy | rougee | 2026-08-25 07:21:43.327644+00:00 | succeeded | 988084523252a05983822bfe403fab3473b4502aaac1c695e7b835e0ff86673c | 13,778 | null | """Official Submittable Champion for One Layer Deeper (Contextual Prelude + Stacked Quadratic Core).
State-of-the-Art Recurrent Architecture:
- Prelude: Causal RoPE Multi-Head Self-Attention Contextual Encoder
- Metric Geometry: Dynamic Skew-Symmetric Lie Group Matrix Exponential W_orth = exp(S - S^T) from Causal Prom... | {
"id": "53147ecb-9c4d-40e9-bd48-4fc1217eab4f",
"created_at": "2026-08-25 07:21:43.327644+00:00",
"db_md5": "21d683c6e5d833061453107ddf7397fe",
"submitter": "Rougee",
"github_login": "rougee",
"run_id": "4952509a-d505-4c53-9100-6baca88445ec",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 21.276240063427096,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.029999999552965166
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 20.60733413696289,
"example_count": 100... |
53152d54-7211-4bb8-a175-963e947bd595 | easy | shreyash-chonkie | 2026-08-26 12:51:18.663169+00:00 | succeeded | 0ffa41bae3794c6c2cfed411b972fd8a722ac40d3c09eece9d83a233a19f8286 | 12,260 | null | """A bounded-local compiler driving a generic learned stack executor."""
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,
ass... | {
"id": "53152d54-7211-4bb8-a175-963e947bd595",
"created_at": "2026-08-26 12:51:18.663169+00:00",
"db_md5": "46b828b9251130c2f5e9b4bec6f08159",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "fb5d26d7-2f72-4b55-aa08-992bb37474de",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 3.48241126537323,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.510441541671753,
"example_count": 100,
"exact_... |
5321c438-4331-41a2-8912-6f7db0007674 | easy | alirezashirvani-jr | 2026-09-01 00:42:35.189709+00:00 | succeeded | 636490a89471fd04c6d9dad2e3f05b1e061da391f2509373e45f1b2a5bcb8893 | 18,014 | null | from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, TokenLossBatch, assert_model_state
PAD = 0
MODULUS = 2
VALUE = 3
DEPTH = 4
ANSWER = 5
EOS = 6
DIGIT0 = 7
DIGIT_COUNT ... | {
"id": "5321c438-4331-41a2-8912-6f7db0007674",
"created_at": "2026-09-01 00:42:35.189709+00:00",
"db_md5": "7fc31f691c61e5284cabecb8709b3bcd",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "085e5789-563f-442a-a7e3-91776b118d6b",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 22.56314671613884,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002083333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 22.080081058724577,
"example_count": 600... |
532c8523-8036-49fc-8015-4b5dbb9069cf | easy | c-nnr | 2026-08-04 11:17:50.120486+00:00 | succeeded | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "532c8523-8036-49fc-8015-4b5dbb9069cf",
"created_at": "2026-08-04 11:17:50.120486+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "c-nnr",
"github_login": "c-nnr",
"run_id": "36a02241-d442-4b4a-9c5f-fef44b3d486d",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 3.014359712600708,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.94218373298645,
"example_count": 100,
... |
5330c41f-ecf4-4dfa-ad3c-4beeff9031f8 | easy | alirezashirvani-jr | 2026-08-31 23:11:16.420473+00:00 | succeeded | 833072d99b482a874a60ee8cfe1cca4b839a6255660346ef90e5069b04df0f06 | 25,284 | null | from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
HP = {
"width": 0,
"maximum_outer_steps": 64,... | {
"id": "5330c41f-ecf4-4dfa-ad3c-4beeff9031f8",
"created_at": "2026-08-31 23:11:16.420473+00:00",
"db_md5": "419d665ac4de16d355ed6fa3232bf1da",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "fe2f3ca2-560e-4396-81fc-6c8ac02e8296",
"tier": "easy",
"dataset_id": "e1",
... | {
"score": {
"mean_loss": 2.0831801891326904,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0621159076690674,
"example_count": 10... |
5337f7a8-4d17-41cf-a9fa-8b58664fccc2 | easy | AFeinfield | 2026-08-18 23:41:08.927928+00:00 | succeeded | dd6fa161e518fbc52f4ee06609ebf626a2e6547dc0d13279f3c9a2df6e87d74f | 13,788 | null | """083: conv kernel 5 -> 9 in the multiply sublayer.
Direct test of the mechanism story: the conv gathers neighbouring digits
so the bilinear form can multiply them, so a wider kernel should let more
digit PAIRS interact per iteration.
Original 073 notes follow.
073: BILINEAR multiply + K=5 agreement heads.
The multi... | {
"id": "5337f7a8-4d17-41cf-a9fa-8b58664fccc2",
"created_at": "2026-08-18 23:41:08.927928+00:00",
"db_md5": "7f429cd8b404c5ed7219db333f350708",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "da2d9733-c209-4d65-8176-e97a2bd9c61d",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 16.641676902770996,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012500000651925802
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 17.18892478942871,
"example_count": 300... |
533ad4bb-a5bf-4939-94b8-a9bae23bf697 | easy | khushidahi | 2026-08-05 20:29:53.096471+00:00 | succeeded | 73d4cede9dbc184058c67227389ee6579308a4674db84a3c48a5d1793642cf4f | 16,416 | 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": "533ad4bb-a5bf-4939-94b8-a9bae23bf697",
"created_at": "2026-08-05 20:29:53.096471+00:00",
"db_md5": "510c56bdfd21bc99e814c0a1de01e2d6",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "72219512-f787-475f-92b5-ac83d58412c9",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.2180862426757812,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166667005047202
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2157392501831055,
"example_count": 60... |
53460b1c-9713-43b2-8c57-811124b08475 | easy | viridale | 2026-08-04 22:19:04.187690+00:00 | succeeded | 8875faef1f48efa4f377699f4fb161a34110c0b5553a6bb4a5f3d3620ad1dd9a | 7,865 | null | """Clean autonomous token-native tied recurrence.
hypothesis: v166's distinct random step embedding at every refinement weakens the
shared-algorithm bottleneck and injects fresh noise through all 16 applications;
removing only that embedding makes one learned transition genuinely autonomous.
axis: arch
target:... | {
"id": "53460b1c-9713-43b2-8c57-811124b08475",
"created_at": "2026-08-04 22:19:04.187690+00:00",
"db_md5": "6cd710c9cb8b0ebf8bd391e4adfe93a8",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "2d14c503-d212-47ea-920d-930d8987d901",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 2.5604886267928375,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.015
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6545177935350326,
"example_count": 800,
"e... |
53464ae7-6bc7-4947-b923-84ca8a3a7008 | easy | DESU-CLUB | 2026-08-30 19:49:14.990320+00:00 | succeeded | 512145c1f969ba1a12f4f69ba1d8af04597239399c6295ad343d36f5c752aa3b | 16,039 | null | """Discrete-state loop over a place-value tape, trained with orthogonalize-only Muon.
The recurrent state carried between squarings is the *digit string*, not a
residual stream. Each outer iteration reads (N, x) off an LSB-first place-value
tape, emits digit logits for one squaring, hard-argmaxes them (straight-throu... | {
"id": "53464ae7-6bc7-4947-b923-84ca8a3a7008",
"created_at": "2026-08-30 19:49:14.990320+00:00",
"db_md5": "fe24752b5b6d14ae8fc4176d0aeb273a",
"submitter": "Low Keng Hoong, Warren",
"github_login": "DESU-CLUB",
"run_id": "14370923-0fbc-4070-a434-6e45f4447083",
"tier": "easy",
"dataset_id": "e6",
"sta... | {
"score": {
"mean_loss": 1.0996450108208067,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6725225384170945
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.8672475218772888,
"example_count": 60,
... |
534e06c7-016d-4b8e-9b9c-5d4f1cd42663 | easy | erdavis0 | 2026-08-27 02:17:41.583154+00:00 | succeeded | ca6f1b5c93401e93c705f4189e451ef833a9103266142696e3b482501ff8d97a | 31,305 | null | """Typed-prefix reversible transition B1: CE plus selector-oracle KL.
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": "534e06c7-016d-4b8e-9b9c-5d4f1cd42663",
"created_at": "2026-08-27 02:17:41.583154+00:00",
"db_md5": "3f9846560514c463a6ae282585eac958",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "3d36bbb9-d3c0-4b69-9bd0-e027532fbd31",
"tier": "easy",
"dataset_id": "e9",
"status": "succeeded",... | {
"score": {
"mean_loss": 7.156833648681641,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08777778036892414
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.643091201782227,
"example_count": 90,
... |
53578050-1c40-495f-af3f-7fd9b601793b | easy | DDanlov | 2026-08-08 09:53:16.285363+00:00 | failed | 98969dee575a7e37830e2ef134b47a045a9d38bd79048d685d71a1ba8f949e38 | 8,530 | 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
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class ModelConfig:
def __init__(self, vocab_size: int, max_seq_len: int):
self.vocab_size = vocab_size
self.max_seq_le... | {
"id": "53578050-1c40-495f-af3f-7fd9b601793b",
"created_at": "2026-08-08 09:53:16.285363+00:00",
"db_md5": "4d4c7b6480bfc30db0c1f60903fdc6ff",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "175851e6-478e-48df-957b-4e1b7bd0cb59",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
5361df38-e882-4649-ae2e-3bb3cb37a06b | easy | blake-camp-surge | 2026-08-05 17:58:33.481010+00:00 | succeeded | 5eaa9ca09e2858a94051093f64351d6af6b9c8a4beb9d5e294f4f5bcbd32cbc2 | 18,090 | null | """Width-16 linear-memory DAN with differentiable scratch nodes."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_m... | {
"id": "5361df38-e882-4649-ae2e-3bb3cb37a06b",
"created_at": "2026-08-05 17:58:33.481010+00:00",
"db_md5": "256bc3ec2a6070d7f9f1e228fd1f82d6",
"submitter": "Blake Camp",
"github_login": "blake-camp-surge",
"run_id": "82b336a9-c13e-459e-a6ce-9d72dd8a5793",
"tier": "easy",
"dataset_id": "e3",
"status":... | {
"score": {
"mean_loss": 2.1275426934811,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013750000121071934
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1262525738815223,
"example_count": 800,
... |
5371a73c-5bf6-475c-9f82-00edcb6364a7 | easy | blake-camp-surge | 2026-08-03 03:20:13.610022+00:00 | succeeded | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "5371a73c-5bf6-475c-9f82-00edcb6364a7",
"created_at": "2026-08-03 03:20:13.610022+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Blake Camp",
"github_login": "blake-camp-surge",
"run_id": "0990819b-d727-4ddd-8110-4bf20f4d6aef",
"tier": "easy",
"dataset_id": "e3",
"status":... | {
"score": {
"mean_loss": 3.9928400392458423,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000009313226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.297301549594544,
"example_count": 800... |
537c1d66-2f59-4ecd-aeb6-fbb0bfcaf8cf | easy | viridale | 2026-08-13 23:39:25.688022+00:00 | succeeded | 26b7fac33a7da881229059b441682ee592208c0235eac0b58f68f6eb361abf75 | 20,854 | null | """Prompt-amortized conditional diffusion over the complete answer.
hypothesis: prompt-only endpoint rollouts begin outside the exact answer basin; one
prompt-forward complete reverse table lets supplied final strings train the full
local correction field without a label-conditioned learned call.
axis: arch/lo... | {
"id": "537c1d66-2f59-4ecd-aeb6-fbb0bfcaf8cf",
"created_at": "2026-08-13 23:39:25.688022+00:00",
"db_md5": "b0e73c473fb93420296c8658fad0150c",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "f189a344-a98b-47b4-ba71-cca8584c3b81",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 16.709707260131836,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01833333307877183
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 16.512603759765625,
"example_count": 100... |
53834170-f35c-47b0-9540-c628cd3953e5 | easy | viridale | 2026-08-06 04:51:33.829195+00:00 | succeeded | c41c312abc57556d5a51826461ed92cdba8e8006af049fc0b060c461aca05c2a | 13,097 | null | """Clean bounded same-stage gauge-conjugated canonical recurrence.
hypothesis: one shared nonlinear cell may implement reusable computation only after
each opaque prompt/stage is transported into a learned bounded local coordinate
chart; same-stage conjugacy makes every chart causal and non-canceling.
axis: ar... | {
"id": "53834170-f35c-47b0-9540-c628cd3953e5",
"created_at": "2026-08-06 04:51:33.829195+00:00",
"db_md5": "86cd2edf5569445426ba558e7145d5cf",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "39dd3ed4-872a-4eba-87e2-bdf0a6fc9c1e",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 4.238504648208618,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03166666720062494
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.270853042602539,
"example_count": 100,
... |
538c0870-143d-467d-9dde-f6960af1fb9a | easy | alirezashirvani-jr | 2026-09-01 00:55:05.567712+00:00 | succeeded | 98e07da4d2c504b7f3b20c4e9add7a4fd0986082b694299d7e68304c50d87089 | 33,618 | null | from __future__ import annotations
import math
import time
from contextlib import nullcontext
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
MLP_EX... | {
"id": "538c0870-143d-467d-9dde-f6960af1fb9a",
"created_at": "2026-09-01 00:55:05.567712+00:00",
"db_md5": "8ffbc3f7a281baf3daa3404aeeb77368",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "55ddc383-3f0d-4f42-921a-9c36032261b6",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 3.0754146491353573,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03416666672875484
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.101840019226074,
"example_count": 600,... |
538dd6e4-8e08-40b8-9627-97563273a9ee | easy | k-penchev | 2026-08-22 23:41:28.819740+00:00 | succeeded | f919c01ef6da3a7f63571a58716fb19ee21d147d682354315afc48b2008089f9 | 9,580 | null | """Schedule-Free gated recurrence with T-conditioned evaluation depth."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
... | {
"id": "538dd6e4-8e08-40b8-9627-97563273a9ee",
"created_at": "2026-08-22 23:41:28.819740+00:00",
"db_md5": "df829fabf564e8ab0083ee383173f5db",
"submitter": "Kaloyan Penchev",
"github_login": "k-penchev",
"run_id": "ff0f2c6c-96e3-4c23-bc4c-38088fc68be8",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 2.229616165161133,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03500000108033419
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.388570785522461,
"example_count": 100,
... |
539234bb-6f9d-4d74-bf8b-84492e1019c0 | easy | DDanlov | 2026-08-22 14:12:24.791846+00:00 | succeeded | f2979540ccfca3ba3cdb205cc6431cbfc54f3dcd1e7730cf39444a60c6eb1cf0 | 28,998 | null | """
Unified Prompt-Only Deliberation Model (85.3M Parameters) - Easy Tier (e5)
Strict Prompt-Only Deliberation: Deliberation engine processes ONLY prompt tokens + 4 pause registers.
Answer digits are predicted via causal readout cross-attending to the unrolled thought manifold.
"""
from __future__ import annotations
im... | {
"id": "539234bb-6f9d-4d74-bf8b-84492e1019c0",
"created_at": "2026-08-22 14:12:24.791846+00:00",
"db_md5": "67e23ae9dcc0b361bf16f37b1a3f9d7f",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "730c302b-882e-4874-ba36-d53976759870",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 13.891209555836257,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0062500000372529025
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 14.381457717963933,
"example_count": 6... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.