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 |
|---|---|---|---|---|---|---|---|---|---|---|
53948952-606b-42f3-9e5b-5c0b96e1d020 | easy | liam-gb | 2026-08-16 19:51:56.322197+00:00 | succeeded | 73377d899037f347ce9872a62e038450eb59f5c03a550d9d438f7ba6dc0e4190 | 18,288 | null | """rns_pm_gather_ct_widet_l32nore_muon: single change vs rns_pm_gather_ct_widet_l32nore —
the optimizer. Dense 2-D matrices step with Muon (momentum + Newton-Schulz
orthogonalisation, the standard practice split); lookup tables and all 1-D /
scalar parameters keep AdamW exactly as the base ships them (tables lr 3e-2,
n... | {
"id": "53948952-606b-42f3-9e5b-5c0b96e1d020",
"created_at": "2026-08-16 19:51:56.322197+00:00",
"db_md5": "8e64703b1ca5e11db9484c1cff485ce2",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "c351ee8a-1e70-476d-8036-37d917c11f31",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.802015873091715,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05625000010554989
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.610459327697754,
"example_count": 600,
... |
53966f7d-9216-4f55-af85-0210b2718baa | easy | EyimofeA | 2026-08-09 21:39:47.454309+00:00 | succeeded | d7200ea4d2d53af87232083cecf53c4052f246fcdb8ea89928bee7209d9ef51d | 9,071 | null | """Fable Hard arch + train-time T cap + AdamW (timeout-safe).
Merged from `fable_hard_h1_muon` architecture with two changes aimed at the
Hard wall-clock timeout (job 5b363135…, MAX_LOOPS=64 unroll + Muon NS + no
eval_batch_size):
1. **T handling:** parse T from the prompt (same derived features). During
`train()`... | {
"id": "53966f7d-9216-4f55-af85-0210b2718baa",
"created_at": "2026-08-09 21:39:47.454309+00:00",
"db_md5": "df318dd463e79dea4d933da4445d6cad",
"submitter": "mof",
"github_login": "EyimofeA",
"run_id": "5155edb5-1b32-4263-92de-5c670ce2d030",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.227804660797119,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012083333684131503
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2237331867218018,
"example_count": 300... |
5397e3c3-c322-42d2-b861-6e282dc7badc | easy | erdavis0 | 2026-08-23 15:49:48.424806+00:00 | succeeded | a088f356590cfba5ad21d2d47f6b52ef6ea3fd6c700f05b01303b3ce92032250 | 23,907 | null | """Reversible augmented-state transducer over aligned learned digit tapes.
Public delimiter syntax only routes opaque token categories into distinct,
right-aligned N, X, and T tapes. Learned local cells compile separate
interaction, carry, and modulus-conditioned workspaces. N is isolated from
interaction and carry ... | {
"id": "5397e3c3-c322-42d2-b861-6e282dc7badc",
"created_at": "2026-08-23 15:49:48.424806+00:00",
"db_md5": "3e958839de80efbf59a22b9e0b2a18fd",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "570905cc-6b4d-46a1-a2c0-d05f601b1ddc",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 6.9913530349731445,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06999999843537807
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.553481101989746,
"example_count": 100,... |
53a058e6-5fe5-4180-a676-8bb5411995d6 | easy | ehonig | 2026-08-19 13:28:13.619882+00:00 | succeeded | f1151cf7fb93a871711b7d29f5a64d0f28eb0a2817c1f9323f9fd01d1be1176c | 15,765 | null | """A plain Transformer for repeated modular squaring, and nothing else yet.
This is a deliberate restart. The previous submission accumulated a field
parser, a carry scan, a learned reciprocal, periodic features, a digit
bottleneck, a depth selector and two cell types, most of them tuned around a
memorisation table th... | {
"id": "53a058e6-5fe5-4180-a676-8bb5411995d6",
"created_at": "2026-08-19 13:28:13.619882+00:00",
"db_md5": "5a8b22e6d9eb37422b59a5a381a7467c",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "1f9470bc-8b43-4558-a101-b3d96bb6adb1",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 3.237252116203308,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006249999860301614
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.300570011138916,
"example_count": 800,... |
53a67fd4-e906-4626-835c-bbc0d4f4161d | easy | ddoan | 2026-08-14 11:05:49.224923+00:00 | succeeded | 4b5e9d4891d24ab7ec629c0b1414e860f449590b277456fedae81d697d039ccd | 10,413 | null | """Place-aligned active memory with hard-gated diagonal CGRU updates."""
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": "53a67fd4-e906-4626-835c-bbc0d4f4161d",
"created_at": "2026-08-14 11:05:49.224923+00:00",
"db_md5": "cf77a5c3ed3605f339d3ff420a9a3eae",
"submitter": "Doug Doan",
"github_login": "ddoan",
"run_id": "ccfd9992-df1a-4b43-ae5f-b614f470dfa4",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.1469511625193705,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000009313226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1467866442210544,
"example_count": 80... |
53a90f83-35cd-48a6-99cd-38d60429b609 | easy | christopherhwood | 2026-08-06 17:12:49.286708+00:00 | succeeded | 08f270e142bf9ae677ad2248daf54b5fe01661bb2d27068d2eb63842079e396f | 19,025 | null | """Canonical categorical cellular program for a hidden recurrence.
The model turns the positive endpoint-identification argument into an explicit
architecture. The state that survives every local update contains only a
distribution over the ten observable decimal digits and a small categorical
controller. One transla... | {
"id": "53a90f83-35cd-48a6-99cd-38d60429b609",
"created_at": "2026-08-06 17:12:49.286708+00:00",
"db_md5": "8462f764a72015c08a67183188894db9",
"submitter": "Chris Wood",
"github_login": "christopherhwood",
"run_id": "3c4451f6-815c-4a30-b9e2-fb70d060e94f",
"tier": "easy",
"dataset_id": "e5",
"status":... | {
"score": {
"mean_loss": 2.1707518700341253,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.165414011531881,
"example_count": 600... |
53aae23f-4c85-45e7-853d-4b6e305cf73a | easy | yashkant | 2026-08-10 20:15:46.124150+00:00 | succeeded | 102d896087066490f6ebb4f18324d4d7ca2ac60bc688db06d565ff2e52b46913 | 31,137 | null | """Round1282 shared continued-fraction denominator-recursion pair.
The complete Round1141 D416 MemGLU carrier remains byte-identical. Its pooled
representation seeds a positive terminal denominator. Every shared call adds
one tied continued-fraction level, ``P(x) + softplus(beta)/(state + epsilon)``.
The matched contr... | {
"id": "53aae23f-4c85-45e7-853d-4b6e305cf73a",
"created_at": "2026-08-10 20:15:46.124150+00:00",
"db_md5": "df57b8b95b0a939244b6cb428b1315f0",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "ef7f0754-a427-42ad-ac3c-9f555cd1b147",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 4.589303970336914,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06999999657273293
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.152830123901367,
"example_count": 100,
... |
53b9aa16-e72f-4e14-9832-4cda66453fa8 | easy | repst | 2026-08-22 20:21:43.794634+00:00 | failed | 238fc2ad39a9d0eeff878a0c6b35026b6c2bdb8ac9ea5c08566fda8f10dbfddd | 14,055 | null | # R515: unified EOS autoregressive semantic microcode grammar.
#
# LEARNED: anonymous numeric-field route, first instruction, shared conditional
# next-instruction-or-EOS grammar, and digit-token vocabulary. All begin
# randomly and receive only evaluator-owned output-token gradients.
#
# AUTHORED: decimal span tokeni... | {
"id": "53b9aa16-e72f-4e14-9832-4cda66453fa8",
"created_at": "2026-08-22 20:21:43.794634+00:00",
"db_md5": "944ddd22274abc4bbcf94b4beca7438c",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "eba9cf45-2b50-420b-96e8-98c318b1f907",
"tier": "easy",
"dataset_id": "e1",
"status": "failed"... | null
|
53bc392d-16e7-4a69-899d-47b3633e1b4c | easy | c-nnr | 2026-08-04 13:24:29.098580+00:00 | succeeded | a011fdd1780e6f769d14db2edb2126d49164b5e1d1113db368410ad88b332639 | 6,459 | null | """One Layer Deeper E1 final-15 experiment: Uncalibrated high-LR recurrence stopped near the update count of the original E1 TTC winner."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBund... | {
"id": "53bc392d-16e7-4a69-899d-47b3633e1b4c",
"created_at": "2026-08-04 13:24:29.098580+00:00",
"db_md5": "979e77135420b2f11682430ada3d5cda",
"submitter": "c-nnr",
"github_login": "c-nnr",
"run_id": "80805a19-8f61-4762-9c51-1a1071f8d19d",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.2645766735076904,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.322803020477295,
"example_count": 100,
... |
53c0000f-310f-4eab-bf42-68a558e0b79b | easy | 0Chris5R | 2026-08-01 07:55:30.575519+00:00 | succeeded | 5eb6a6ad6663ebf56f403eafc24d3b3b392215979e2910cbf9120747ccd0338b | 23,576 | null | """Shared-digit Neural GPU with late full-objective EMA evaluation.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the sam... | {
"id": "53c0000f-310f-4eab-bf42-68a558e0b79b",
"created_at": "2026-08-01 07:55:30.575519+00:00",
"db_md5": "d1ecdef9362744b111123e441811f4a8",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "e9363eee-68ee-4386-8e80-1ca3c552b409",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.5850431161625216,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.027083333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.401562899217477,
"example_count": 600... |
53c7e550-25ab-46e9-9ebe-9817ed82891b | easy | erdavis0 | 2026-08-19 07:53:40.093522+00:00 | succeeded | e6d75591b123483c27b5a4344aa991e325184d79f5ed87c2a9fa7d7390f5f5bd | 10,552 | null | """Generic learned-slot transducer with a shared low-rank latent operator."""
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,
TokenL... | {
"id": "53c7e550-25ab-46e9-9ebe-9817ed82891b",
"created_at": "2026-08-19 07:53:40.093522+00:00",
"db_md5": "fb39c040e0caf73a4ee6e4c2f9f4d564",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "9608f538-6bfc-43f7-b2b0-111d794d0c82",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 1.8356329202651978,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04833333194255829
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.7821825742721558,
"example_count": 100... |
53ca978b-d5cc-46a5-bf51-315a5cacdecb | easy | jordanrubin | 2026-08-11 01:51:52.084559+00:00 | succeeded | 6bf1f983b99635ca286c20e7c917ecfc29ba2a2cabec69ecb8c0e82b728767e2 | 21,669 | null | """Weight-tied MLP-cell loop over soft digit states (mlploop).
Cell = MLP over [expected digits, pairwise digit products, RNS residue
simplices (fixed differentiable mixing over Z_p), N digits]. The cell that
learns one-step modular squaring from direct pairs (52% unseen at 14k rows,
day-20 screen) inside the exact-T ... | {
"id": "53ca978b-d5cc-46a5-bf51-315a5cacdecb",
"created_at": "2026-08-11 01:51:52.084559+00:00",
"db_md5": "a2cac839e289a396335dd6a8ec69d716",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "04c58b0a-884e-4fed-ae18-d91831dc978f",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.2316492415605813,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333230887851
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1653873364487035,
"example_count": 60... |
53cc4d50-91bb-4461-b8e5-57f7b3579879 | easy | yashkant | 2026-08-10 20:55:57.221888+00:00 | succeeded | fe67b8451516bb96afdc2bb6573fdef7fdc0145be69a974a112843bb8c58261c | 19,615 | null | """Round 560 width768/rank64 AdamW LR3e-3 EMA0.997."""
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
... | {
"id": "53cc4d50-91bb-4461-b8e5-57f7b3579879",
"created_at": "2026-08-10 20:55:57.221888+00:00",
"db_md5": "490676b487eee44ecaab723ad278407a",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "054b1905-8d5e-46bf-85f3-78b17216a545",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 3.846782298852517,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.015625000055879357
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.8374806949958913,
"example_count": 800... |
53ce3c3e-2c87-4121-8f64-ed5c35e9f74e | easy | erdavis0 | 2026-08-27 02:34:14.682772+00:00 | succeeded | f696b4a0dbdab40d944b3bb47ef7b8e716ff6e0849099f9cf030611313221c72 | 31,312 | null | """Typed-prefix reversible transition B3: sequence mixture plus balance.
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": "53ce3c3e-2c87-4121-8f64-ed5c35e9f74e",
"created_at": "2026-08-27 02:34:14.682772+00:00",
"db_md5": "aa01dfb95948365c687b75f02f03cd86",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "8b2c32a5-bb4d-45f2-a8a6-3ae5fb86bdc3",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 9.160211086273193,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.071666669100523
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 10.598641395568848,
"example_count": 100,
... |
53d04548-38dc-4504-94b4-05955305b9c8 | easy | sapient-sapiens | 2026-08-14 18:18:10.284588+00:00 | succeeded | 28ba469aae3e42ca679270105f219b6bd47433fa80f83dc154b3c3e55e44c7c3 | 14,299 | null | """Digit-space recurrent operator for repeated modular squaring.
Nothing here converts a token into a decimal number. Digits stay as embeddings,
so what a digit is worth, and every carry and reduction, is learned. That also
sidesteps a hard numeric wall: at Hard's modulus scale `x^2` runs past the 53-bit
exact-integer... | {
"id": "53d04548-38dc-4504-94b4-05955305b9c8",
"created_at": "2026-08-14 18:18:10.284588+00:00",
"db_md5": "1efadda9e01ae84ec2857432a666c542",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "c10c59c6-5b9f-4cbd-a688-a1043637d330",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 4.027323438045093,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08833333282421033
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.712823867797852,
"example_count": 100,
... |
53d2e119-4bfa-427a-a6b7-b5e78b0c9995 | easy | Yalyenea | 2026-08-11 17:28:45.139827+00:00 | succeeded | 4381420cfc6e7616d183b778888d0c81f5641f919d6481d2c69a3eb19101db3f | 7,018 | null | """Official causal-LM R4 sandwich loop with scratchpad tokens."""
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 = 512
NUM_H... | {
"id": "53d2e119-4bfa-427a-a6b7-b5e78b0c9995",
"created_at": "2026-08-11 17:28:45.139827+00:00",
"db_md5": "f4e267a24e62ce2d04423cdcf3ddee62",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "b1ecaa5d-706a-4dea-ad34-6ceaca097707",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 3.0905981063842773,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03666666615754366
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.754412889480591,
"example_count": 100,... |
53e75429-3a28-42f9-9155-712ca58d5621 | easy | khushidahi | 2026-08-05 23:08:45.122755+00:00 | succeeded | 78722a5d1924ebffb0fd3202d79bc36fd88e4b9c338b95facff9941a9873ceea | 25,710 | null | """R16 hard-transfer hybrid experiments for One Layer Deeper.
Motivated by complementary Hard behavior:
- the generic sequence recurrent model retained seen-N first-rung progress;
- the structured decimal-tape model retained OOD-N first-rung progress.
The dual variants co-train both learned representations, supervise... | {
"id": "53e75429-3a28-42f9-9155-712ca58d5621",
"created_at": "2026-08-05 23:08:45.122755+00:00",
"db_md5": "9ac462cd537ec2d817a2be39e296f285",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "4d032a8c-27f4-43d8-8854-fee8651d3531",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.0725883003686976,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0053124999999999995
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0703039804856815,
"example_count": 1... |
53ed30f4-0913-49a0-b7a0-0e576083053a | easy | isaac0804 | 2026-08-10 21:34:11.303329+00:00 | succeeded | cb39b32d05e6ede60f130d3b5e6ed7c53943ceab66936bd9040fac23c4cdc9ee | 10,276 | null | """Clean loop (no damped gate, no step-embedding) + a simplified xHC
(Expanded Hyper-Connections) variant: N=16 parallel residual streams,
mHC-style doubly-stochastic "keep" mixing across ALL 16 streams every
step, but new sublayer content is written back to only k=4 of the 16
streams (sparse write-back), plus a small ... | {
"id": "53ed30f4-0913-49a0-b7a0-0e576083053a",
"created_at": "2026-08-10 21:34:11.303329+00:00",
"db_md5": "0e431d74437457a24951b09441ae4ea9",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "17e47bfa-813a-4e4f-80d0-26430e5e9b6a",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 2.2105359260927298,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000009313226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.234215588744167,
"example_count": 800... |
53ed973d-9b6a-45b1-b331-e3cefe8290a1 | easy | Alan-Qin | 2026-08-26 15:06:46.432749+00:00 | failed | 850268f2eb903481d890f47540462e8b0f962c46dacdd82695ef3fbd33812677 | 24,977 | null | """v91h17 = v91 with LR-cosine horizon 1700 (Medium tiers are 600s hosted ~1.7k steps, NOTES #172; cosine must land inside the real budget).
v91 = v63lf + w-hat auxiliary head on the proven gated MLP (route B v2:
same readout, added CRT-quotient supervision; measures whether the MLP can
learn the universal quotient in... | {
"id": "53ed973d-9b6a-45b1-b331-e3cefe8290a1",
"created_at": "2026-08-26 15:06:46.432749+00:00",
"db_md5": "eb439a22cbe965b79c1780dd0ee6dc34",
"submitter": "Zeyu Qin",
"github_login": "Alan-Qin",
"run_id": "8efc8ccc-3b15-4d64-8cba-d1db859b9ce8",
"tier": "easy",
"dataset_id": "e6",
"status": "failed",... | null
|
53edf907-eeff-4929-8c5a-9f27f48bf3d8 | easy | christopherhwood | 2026-08-08 07:32:10.680784+00:00 | failed | 2b4af0d965012d4f21bb29e5bb6fc34287a1dbd00c5f6edcb5755f82234bcfd7 | 20,295 | null | """Standalone benchmark-native growth of straight-line recurrence programs.
The persistent state contains capacity for programs but no nonterminal program
at construction time. Optimizer steps retain endpoint-ranked parents, append
one instruction on GPU, and overwrite the remaining inactive/pruned rows.
"""
from __... | {
"id": "53edf907-eeff-4929-8c5a-9f27f48bf3d8",
"created_at": "2026-08-08 07:32:10.680784+00:00",
"db_md5": "4d7e6bf08cd5cfee92475fb65d96ab04",
"submitter": "Chris Wood",
"github_login": "christopherhwood",
"run_id": "274431be-1bc6-4937-a4fa-695f8acdd2d3",
"tier": "easy",
"dataset_id": "e1",
"status":... | null
|
53ef7889-9800-4b4a-a559-ae2eeedc05ae | easy | yashkant | 2026-08-12 04:39:14.206524+00:00 | succeeded | 83b038954234eaa72b2073621b6c8691a2329a752b69985ce1e29bb69444e804 | 32,160 | null | """Round1787 contrastive set-code centering on the W12/D1 carrier.
This evolutionary graft keeps the carrier, one shared projection per parsed
T, recurrent feedback, optimizer, LR, EMA, and parameter count fixed. It
removes the shared mean from the two learned Dykstra set codes and doubles the
remaining contrast, mak... | {
"id": "53ef7889-9800-4b4a-a559-ae2eeedc05ae",
"created_at": "2026-08-12 04:39:14.206524+00:00",
"db_md5": "54fdd172fa6f5c11c4831555174ca8dc",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "2ed2b6b2-1e38-44f7-83e5-435c6dbe69f8",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 4.435919880867004,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.3306250274181366
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.269965648651123,
"example_count": 300,
... |
53f131d1-9a2b-4839-b0ee-84864433c61d | easy | oupadhyay | 2026-08-07 12:43:51.636195+00:00 | succeeded | d9fe46d380639e44f497127eb4fcfbb43643dd68644e1e8f3155f85bad84b9cc | 8,697 | null | """N-conditioned factorized ordered digit-pair context with a tied reducer."""
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle, Submission, assert_model_state
PAD, N, X, T, ANS, DIGIT = 0, 2, 3, 4, 5, 7
D, PLACES, MICROPHASES, MAX_STEPS = 64, 4, 1, 64
STATE_ELEM... | {
"id": "53f131d1-9a2b-4839-b0ee-84864433c61d",
"created_at": "2026-08-07 12:43:51.636195+00:00",
"db_md5": "98a660c64e7ac2316a3d496ca06d2202",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "de47680c-f289-42bc-b8f9-5d75bffbab4a",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.1983713999204584,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2013249301055087,
"example_count": 60... |
53f7b5d9-c34e-4012-947c-4ed1ea46bca2 | easy | DDanlov | 2026-08-10 09:15:16.391994+00:00 | succeeded | 7204b1b334bc6180224075dd215b4c7bd621deca0940e9f4282a88a2bbe6a80b | 15,821 | 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 RMSNorm(nn.Module):
def __init__(self, dim: int, eps: float = 1e-6):
super()... | {
"id": "53f7b5d9-c34e-4012-947c-4ed1ea46bca2",
"created_at": "2026-08-10 09:15:16.391994+00:00",
"db_md5": "414382daff390477b79c9a4f9c11fe0e",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "07a69fdf-1e0f-43e8-bfb5-dfd37ae1e55b",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 31.899843018548445,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003333333432674408
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 41.57046890258789,
"example_count": 100... |
53f8d8d0-6917-4f19-a97f-52d2973633b0 | easy | DDanlov | 2026-08-13 22:40:16.229059+00:00 | failed | ed516e6e827a1f9c4e6504f548c38b93bd3430d645ded385c96a9ebec0321cd3 | 22,147 | null | import torch
import torch.nn as nn
import torch.nn.functional as F
import math
from dataclasses import dataclass
from typing import Optional
from torch.optim import Optimizer
try:
from client import Submission, assert_model_state, OptimizerBundle
except ImportError:
Submission = None
assert_model_state = N... | {
"id": "53f8d8d0-6917-4f19-a97f-52d2973633b0",
"created_at": "2026-08-13 22:40:16.229059+00:00",
"db_md5": "45bd94ea4d474d8771bffcc825a33b57",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "7af46e82-ba20-4b9f-8d83-f247a6986bc6",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
53f9ac7b-b730-4635-978d-5d6a328a3c9e | easy | jordanrubin | 2026-08-13 17:47:15.434556+00:00 | succeeded | e2245c74dfcfe7d00f8aeee61890a07c07a0d89977c245e3844890ce7f77b424 | 27,175 | null | """Weight-tied MLP-cell loop over soft digit states (mlploop).
Cell = MLP over [expected digits, pairwise digit products, RNS residue
simplices (fixed differentiable mixing over Z_p), N digits]. The cell that
learns one-step modular squaring from direct pairs (52% unseen at 14k rows,
day-20 screen) inside the exact-T ... | {
"id": "53f9ac7b-b730-4635-978d-5d6a328a3c9e",
"created_at": "2026-08-13 17:47:15.434556+00:00",
"db_md5": "3f710550312ea469d13b008d83980242",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "65c8fe9b-40d2-4da8-a1d3-70d5e0b94bcf",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.8197203691112236,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02666666591539979
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2129659159568393,
"example_count": 600... |
53fa7ccd-c362-4f12-9a00-42941be64057 | easy | liam-gb | 2026-08-16 23:11:09.028073+00:00 | succeeded | 24fc878b19987564e88eccf1d98937ae5735fa8ab620fdef470b9eacfec14e2f | 16,740 | null | """rns_pm_gather_ct_widet_l32nore_r8: single change vs
rns_pm_gather_ct_widet_l32nore — the batch-reuse policy returns True, so the
evaluator reuses each fetched batch for the maximum eight optimizer updates.
WHY. On the granular Easy suites one batch is 85-97% of the training split,
so every step ends an epoch and th... | {
"id": "53fa7ccd-c362-4f12-9a00-42941be64057",
"created_at": "2026-08-16 23:11:09.028073+00:00",
"db_md5": "46c5decfff7d97a850fb297981e9d310",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "24321849-08dc-4e5c-ab67-7501937de055",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.377949476242065,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.3843750227242708
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 10.753095626831055,
"example_count": 300,
... |
540643f3-b2ff-423b-95c5-5001ee2ab9d9 | easy | viridale | 2026-08-03 09:30:52.131689+00:00 | succeeded | 2ab7cba2961297bfd1d3fd25e786d3b53a173f72d2647e0b47600e40a5f5622b | 134,838 | null | """Prompt-depth I/O wrappers over the exact global source posterior.
hypothesis: the hidden source edit may use prompt T in initialization or the
returned observation, loci still absent after the T-conditioned clock and
transition closure despite being explicitly elevated by the API hint.
axis: arch
target: re... | {
"id": "540643f3-b2ff-423b-95c5-5001ee2ab9d9",
"created_at": "2026-08-03 09:30:52.131689+00:00",
"db_md5": "317a01daef78471923437720d63f2031",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "e4e80d96-0413-4cee-a4a2-84296490b555",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 0.003014546020381487,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0030145461205393076,
"example_count": 100,
... |
540e05c1-5e4e-4565-bad2-8c4057017186 | easy | jordanrubin | 2026-08-11 00:37:44.789383+00:00 | succeeded | 5627289ebca0e4853183d79604695f32d270f428a44ccf506384676ba86c6797 | 20,108 | 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": "540e05c1-5e4e-4565-bad2-8c4057017186",
"created_at": "2026-08-11 00:37:44.789383+00:00",
"db_md5": "bdc145351ed7c8a97c4c6df2f8d43449",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "1837d6c8-f8ae-4717-8c41-e7228b355447",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.7823984582758667,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006249999860301614
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1775721426501935,
"example_count": 60... |
540eb153-4839-4dec-adb9-3d1d0c7d13dd | easy | sapient-sapiens | 2026-08-14 17:28:12.801133+00:00 | succeeded | 28ba469aae3e42ca679270105f219b6bd47433fa80f83dc154b3c3e55e44c7c3 | 14,299 | null | """Digit-space recurrent operator for repeated modular squaring.
Nothing here converts a token into a decimal number. Digits stay as embeddings,
so what a digit is worth, and every carry and reduction, is learned. That also
sidesteps a hard numeric wall: at Hard's modulus scale `x^2` runs past the 53-bit
exact-integer... | {
"id": "540eb153-4839-4dec-adb9-3d1d0c7d13dd",
"created_at": "2026-08-14 17:28:12.801133+00:00",
"db_md5": "1efadda9e01ae84ec2857432a666c542",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "623db432-61e2-4161-8404-9498064027bf",
"tier": "easy",
"dataset_id": "e2",
"status": "su... | {
"score": {
"mean_loss": 3.0428291029179526,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000049670537
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.0167313704784187,
"example_count": 30... |
5410f7cd-35c0-4ec5-ac70-289ddee7e86b | easy | sapient-sapiens | 2026-08-31 02:11:59.778839+00:00 | succeeded | aa6d7b69fea87cdac919568609a900aaa3d1f1244e7908e0b0202b743cf8dfa8 | 26,449 | null | """Eligible full-E5 reducer-heavy probabilistic digit transducer.
All prompt digits, public A/X states, candidates, and outputs use one shared
ten-category decimal chart. Every square is four tied Horner steps. Every
Horner step applies one shared modulus-blind formation block twice and one
shared, reducer-heavy, tw... | {
"id": "5410f7cd-35c0-4ec5-ac70-289ddee7e86b",
"created_at": "2026-08-31 02:11:59.778839+00:00",
"db_md5": "4f6af82d1b68882259bc7af73b87d66b",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "b4b16aa8-6a28-4569-8326-86927e52d1aa",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.2027423303251203,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002083333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.207416863184873,
"example_count": 600... |
54154a95-0b13-41f2-8da3-0ac414a884bd | easy | sapient-sapiens | 2026-09-01 02:44:28.511079+00:00 | succeeded | b3033c13a09b4c0fc9f4682b70d334ba1b3d3b8c73fe07b1d412dee8cdda869d | 13,829 | null | """E5 one-shot square tied only across T with full outer BPTT."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
VARIAN... | {
"id": "54154a95-0b13-41f2-8da3-0ac414a884bd",
"created_at": "2026-09-01 02:44:28.511079+00:00",
"db_md5": "5be975bfd7d0e70f20c47bc852bd7f66",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "665aafc1-f212-4d93-b124-4d950065b99d",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1792981435738144,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0050000000248352685
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.18386229498504,
"example_count": 600... |
54184265-e998-4578-ae8e-4e889204f1b8 | easy | shirvani-jr | 2026-08-29 01:23:02.367485+00:00 | succeeded | d3ecbf2ad9f43cab9b2e3e569533f2bb6a8eb331bc3e79329bb147a9950b63b4 | 48,550 | 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": "54184265-e998-4578-ae8e-4e889204f1b8",
"created_at": "2026-08-29 01:23:02.367485+00:00",
"db_md5": "5de01dda85bf263313d5980c95494521",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "6cc31ed4-56d3-4f47-abb9-19a4344bc25c",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.8840450392113075,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.017916666896392902
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.9186904430389404,
"example_count": 60... |
541a5ba0-62a0-420e-8487-e66c94e5898a | easy | yashkant | 2026-08-09 02:44:41.585939+00:00 | succeeded | 9084788c57d70e9371c16efd7b28d9e381cd80dfe176a2de867c67d09ded40c6 | 22,836 | null | """Round 211 common TRM-style x/y/z shared-transition template."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
from torch import... | {
"id": "541a5ba0-62a0-420e-8487-e66c94e5898a",
"created_at": "2026-08-09 02:44:41.585939+00:00",
"db_md5": "3e558755dcaae2ce6d98f73727e41002",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "34a41a6b-e183-46aa-8345-7834bffd9b23",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.143457544447334,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.146639347076416,
"example_count": 600,... |
54285e68-fb60-4e05-8e7d-00f89ee292d3 | easy | jordanrubin | 2026-08-14 01:09:32.288646+00:00 | succeeded | fc98447fc19a656845f48580dd88f53c0b2d7d00c35e32a8fe2d3dfff79a73f4 | 29,581 | null | """Weight-tied MLP-cell loop over soft digit states (mlploop).
Cell = MLP over [expected digits, pairwise digit products, RNS residue
simplices (fixed differentiable mixing over Z_p), N digits]. The cell that
learns one-step modular squaring from direct pairs (52% unseen at 14k rows,
day-20 screen) inside the exact-T ... | {
"id": "54285e68-fb60-4e05-8e7d-00f89ee292d3",
"created_at": "2026-08-14 01:09:32.288646+00:00",
"db_md5": "d08ba312cd6460ffc449c5d2476626a5",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "ff1e250c-63ec-45f8-945f-30f5395bb0ca",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.853500750911687,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.020833333022892475
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2001122136019804,
"example_count": 600... |
542ebd20-d06f-4ec8-b3f1-6c64bb9d449b | easy | yashkant | 2026-08-11 22:26:24.781060+00:00 | succeeded | 5ee996ca12c0c91f90c26eafbe1369a68ecb2fe2b962c8228bc5963a2f02f67b | 34,238 | null | """Round1282 shared continued-fraction denominator-recursion pair.
The complete Round1141 D416 MemGLU carrier remains byte-identical. Its pooled
representation seeds a positive terminal denominator. Every shared call adds
one tied continued-fraction level, ``P(x) + softplus(beta)/(state + epsilon)``.
The matched contr... | {
"id": "542ebd20-d06f-4ec8-b3f1-6c64bb9d449b",
"created_at": "2026-08-11 22:26:24.781060+00:00",
"db_md5": "b9d712aa911475eaf7ccde95756d9fa4",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "5e3491fd-46e0-4967-b3f6-624a574e070d",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.149498187743264,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1550062301447572,
"example_count": 600,
"ex... |
543bb1ee-17f5-4de1-b184-c83cfb1ec344 | easy | erdavis0 | 2026-08-29 01:24:34.270611+00:00 | succeeded | c7f78ef5748d23ee6c570a69dda0f9e2f1d45cb5b22c279e1ef01c84faeb2f27 | 17,575 | null | """Sealed N-conditioned circulant operator over an X-initialized latent tape.
Public delimiters only route opaque categories into independent right-aligned
N, X, and T tapes. N alone compiles a small bank of learned cyclic place
kernels and a continuous program vector; X alone initializes the recurrent
state; T is co... | {
"id": "543bb1ee-17f5-4de1-b184-c83cfb1ec344",
"created_at": "2026-08-29 01:24:34.270611+00:00",
"db_md5": "c929d07e2f2fc815762284c89335e51b",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "c1f681e5-b6e9-4b9d-b448-e6ba9a0ed996",
"tier": "easy",
"dataset_id": "e9",
"status": "succeeded",... | {
"score": {
"mean_loss": 6.067757606506348,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07333333604037762
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.7995710372924805,
"example_count": 90,
... |
544d9a35-b87b-40f1-9b91-0c08b91636a8 | easy | DDanlov | 2026-08-19 19:55:56.589338+00:00 | succeeded | cd3a7e0a81fb5cba6b1a0fd74ecdbb9fa8f78d3b247f19b7ca56805aad05d6ca | 17,724 | null | """
Dynamic Submission for One Layer Deeper API Benchmark
Config: dim=1952, num_heads=32, d_ff_mult=4, fixed_trec=64
"""
from __future__ import annotations
import math
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
... | {
"id": "544d9a35-b87b-40f1-9b91-0c08b91636a8",
"created_at": "2026-08-19 19:55:56.589338+00:00",
"db_md5": "015ce2f5b55abe4bc9324b604adf5717",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "bec384bb-f58a-40d5-8202-bb5f77915812",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.3976504826828773,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0070833333457509685
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.471624405929326,
"example_count": 60... |
54556a38-ff67-46d1-9d20-a3e001cfb532 | easy | viridale | 2026-08-11 09:55:39.100604+00:00 | succeeded | 8b4ca0685454fd48ad2444b06445a18199bf4f6055ec40be99b0738206ac13cf | 43,039 | null | """Persistent-register Sidon transducer.
hypothesis: the frozen H2 gate proves that v507's within-step feature cannot
preserve a previous recurrent value. Add one learned delay-line register initialized
from the prompt, shift the old current value into it after every active step, and
expose all pair intera... | {
"id": "54556a38-ff67-46d1-9d20-a3e001cfb532",
"created_at": "2026-08-11 09:55:39.100604+00:00",
"db_md5": "0b2307e9e4731440df75e160a180c40f",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "8ddbb074-6968-4d57-af98-a5424ab67b87",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 0.15925879776477814,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.15121066570281982,
"example_count": 100,
"e... |
545952e7-c1a5-4cbc-8cff-20adc592542a | easy | karanganesan | 2026-08-06 05:02:59.586333+00:00 | succeeded | 1e7b60fbabe8c4ab1d491d652739779ea3a2a8eefa4baec7a3229572ceeeb3cb | 25,599 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "545952e7-c1a5-4cbc-8cff-20adc592542a",
"created_at": "2026-08-06 05:02:59.586333+00:00",
"db_md5": "64900cfd79e33625560a670b6eb9044d",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "9c09eae4-f884-45ae-987c-703b1930ab25",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.3002102375030518,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666716337204
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2797272205352783,
"example_count": 10... |
54615f28-726e-46d4-9bfa-08a7f377da82 | easy | yashkant | 2026-08-14 15:34:36.231771+00:00 | succeeded | c2206a8bdc7e40bd61218c448683aae15a7023285b33f88f47293c6aaeab53dc | 41,540 | null | """Round1701 Dykstra width16/hidden32 architecture pair on the D416 MemGLU carrier.
The fixed Round1141 carrier first produces its ordinary pooled representation.
A compact primal state and two Dykstra correction memories are initialized
from that immutable representation. Every parsed public-T cycle applies two
cond... | {
"id": "54615f28-726e-46d4-9bfa-08a7f377da82",
"created_at": "2026-08-14 15:34:36.231771+00:00",
"db_md5": "265f8a39a3d4f85a2489ebead8d7c850",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "80549ebf-e1cd-444f-9a90-49ca38d2d1d9",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 2.1331200065480345,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018125000158324838
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.132100338588806,
"example_count": 800... |
5476937f-7ef3-43ab-8adb-e04dc5ea6ce2 | easy | khushidahi | 2026-08-07 02:25:32.131159+00:00 | succeeded | 26e4fa0f318257faea3683f4d09f59288770e7f96c4293559ff7f09c847b5d76 | 18,882 | null | """R6 structured decimal-workspace model for One Layer Deeper.
The model does not implement multiplication, modular reduction, or the public
recurrence. It uses the public tokenizer structure to build learned, right-
aligned decimal tapes for N, X, and T, then applies a shared neural transition
to a mutable work tape.... | {
"id": "5476937f-7ef3-43ab-8adb-e04dc5ea6ce2",
"created_at": "2026-08-07 02:25:32.131159+00:00",
"db_md5": "03a6282f69744bcc4dbbb62a12454126",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "f1980fc5-0944-4844-b020-41036ff93ea9",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.077160954475403,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007395833497866988
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.074944257736206,
"example_count": 1200... |
5478516f-a4aa-4403-af28-f908fb1133e2 | easy | Ranoobaba | 2026-08-16 18:19:18.229782+00:00 | succeeded | 34fa8bc0b3c28ac6f833dd7dc34de2ba2db73514c3eadc328a681f935343a408 | 6,310 | null | """Latent-register recurrence: iterate a small register state, not the
whole sequence. Cheap iterations buy more of them; the sequence is encoded
once and reread by the registers every step (input injection stabilizes
deep recurrence)."""
from __future__ import annotations
import math
import torch
import torch.nn.fu... | {
"id": "5478516f-a4aa-4403-af28-f908fb1133e2",
"created_at": "2026-08-16 18:19:18.229782+00:00",
"db_md5": "fdc5297869555a1222c678cbf0e91847",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "26a80729-5ce3-4141-a5d0-856a916a5d07",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 4.202333846888934,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03333333276212216
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.036709308624268,
"example_count": 100,
... |
547a6d28-c5e1-4dc4-8295-aa9d750435de | easy | jordanrubin | 2026-08-12 04:13:34.019210+00:00 | succeeded | a8113fa5fd30cd348b465897ea29b52cff29e3c0818fa036daff3b32834a3bbe | 22,495 | 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": "547a6d28-c5e1-4dc4-8295-aa9d750435de",
"created_at": "2026-08-12 04:13:34.019210+00:00",
"db_md5": "73f7cb6eef76bd925bbd37d57d0aa13b",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "2dd03b03-9940-4fdf-bd2c-1d4889609f6f",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.564951182451688,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666405896345
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1649734528877276,
"example_count": 600... |
5480030b-2cba-4fdd-b982-156796faacf0 | easy | DDanlov | 2026-08-28 22:04:15.469578+00:00 | succeeded | ebd425885c99cc125bab22f449759041f486d392a62600c295cb310c80dc9f3c | 15,064 | null | """
Trial 30: Model C: D=512, H=8, M=4, u=2 (T=8)
"""
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, To... | {
"id": "5480030b-2cba-4fdd-b982-156796faacf0",
"created_at": "2026-08-28 22:04:15.469578+00:00",
"db_md5": "af4582082a1ac77499d12e11ecd598ec",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "41a2dc18-930a-425c-81b6-5aeb8d25c19c",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.2830499005107954,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.284268800987791,
"example_count": 600,
... |
5488ac41-6894-4b51-809a-3ae721d921b2 | easy | erdavis0 | 2026-08-29 00:55:34.523127+00:00 | succeeded | 516e9d9d041675bd73a4c56b612d0acbc7225da7b8d845cdabab52ec3efc5ebc | 20,130 | null | """Curvature table with a cached gentle modulus-relative residual."""
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... | {
"id": "5488ac41-6894-4b51-809a-3ae721d921b2",
"created_at": "2026-08-29 00:55:34.523127+00:00",
"db_md5": "dac67f1d3596515bff6b89e826ff98e0",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "d5b2dcc9-8a9c-44d2-9f68-2388bbc7affa",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 0.21643178164958954,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9815315902233124
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.22573675215244293,
"example_count": 60... |
548b8d57-f50d-4379-8500-f7b9e5778e1f | easy | sapient-sapiens | 2026-08-25 07:51:03.383533+00:00 | succeeded | 5ea3ba35526716f84acf85a285f887631fd8a54cc9f78047506f804b9acddb02 | 18,917 | null | """Standalone fixed-depth small-model E5 scaling candidate.
Generated variants differ only in the CONFIG literal. Inputs remain
categorical token/group/left-relative embeddings. No N/X/T field is decoded
numerically, fixed depth is independent of T, and only the final answer is a
training target.
"""
from __future__ ... | {
"id": "548b8d57-f50d-4379-8500-f7b9e5778e1f",
"created_at": "2026-08-25 07:51:03.383533+00:00",
"db_md5": "5dffd63464b3f52721a02c3f13a19173",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "ff488c63-0917-4f9f-93e2-f61454141af7",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 4.612187417162571,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013750000024835268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.630122457919099,
"example_count": 600,... |
5494acc6-9b86-400a-a2eb-a028daf590b0 | easy | Yalyenea | 2026-08-11 18:40:11.986732+00:00 | succeeded | 9349412015dd8ee467369dc092aa2f3c90f330424a7c34316f295500d99d3982 | 11,570 | null | """Full-time cosine wide official causal-LM R8 sandwich loop with shared input reinjection and hyperspherical Muon."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpe... | {
"id": "5494acc6-9b86-400a-a2eb-a028daf590b0",
"created_at": "2026-08-11 18:40:11.986732+00:00",
"db_md5": "36c40e3661d0c6ca2e130c3f466779dd",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "c8e182fb-6547-436b-878b-b79d06589ada",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 6.157758592485308,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00791666703298688
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.112541198730469,
"example_count": 600,
... |
549630b7-7e7c-4336-84a7-63bde27ea820 | easy | gauravmishra | 2026-08-08 02:06:34.090611+00:00 | succeeded | adcc65d01978fe5961c3181b18a0e7c523092969f7a0452f93553e5543ffd75b | 3,881 | null | """Four-pass Transformer with two blocks applied in an A-B-A-B pattern."""
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 = ... | {
"id": "549630b7-7e7c-4336-84a7-63bde27ea820",
"created_at": "2026-08-08 02:06:34.090611+00:00",
"db_md5": "3340cb471ec609c7e01b51846d4354b3",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "275f70dc-5caa-4f80-8445-e2f0500d3b88",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 4.985328725587346,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0066666666915019355
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.675028418211659,
"example_count": 600... |
549fb8e9-92e5-4c23-b002-53273e5d564b | easy | ehonig | 2026-08-20 09:48:51.935414+00:00 | succeeded | 773fa6b1624b440ec08cadf1129bcf5827d57d33621b66e7c1c39603577f84e0 | 24,718 | 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": "549fb8e9-92e5-4c23-b002-53273e5d564b",
"created_at": "2026-08-20 09:48:51.935414+00:00",
"db_md5": "f669ab457dd4c328920094cea8545bb2",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "1869e9f1-3d3b-4417-ac79-a7ae08e2e989",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.1816684007644653,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1954381465911865,
"example_count": 100,... |
54a048b6-a08f-470b-8318-60f4b3bbac27 | easy | nikolageorgiev2000 | 2026-08-04 15:14:43.660187+00:00 | succeeded | 8bd87bceda90d6a73910147f9933a170bf9135c0256ad3ebb108507adf0d70b8 | 29,485 | null | """T-composed categorical-digit reasoner with fully learned pair/reduction maps."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
asser... | {
"id": "54a048b6-a08f-470b-8318-60f4b3bbac27",
"created_at": "2026-08-04 15:14:43.660187+00:00",
"db_md5": "ad3f2141ad181c2babb2d9a9e918b400",
"submitter": "Nikola Georgiev",
"github_login": "nikolageorgiev2000",
"run_id": "e037cbe8-997b-4fdb-a564-66750358f3cb",
"tier": "easy",
"dataset_id": "e5",
"s... | {
"score": {
"mean_loss": 2.1917390170906446,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013750000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1894343359053403,
"example_count": 60... |
54a29a8b-da98-400b-81ca-032acf37ab3d | easy | jordanrubin | 2026-08-07 01:06:29.787933+00:00 | succeeded | 6ef54f3395f2231317db29dfa9a4e59f2048ce85b480b685bdc75d899e8273ca | 14,787 | null | """Autonomous digit-state recurrent Transformer for repeated modular squaring.
The model is deliberately organized around one learned transition:
p_0 = right_aligned_decimal_digits(x)
p_{k+1} = F_theta(p_k, decimal_digits(N))
The same two-block Transformer cell is applied exactly T times. T is used only
as ... | {
"id": "54a29a8b-da98-400b-81ca-032acf37ab3d",
"created_at": "2026-08-07 01:06:29.787933+00:00",
"db_md5": "a23fb0fce9b64d80be6a466e757842a3",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "212985ed-cf04-4096-b962-1f1b9863d1bd",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 4.112897895354655,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0041666665735344095
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.12093846012124,
"example_count": 600,... |
54ae37d6-3200-4377-8c2d-6bcd9dbaed7a | easy | yunjiangster | 2026-08-28 04:46:55.092525+00:00 | succeeded | ad88c332f2975331d22257fa785ba443972b0f52840e3a58263c036fdde47048 | 10,651 | null | """Shared neural one-step transition experiment for One Layer Deeper.
Purpose
-------
This is deliberately *not* a symbolic arithmetic solver. It uses the public
Easy/Medium prompt format only to parse N, x, and T, then learns a shared
neural digit-to-digit transition F_theta. The same F_theta is applied T times.
T... | {
"id": "54ae37d6-3200-4377-8c2d-6bcd9dbaed7a",
"created_at": "2026-08-28 04:46:55.092525+00:00",
"db_md5": "914df0962a28bcde9a86fe7f065b7d1d",
"submitter": "Yunjiang Jiang",
"github_login": "yunjiangster",
"run_id": "cc2df748-968b-4010-b738-a09bf9c68cb7",
"tier": "easy",
"dataset_id": "e1",
"status":... | {
"score": {
"mean_loss": 2.2577664852142334,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2299840450286865,
"example_count": 100,
"exa... |
54b21723-c306-4281-bcee-c965d6f5ea0e | easy | viridale | 2026-08-08 19:53:19.138481+00:00 | succeeded | 1b5693fbaedf9918b23f46a0006a355cc18783f729cc94ff4fb48ebd745633e2 | 9,718 | null | """Dataset-global state table with canonical terminal-state likelihood.
hypothesis: v355 reached 37.67% while training transition rows only through a jointly
learned decimal chart. A generic categorical encoding of each supplied final
answer directly anchors the terminal state, removing chart permutation and ... | {
"id": "54b21723-c306-4281-bcee-c965d6f5ea0e",
"created_at": "2026-08-08 19:53:19.138481+00:00",
"db_md5": "d11875debc729a76b3524ac58ed2b51b",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "c9a50632-6294-45d4-bb1f-107dd1baaaad",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 3.3281203508377075,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4116666838526726
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.120719909667969,
"example_count": 100,
... |
54b2d82e-8b4a-4b81-a06e-879b8bab5f76 | easy | newjordan | 2026-08-09 20:20:27.680982+00:00 | succeeded | a729f6bb0df1dfe48b68658b6d747e9d765131484ce839d06a15dac32399a53d | 94,093 | null | """Helix-Crawler — a recursive-weight-shared "crawler" ALU for modular squaring.
The task is r_{i+1} = r_i^2 mod n, chained T times: a *recurrence* over a single
operator. This build embodies the crawler / Frugendorff lineage (spark:~/sota_crawler)
mapped onto that structure. n never enters as a table row (unseen modu... | {
"id": "54b2d82e-8b4a-4b81-a06e-879b8bab5f76",
"created_at": "2026-08-09 20:20:27.680982+00:00",
"db_md5": "c3aa11c6494d5acd02af87ecf0d026d9",
"submitter": "Frosty40",
"github_login": "newjordan",
"run_id": "5a8b3653-da28-4145-9aab-2200959ba248",
"tier": "easy",
"dataset_id": "e4",
"status": "succeed... | {
"score": {
"mean_loss": 3.6930176343657465e-07,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.6912111298766474e-07,
"example_count": 1200,
... |
54bca1dd-9c22-461c-8d82-a56067058f6a | easy | DDanlov | 2026-08-15 20:53:58.621632+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": "54bca1dd-9c22-461c-8d82-a56067058f6a",
"created_at": "2026-08-15 20:53:58.621632+00:00",
"db_md5": "623970f9d216ffb45b2ef4f8507de941",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "75324b75-3fd3-44c0-a118-d76eabf92f36",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.2135479610762587,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02166666684051355
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1131612654052563,
"example_count": 100... |
54c21940-6825-4b53-810b-eb0d0415c47d | easy | isaac0804 | 2026-08-06 15:00:48.003056+00:00 | succeeded | 826b1466551a1d501e2aeca958016316d26abc2498d3cfffe917cebf9659c8db | 4,191 | null | """Tiny single-pass Transformer + high weight decay, for grokking-curve tests.
Same architecture as baseline_adamw, shrunk (D_MODEL 128->32) so many more
optimizer steps fit in a fixed CPU wall-clock budget, and weight_decay raised
1.0 (Power et al. 2022's grokking setting) to accelerate any memorize->
generalize... | {
"id": "54c21940-6825-4b53-810b-eb0d0415c47d",
"created_at": "2026-08-06 15:00:48.003056+00:00",
"db_md5": "1e787d89c2b1748aa5e932ad87dd6e14",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "0821f7bc-bfe2-4b95-bdd8-cf57c9756f08",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 2.189180904712501,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005625
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.225283050918258,
"example_count": 800,
"... |
54c2eeb7-fd46-4b1f-bfb6-9b2cf41b2460 | easy | pepinu | 2026-08-18 15:52:42.314525+00:00 | succeeded | a640950df5c33c6c1dbd2b745e696e7ba765e7d3b03b5ee5e3016b3482978deb | 17,922 | 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": "54c2eeb7-fd46-4b1f-bfb6-9b2cf41b2460",
"created_at": "2026-08-18 15:52:42.314525+00:00",
"db_md5": "e228f120ef9e981d90241e06dbc44737",
"submitter": "pepinu",
"github_login": "pepinu",
"run_id": "1462c81f-5952-4a82-a786-3912d491301f",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
... | {
"score": {
"mean_loss": 33.775850150731316,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6575000011920928
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 42.571124712759875,
"example_count": 800,... |
54c6ef10-4c25-4048-bb34-e2a3f76c09f0 | easy | c-nnr | 2026-08-04 11:52:17.183931+00:00 | succeeded | 0dbb8120bed934bc723c8c15d2353947ffb48c6f58a80098b7e43247d1231770 | 4,524 | null | """One Layer Deeper experiment: Small TRM-style x/z/y iterative refinement model: 2 train cycles, 4 eval cycles."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submissio... | {
"id": "54c6ef10-4c25-4048-bb34-e2a3f76c09f0",
"created_at": "2026-08-04 11:52:17.183931+00:00",
"db_md5": "9e341e8f346c8780a8bfa871a377e82f",
"submitter": "c-nnr",
"github_login": "c-nnr",
"run_id": "a38ec771-6007-42ab-afaa-7308d87aaae0",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.60512113571167,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03166666720062494
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1785054206848145,
"example_count": 100,
... |
54cb5a0e-0d33-4a83-9e9f-0acf42f75513 | easy | jordanrubin | 2026-08-08 10:58:35.038576+00:00 | succeeded | 19e434c7e9fc9d9d601292a0f5c0a4083813bc7790caf87c0582fc1bbcc2cad8 | 16,909 | null | """Autonomous digit-state recurrent Transformer for repeated modular squaring.
The model is deliberately organized around one learned transition:
p_0 = right_aligned_decimal_digits(x)
p_{k+1} = F_theta(p_k, decimal_digits(N))
The same two-block Transformer cell is applied exactly T times. T is used only
as ... | {
"id": "54cb5a0e-0d33-4a83-9e9f-0acf42f75513",
"created_at": "2026-08-08 10:58:35.038576+00:00",
"db_md5": "671c15c2db39c821999bae96a7defa51",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "aab283fa-e653-409e-af8f-cba2f1490ae6",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 5.008879739447416,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010416666356225809
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.097648594411499,
"example_count": 600,... |
54d2f6ea-a31f-478d-8734-5d49091867f9 | easy | yashkant | 2026-08-16 00:03:59.707615+00:00 | succeeded | 571da1e896f343c1988d5debab68ab8bf9394edae3be0448576cf37ece4a0f69 | 44,423 | null | """Round1758 Dykstra/slotwise parent plus a categorical-N RBF read.
Every Dykstra projection recomputes a tiny state-dependent radial-basis
read over immutable categorical N-token features and reinjects an
immutable X-field anchor. The matched control executes both RBF reads
but selects a learned static query, removi... | {
"id": "54d2f6ea-a31f-478d-8734-5d49091867f9",
"created_at": "2026-08-16 00:03:59.707615+00:00",
"db_md5": "764feb58c30ea5da69f0142488f8fd97",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "d243f796-97cf-49ed-8b25-bce0f00df481",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 2.124471716839776,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018750000009313227
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.123963118100999,
"example_count": 800,... |
54d58b0c-e9b9-45e2-a260-421fe4daaca2 | easy | Yalyenea | 2026-08-12 06:31:56.780852+00:00 | succeeded | 778b1a6acb91850097c48bb1baf07e652fd4d4bb0100eeead6243d2538a00ab2 | 10,638 | null | """Adaptive decimal-slot recurrence with an exact residual macro-state path."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_sta... | {
"id": "54d58b0c-e9b9-45e2-a260-421fe4daaca2",
"created_at": "2026-08-12 06:31:56.780852+00:00",
"db_md5": "a22308560fec9864e2c4d99f74e90326",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "62d15287-1f39-49ea-90ed-21a3d60016c9",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.218847845722963,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0050000002374872565
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2606751918792725,
"example_count": 60... |
54d8e514-4ea6-4fd0-9c58-174aa33cc748 | easy | DDanlov | 2026-08-26 20:33:52.183772+00:00 | succeeded | 92e990820445c6aad3f378394bcb1eef1196f4fdae38cd6c200a1046fb24d6e6 | 34,344 | null | import math
import time
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from onelayerdeeper.api import Submission, OptimizerBundle, ModelSpec, OptimizerSpec
except ... | {
"id": "54d8e514-4ea6-4fd0-9c58-174aa33cc748",
"created_at": "2026-08-26 20:33:52.183772+00:00",
"db_md5": "88544f1c864da1b16f9256216b68f6d9",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "9d666291-dca3-4e1f-b409-d577f0a262fd",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.2986836853467114,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2995289323575827,
"example_count": 600,... |
54da3cfd-9726-444c-b505-ae00eec29363 | easy | DESU-CLUB | 2026-08-30 23:56:04.879967+00:00 | succeeded | 32c30c99e775c03d265a296a458ac7ffef1d03759f70dd9d1d99c6d7731a9f23 | 15,591 | 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": "54da3cfd-9726-444c-b505-ae00eec29363",
"created_at": "2026-08-30 23:56:04.879967+00:00",
"db_md5": "a2af0b94f02313e37a59f80b488b7cc8",
"submitter": "Low Keng Hoong, Warren",
"github_login": "DESU-CLUB",
"run_id": "b2e5ff34-ec33-47e6-b0bb-6751ec152a82",
"tier": "easy",
"dataset_id": "e5",
"sta... | {
"score": {
"mean_loss": 2.7803589417980845,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3300351386914873,
"example_count": 60... |
54dce4ee-aecf-45e6-b695-908f553e22ce | easy | chad-atexpedient | 2026-08-06 03:17:58.031362+00:00 | succeeded | 495b650f34d32ccba3cd49fd0dc04fe17890232045d8ce4bd404ee47e5dfdcac | 11,720 | null | """C04: R028F with Grokfast gradient EMA filtering."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
LANES = 6
CHANNELS = 24
MUTABL... | {
"id": "54dce4ee-aecf-45e6-b695-908f553e22ce",
"created_at": "2026-08-06 03:17:58.031362+00:00",
"db_md5": "e9b57fe86f52490e3b76055eb14bf3b7",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "99b7395e-e0fd-438f-af29-136a5e1f55d1",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.209896613158933,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002083333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2112842001722535,
"example_count": 600... |
54e67aff-65e8-4cd2-bdf0-33b36c720bd6 | easy | erdavis0 | 2026-08-22 01:59:48.500922+00:00 | succeeded | 99321c695c192b9ea1a09ae05c38e09a8617a6fcdd3b383274e0466214741961 | 6,150 | null | """Compact latent field router with right-aligned output queries."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
WIDTH = 72
HEADS... | {
"id": "54e67aff-65e8-4cd2-bdf0-33b36c720bd6",
"created_at": "2026-08-22 01:59:48.500922+00:00",
"db_md5": "b69b6bf896dd99c570f9138aa9dd1cde",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "40fc367c-a951-4da4-ab82-81982f0b40d4",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 7.284983158111572,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08175675990059972
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.377727508544922,
"example_count": 60,
... |
54e72dbb-19b1-4929-86be-1d0ef8cb0311 | easy | viridale | 2026-08-15 18:28:47.005911+00:00 | succeeded | fcd808694223d067b89dd44b0530ce6d7e94f3aa5cc16944b141a7566edb0187 | 29,553 | null | """Boundary-preserving local evolution in a sparse cubic character fiber.
hypothesis: after the exact four-coordinate T1 boundary is latched, preserve complete
elite dependencies and explore the remaining recurrent fiber by one-coordinate
moves or cubic-owner switches instead of independent redraws alone.
axis... | {
"id": "54e72dbb-19b1-4929-86be-1d0ef8cb0311",
"created_at": "2026-08-15 18:28:47.005911+00:00",
"db_md5": "062bfc3c613abb515ba2c57f62a3ad80",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "afe003fc-3f39-49c2-bf32-3117f328d4db",
"tier": "easy",
"dataset_id": "e6",
"status": "succe... | {
"score": {
"mean_loss": 0.15029456466436386,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.15949773788452148,
"example_count": 60,
"ex... |
54e9df6d-f043-4fe2-a66c-751906015093 | easy | anirudh-chakravarthy | 2026-08-12 18:36:32.593962+00:00 | succeeded | 7b1d847457f17219a1dafd673ac58eaffacbea85d74876cefaf2acda0987c720 | 11,282 | null | """A diagonal Neural GPU over binary-coded decimal number cells."""
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": "54e9df6d-f043-4fe2-a66c-751906015093",
"created_at": "2026-08-12 18:36:32.593962+00:00",
"db_md5": "acdf86e669298d06ba4f5b816fe0c319",
"submitter": "Anirudh S Chakravarthy",
"github_login": "anirudh-chakravarthy",
"run_id": "72e95d86-453c-430d-a989-0f1e86dc8a73",
"tier": "easy",
"dataset_id": "... | {
"score": {
"mean_loss": 2.1513120954527363,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666679084302
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1507451106614597,
"example_count": 60... |
54eeb77b-5ebf-4899-8557-bdfb2737fdbd | easy | khushidahi | 2026-08-05 23:07:11.485730+00:00 | succeeded | 4eaddb18b3b2d01f5ae0c48f733438004a5e76fe7b14eac3d52041dd650600e0 | 25,722 | null | """R16 hard-transfer hybrid experiments for One Layer Deeper.
Motivated by complementary Hard behavior:
- the generic sequence recurrent model retained seen-N first-rung progress;
- the structured decimal-tape model retained OOD-N first-rung progress.
The dual variants co-train both learned representations, supervise... | {
"id": "54eeb77b-5ebf-4899-8557-bdfb2737fdbd",
"created_at": "2026-08-05 23:07:11.485730+00:00",
"db_md5": "ad11151e92602b8dfd0524cc2ca0b34f",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "2cad5d3d-6a37-4ddc-8bbc-9cbdcf209b68",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1776080545004426,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012916666840513548
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1807351112365723,
"example_count": 60... |
54f213a2-87be-4810-9883-dd7a6e1dc840 | easy | oupadhyay | 2026-08-28 01:58:33.167384+00:00 | succeeded | b66ec811ad9507b412685e6dc807e62a1f9a3ae7dc62e7ea9660d46613a3ca80 | 4,043 | null | """Literal polynomial cell."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle,Submission,TokenLossBatch,assert_model_state
D,H=32,16
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_size,s.max_seq_len=vocab_size,max_seq_... | {
"id": "54f213a2-87be-4810-9883-dd7a6e1dc840",
"created_at": "2026-08-28 01:58:33.167384+00:00",
"db_md5": "eda301b1313164e440557227fa9fa82f",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "a1dea92b-f1bb-468f-9dc5-8426cc247538",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 2.4071496725082397,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07027027569711208
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3268535137176514,
"example_count": 60,... |
54f32c67-2987-408f-abc7-0dd5ee703ee5 | easy | sapient-sapiens | 2026-08-17 11:58:02.006371+00:00 | succeeded | 2c81a075b20053268cec22c72deface05595d31c759ee2bcd2cc116301c4e4dd | 17,309 | null | """Parameterized fixed-K recurrent candidate for the additive sweep.
Knobs at the top are the only intended experimental variables. T remains an
ordinary prompt field and is never parsed into a loop count.
"""
from __future__ import annotations
import math
import time
import torch
import torch.nn.functional as F
fr... | {
"id": "54f32c67-2987-408f-abc7-0dd5ee703ee5",
"created_at": "2026-08-17 11:58:02.006371+00:00",
"db_md5": "b9ccb0a726cb5c7eb0fb4e1005c7bcb7",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "ebcfce7f-624f-4657-bb0a-dc4b4c6baece",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.3363742055327386,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005416666703919569
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3312988785082984,
"example_count": 60... |
54f5ccc5-8126-43e2-9cb7-b65824bc20f8 | easy | TZZheng | 2026-08-07 05:11:10.920365+00:00 | succeeded | 1c12a17fc2eb4a2c7b22b30020f49594602a6b119a2f3165ea09654d60af4f90 | 12,526 | null | """Learned recurrent modular-transition model.
The mutable digit state is initialized from x. A learned square-oriented
self-mixing network and an N-conditioned reduction network form one transition,
and exactly the same transition parameters are reused T times. No arithmetic
operation is hard-coded; only the public p... | {
"id": "54f5ccc5-8126-43e2-9cb7-b65824bc20f8",
"created_at": "2026-08-07 05:11:10.920365+00:00",
"db_md5": "c31e0194f81883af2abccc305e25141b",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "0bc83ad9-ab87-4355-9b9a-15c454138d62",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 7.789725303649902,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04333333298563957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.467862129211426,
"example_count": 100,
... |
54f63cad-5794-40c0-9e12-b9e0d49b40df | easy | chad-atexpedient | 2026-08-20 15:58:08.905987+00:00 | succeeded | 88414833fd0aeb52b1fd72ff501abd5bff1bdd818e03c064d33f4529e9f49f46 | 22,987 | null | """X119: RNG-parity-fixed linear digit-coordinate residual.
X118 combined linear and squared digit coordinates at full projection strength.
X119 tests the minimal ordering bias: one centered scalar coordinate, projected
through an isolated module and gated near 0.25. Every X98E tensor remains exactly
seed-parity match... | {
"id": "54f63cad-5794-40c0-9e12-b9e0d49b40df",
"created_at": "2026-08-20 15:58:08.905987+00:00",
"db_md5": "2ac4a5fbc0f6b4fc5c1df2c394f0c125",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "7bcdc231-225f-4b6b-b75c-e40276f098e3",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.4008879237081944,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01041666670391957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4798390982931506,
"example_count": 600... |
54f8d3ae-8705-44fb-bde0-ba214a72aa21 | easy | jayyala | 2026-08-16 23:24:43.227051+00:00 | succeeded | 84301c5daa7cf3516f2dca8586af0e29a12d0caaf01032a2b3ca60113fc8eacd | 20,507 | null | """One learned modular-square step, applied exactly T times.
The underlying map is recovered in a modulus-equivariant form: the integer
product x*x is represented as a circular phase, which wraps automatically,
and a tied coarse-to-fine decoder inverts that phase to a residue. The same
decoder is trained by reconstru... | {
"id": "54f8d3ae-8705-44fb-bde0-ba214a72aa21",
"created_at": "2026-08-16 23:24:43.227051+00:00",
"db_md5": "9e92b4e9a5e58a3ab723917f67bbf1db",
"submitter": "jayyala",
"github_login": "jayyala",
"run_id": "bb9aecb8-3b71-4f9c-9ecc-fc466bea7985",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded"... | {
"score": {
"mean_loss": 35.42687797546387,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333767950535
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 35.67567443847656,
"example_count": 60,
... |
54fd535a-eea0-419f-a4eb-eda7f264cbc0 | easy | sapient-sapiens | 2026-08-14 12:38:37.387661+00:00 | succeeded | 32c22888c589dfc4eace9a4edcef37c675cb93d9ebd934e9eb644db800727ba3 | 16,887 | null | """Fixed-T=2 tied Transformer for the E3/E4 operator-recovery campaign.
This file is the authoritative, standalone hosted submission template. The
variant generator changes only the constants in the marked configuration
section. The model never reads T: its four-block core is applied exactly twice.
"""
from __futur... | {
"id": "54fd535a-eea0-419f-a4eb-eda7f264cbc0",
"created_at": "2026-08-14 12:38:37.387661+00:00",
"db_md5": "28e3533e6fb9162bd88623a88f957a65",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "323ca732-61c7-4a78-a0b9-72efc40d04db",
"tier": "easy",
"dataset_id": "e4",
"status": "su... | {
"score": {
"mean_loss": 2.064139265765891,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0036458333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0653831300654693,
"example_count": 12... |
5504eeaf-b0d3-4a9d-9b01-3af2d6b31807 | easy | AFeinfield | 2026-08-12 23:43:48.009896+00:00 | succeeded | 66007340493fbc23410d6b40178187103204f98fe6ed4019868ce65c6fc580e3 | 5,694 | null | """Looped (weight-tied) bidirectional Transformer, d=256, 8 iterations.
003: 002 + GrokFast-EMA gradient amplification (alpha=.98, lamb=2) on AdamW.
prior. Same parameter count order as a 1-block model; depth comes from reuse.
Warmup + cosine LR. No compile (60s budget).
"""
from __future__ import annotations
import... | {
"id": "5504eeaf-b0d3-4a9d-9b01-3af2d6b31807",
"created_at": "2026-08-12 23:43:48.009896+00:00",
"db_md5": "7b5c2e3c9249ae4a27a36d8cb33b0493",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "73b61fc6-1aeb-4c81-a307-455722dd402f",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.82824045419693,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.038333332166075706
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.7561980485916138,
"example_count": 100,... |
5509421a-ca92-40e0-b951-04e17adc8e91 | easy | chad-atexpedient | 2026-08-09 21:06:04.114835+00:00 | succeeded | 8d7cc7fcb7765f01d6944e6eff0d0dc795e3ebb2732e85f77f102fa0d477ec2e | 11,001 | null | """BL6B: budget-normalized LR continuation with boundary anchor."""
from __future__ import annotations
import torch
import torch.nn.functional as F
import time
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state... | {
"id": "5509421a-ca92-40e0-b951-04e17adc8e91",
"created_at": "2026-08-09 21:06:04.114835+00:00",
"db_md5": "f20c4a9f4d9dede5fb3856e2b6cd752a",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "ed0bff04-d22f-4927-a2cb-c93750a2fa35",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.9845286343978574,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.972655503739156,
"example_count": 600... |
550af55a-f0d4-4eb3-ada3-69ff42e21174 | easy | AFeinfield | 2026-08-27 21:43:26.699111+00:00 | succeeded | bbe9a662fc80ddea32cc2f176fb798793467e10443f728bf6db7371985d12204 | 11,655 | null | """149: 148 + LEARNED QUOTIENT ESTIMATION (throughput fix).
148 learned (loss 2.85->2.67 in 35 steps) but ran at 0.6 steps/s: the
16x9 blind conditional-subtraction structure is launch-bound. 149
replaces it with a learned digit-quotient estimator per alignment (one
subtraction + 2 comparator-gated corrections) and ca... | {
"id": "550af55a-f0d4-4eb3-ada3-69ff42e21174",
"created_at": "2026-08-27 21:43:26.699111+00:00",
"db_md5": "6132b810adde234497da9ac4638dc90e",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "0b0c1da8-9836-4614-9074-1033eddd654d",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.2665218734035495,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0008333333457509677
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2686915108975807,
"example_count": 6... |
550d4784-e749-4863-a3b1-252955e77610 | easy | Ranoobaba | 2026-08-17 06:39:52.611215+00:00 | succeeded | 69344056e4a57f609530bb56e8e8000ad7df7ccf3ed837442cc08facfc61e3b2 | 5,804 | null | """Looped transformer, iteration count conditioned on T read from the prompt.
Core idea: learn one latent "squaring step" operator and apply it T times.
Depth generalization comes from the loop structure itself; the model never
needs more parameters for deeper problems, only more iterations.
"""
from __future__ impor... | {
"id": "550d4784-e749-4863-a3b1-252955e77610",
"created_at": "2026-08-17 06:39:52.611215+00:00",
"db_md5": "eeb6923f196a8dab1be1c7bdeab1f753",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "2ebcb699-0ab4-4fb5-b8c0-55d8277f2aac",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 6.450071699304177,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04166666527589162
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.586390018463135,
"example_count": 100,
... |
550fa8b3-abf0-469d-af06-d29126960a75 | easy | khushidahi | 2026-08-05 23:47:01.392651+00:00 | succeeded | dfdd20a37356db2c57a3feed71047534cc2a8fcb9e127979cabde57801ef66c5 | 20,279 | null | """R21 hierarchical value-and-scratch 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 ta... | {
"id": "550fa8b3-abf0-469d-af06-d29126960a75",
"created_at": "2026-08-05 23:47:01.392651+00:00",
"db_md5": "88e704d39010cf785c3e6dd1f0d962ed",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "36e6e1f6-4f89-4c20-9639-3a3f256f3026",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.2271933555603027,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006875000195577741
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2246828079223633,
"example_count": 30... |
551032cf-51f0-4f3d-8e5c-3afb20276646 | easy | khushidahi | 2026-08-03 22:41:18.349662+00:00 | succeeded | 56c421eedcc11b5ffec1f24752d8e26b87d45a53da8331b29ad6584322ff5041 | 16,587 | 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": "551032cf-51f0-4f3d-8e5c-3afb20276646",
"created_at": "2026-08-03 22:41:18.349662+00:00",
"db_md5": "c41e2ea6346e96903b50e04f355473d9",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "5a645ab1-0322-40db-8946-e22327fbdd15",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.075472354888916,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005208333372138441
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.072113037109375,
"example_count": 1200... |
55197834-c664-4c8f-a98c-f1ca856df8f1 | easy | erdavis0 | 2026-08-27 03:19:22.280413+00:00 | succeeded | 2631fff557194c25861f5cbd2a3339df5a8a8980d811bc985ebcaec80ee4bdc0 | 31,299 | null | """Typed-prefix reversible transition B5: CE/sequence-mixture hybrid.
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": "55197834-c664-4c8f-a98c-f1ca856df8f1",
"created_at": "2026-08-27 03:19:22.280413+00:00",
"db_md5": "2981fa4ef2c40f38bd0d3090c6d64a88",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "6b02d28c-8319-40f8-bd66-dd9147c38320",
"tier": "easy",
"dataset_id": "e4",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.588940282872901,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0034375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.568946799711618,
"example_count": 1200,
... |
55205548-1f83-4c65-8170-5a70a866b7e7 | easy | alirezashirvani-jr | 2026-08-29 09:43:48.781393+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": "55205548-1f83-4c65-8170-5a70a866b7e7",
"created_at": "2026-08-29 09:43:48.781393+00:00",
"db_md5": "050524d36ff9ab09e87d1ff83923e1e0",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "5b3fb598-2439-4703-85bd-30b1907e0b03",
"tier": "easy",
"dataset_id": "e2",
... | {
"score": {
"mean_loss": 1.0033780336380005,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.7885416746139526
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.5910929441452026,
"example_count": 300,... |
55211540-7e96-4e51-a157-4f50b2116f5c | easy | ehonig | 2026-08-20 07:53:40.188715+00:00 | succeeded | 2210b6deec67a288f1df21466b74ea0e58ab3b7bef866c81388a8a9d37fc74a4 | 23,040 | 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": "55211540-7e96-4e51-a157-4f50b2116f5c",
"created_at": "2026-08-20 07:53:40.188715+00:00",
"db_md5": "e530e999def1e0017fd9c3144c5d28a1",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "9dedb6c7-0f7d-4033-8f17-7cfd8b0f6173",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.9937965869903564,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06833333149552345
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9805493354797363,
"example_count": 100... |
55288adb-7a94-42c0-9d2b-16f49af367fa | easy | jordanrubin | 2026-08-03 21:08:20.722718+00:00 | succeeded | 09586f90ffa296e3bdbb25e6a0b059c7579344900d2f7d9eb12b0e902c4848f6 | 15,390 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "55288adb-7a94-42c0-9d2b-16f49af367fa",
"created_at": "2026-08-03 21:08:20.722718+00:00",
"db_md5": "ef7d67d872a65aa96e81ca53991ab27d",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "d3379cc1-8cde-4f5e-8de9-82299cf2afcd",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 3.1913367205449332,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.19083332844699422
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.892922591387741,
"example_count": 300,... |
552a5119-f222-45a7-beee-45d3f418f1bd | easy | repst | 2026-08-23 10:52:20.197232+00:00 | succeeded | 2c7861bb2c63cf2d98e214e0ceba7501c6a7945257f530b6fb91cbf177cc268f | 27,550 | null | # R529: streaming-MDL grammar-frontier recurrent machine.
#
# The R525 factor graph and exact recurrent interpreter are retained, including
# their small dense associative memory for direct credit. Expression syntax is
# now streamed in minimum-operation order and stopped at a semantic compute
# budget, rather than ma... | {
"id": "552a5119-f222-45a7-beee-45d3f418f1bd",
"created_at": "2026-08-23 10:52:20.197232+00:00",
"db_md5": "c128bfd6cace8a3bb540d49c5a1f15ad",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "9ab8a347-33e0-44eb-97a8-0edae6d18451",
"tier": "easy",
"dataset_id": "e7",
"status": "succeed... | {
"score": {
"mean_loss": 0.0005221988831181079,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0005257647135294974,
"example_count": 85,
... |
552a7777-6139-4774-97b3-75b9910251da | easy | DDanlov | 2026-08-21 01:04:20.228977+00:00 | failed | 6fc7c71ce6e63bc1c78e33b1b8d570fc851b2262d81f11e492f1284cfc54d071 | 21,933 | null | from __future__ import annotations
import math
import time
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
except Imp... | {
"id": "552a7777-6139-4774-97b3-75b9910251da",
"created_at": "2026-08-21 01:04:20.228977+00:00",
"db_md5": "2239eb0efc3bdf000c88883800e1daee",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "eb92a510-0150-4a03-923d-27faa9bf8c7c",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
55331648-2b8c-4d7c-9576-394a2f41c2f5 | easy | DDanlov | 2026-08-31 21:09:28.789517+00:00 | succeeded | 240547a2a7bbb916ba49e456fe5679725af4e41d5563fc47c92f02f8aa250be5 | 32,411 | null | """
Unified Continuous Hopfield Equilibrium Propagation Deliberation Engine
=======================================================================
Architecture:
- Deliberation: Continuous Hopfield Multi-Layer Energy System (L=4, D=768, d_ffn=2304, H=16)
- Integration: In-Place Strang-Split Conformal Symplectic Velocit... | {
"id": "55331648-2b8c-4d7c-9576-394a2f41c2f5",
"created_at": "2026-08-31 21:09:28.789517+00:00",
"db_md5": "b362a957b8a2cbfdce8e2731101549e1",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "390c1681-a0a6-4b98-8bf7-1f3ebb3adf6c",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.076433791879473,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05166666701436043
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0416527795257853,
"example_count": 100,... |
55386059-65fa-49ca-90ce-2a82d5185d0b | easy | AFeinfield | 2026-08-25 01:23:42.810608+00:00 | succeeded | 75ab342226f15281aa72a6f45ad54288941b5f1f10e912792e9e241de844a66e | 8,019 | null | """110: 023 (looped d256, the only lineage that FITS e5 in 60s) + the
Aug-19 tied-head init fix. CONTROL arm for the 111/112 abacus ladder.
023 fit variable-N e5 to train_acc 1.0 in 60s (test 0.5% — memorization),
but it predates the tied-head repair: head shares nn.Embedding's default
std ~1.0 weights, so logits sta... | {
"id": "55386059-65fa-49ca-90ce-2a82d5185d0b",
"created_at": "2026-08-25 01:23:42.810608+00:00",
"db_md5": "0ceeae65711a39ea3b4f6baf46673315",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "ab75c27a-43bc-46dc-b045-e43753de2e46",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1452942489184084,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1500721298525685,
"example_count": 600,
"e... |
553e34b5-3660-4e18-9b8a-2fcfe715cc15 | easy | khushidahi | 2026-08-03 21:33:57.913751+00:00 | succeeded | 53f25624b986b3cbb953f0b8495e08d117e31bf91d739cab10b3ad25268d736a | 16,556 | 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": "553e34b5-3660-4e18-9b8a-2fcfe715cc15",
"created_at": "2026-08-03 21:33:57.913751+00:00",
"db_md5": "de31eb19c609ba0aa9075b57c418790b",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "e751098e-7e82-4bdf-80b2-af6463a935b9",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.2122405767440796,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333302289248
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2050814628601074,
"example_count": 60... |
553f3490-0b64-444f-aa56-f80ca40d527b | easy | erdavis0 | 2026-08-27 00:59:13.889784+00:00 | succeeded | f696b4a0dbdab40d944b3bb47ef7b8e716ff6e0849099f9cf030611313221c72 | 31,312 | null | """Typed-prefix reversible transition B3: sequence mixture plus balance.
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": "553f3490-0b64-444f-aa56-f80ca40d527b",
"created_at": "2026-08-27 00:59:13.889784+00:00",
"db_md5": "aa01dfb95948365c687b75f02f03cd86",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "8b4a70b7-a911-47de-8a27-c1d5ff179c79",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",... | {
"score": {
"mean_loss": 3.0157858987174007,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01000000002483527
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.09433732652878,
"example_count": 600,
... |
55418eab-8d4e-4f5d-8a87-8a1a8582a8a7 | easy | sapient-sapiens | 2026-08-15 16:02:39.502374+00:00 | succeeded | f0ade338d4647a177297c12665dcaab286afb28a9b7371cf37ba7e754cf84477 | 12,818 | null | """Fixed-T=2 recurrent Transformer with semantic field-relative embeddings.
The learned four-block operator is applied exactly twice with shared weights.
Inputs remain categorical: token, prompt-field identity, and left-to-right
position within the field. No token is converted to a number and no arithmetic
operation ... | {
"id": "55418eab-8d4e-4f5d-8a87-8a1a8582a8a7",
"created_at": "2026-08-15 16:02:39.502374+00:00",
"db_md5": "8484e435dcdfb0a9bef5f595bfa7ed5d",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "db7659bc-7685-4781-850e-94a7696ea418",
"tier": "easy",
"dataset_id": "e3",
"status": "su... | {
"score": {
"mean_loss": 3.7929077529622415,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.7705462495469226,
"example_count": 800,
... |
55480d6b-1fd7-4227-af5c-04a7782133f0 | easy | jordanrubin | 2026-08-08 09:57:42.629260+00:00 | succeeded | 078cd8887c504bdcfcf580d442de465beb5b3a0ea5e1726a6c95fd2267731c91 | 16,615 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "55480d6b-1fd7-4227-af5c-04a7782133f0",
"created_at": "2026-08-08 09:57:42.629260+00:00",
"db_md5": "1f66d0c9d8b2addfecdef9a9aa694dae",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "a09bfad6-693d-4278-957d-debc045390ca",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.176748477179112,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00458333323088785
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1858301364520205,
"example_count": 600,... |
554b0f0a-478d-474a-8068-4bafe3508de8 | easy | jordanrubin | 2026-08-07 01:13:05.481142+00:00 | succeeded | 6ef54f3395f2231317db29dfa9a4e59f2048ce85b480b685bdc75d899e8273ca | 14,787 | null | """Autonomous digit-state recurrent Transformer for repeated modular squaring.
The model is deliberately organized around one learned transition:
p_0 = right_aligned_decimal_digits(x)
p_{k+1} = F_theta(p_k, decimal_digits(N))
The same two-block Transformer cell is applied exactly T times. T is used only
as ... | {
"id": "554b0f0a-478d-474a-8068-4bafe3508de8",
"created_at": "2026-08-07 01:13:05.481142+00:00",
"db_md5": "a23fb0fce9b64d80be6a466e757842a3",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "7f4f6184-be3c-4fd9-842e-c38364593728",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 7.353684176233344,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.14499999582767487
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.552156925201416,
"example_count": 100,
... |
5550450c-5cd1-438b-a676-04dd6a2de1c0 | easy | lzy54 | 2026-08-29 09:12:14.895913+00:00 | succeeded | e546a3b496e99ae2909d93d5c895ec67a5dc81ada4533f250b0a6a48d2d5b998 | 9,330 | null | """Phase 4 P3: P2 alignment features with RoPE replacing absolute position."""
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_MODE... | {
"id": "5550450c-5cd1-438b-a676-04dd6a2de1c0",
"created_at": "2026-08-29 09:12:14.895913+00:00",
"db_md5": "d0ce7ad93e2065728b688df745d4901f",
"submitter": "lzy54",
"github_login": "lzy54",
"run_id": "47e9a39e-80e5-4f1c-aaf3-e379274a05f9",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.1556623442749325,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000024835268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.15992068068329,
"example_count": 600,... |
555a25ad-e064-4b3a-9067-b70b5d03d50a | easy | oupadhyay | 2026-08-28 02:54:04.331347+00:00 | succeeded | 4fd6f8d23c9f61345a11118872420bd0e2cb8aa775fd57e5807cf74125700569 | 7,796 | null | """BiScan entropy preference with fourfold T1 endpoint weighting."""
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle, Submission, TokenLossBatch, assert_model_state
PAD, N, X, T, ANS, DIGIT = 0, 2, 3, 4, 5, 7
MAX_STEPS = 64
class Config:
def __init__(self... | {
"id": "555a25ad-e064-4b3a-9067-b70b5d03d50a",
"created_at": "2026-08-28 02:54:04.331347+00:00",
"db_md5": "328c3baceb3cfac2c1359d49f58422db",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "7ee795d9-017a-47b0-a579-0bdacc7f48c8",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 2.5138591838209594,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.19819820078240857
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.748443126678467,
"example_count": 60,
... |
555e4c5d-0dfc-47dd-a507-39cd50165df1 | easy | sapient-sapiens | 2026-08-23 08:55:59.393789+00:00 | succeeded | 74f5f20f9611b46db467144d0857c6746c15cb648550515692b7ef2a9c53877b | 20,235 | null | """Approximately 10x-scale adaptive child of the D6 pair/quotient architecture.
The architecture uses the required learned additive token/group/left-relative
representation. Marker-derived field lengths additionally route categorical
digit embeddings into right-aligned lanes; this alternative place alignment is
an ex... | {
"id": "555e4c5d-0dfc-47dd-a507-39cd50165df1",
"created_at": "2026-08-23 08:55:59.393789+00:00",
"db_md5": "250cf977ef786c2b1e6ffe4114c5ec60",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "476106f5-40a2-4c1b-bd4e-7451d88717e2",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.169968150252381,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666716337204
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.17271845116209,
"example_count": 600,
... |
555e86ab-0c91-4b4a-a4c7-f07ff42f3b1a | easy | yashkant | 2026-08-26 07:46:59.401735+00:00 | succeeded | 621e63d0a995520cd2af40db43a35110851477f2e2dad36c0d328f6263cac9f8 | 24,538 | null | """One Layer Deeper HARD submission v7: prime-agnostic repeated-squaring digit machine.
The entire model is three tied digit cells (~86K params):
mul: (digit, digit, carry 0..8) -> digit, carry [schoolbook partial products]
add: (digit, digit, carry 0/1) -> digit, carry [partial accumulation, N-doubling]
... | {
"id": "555e86ab-0c91-4b4a-a4c7-f07ff42f3b1a",
"created_at": "2026-08-26 07:46:59.401735+00:00",
"db_md5": "82455da07da55c204565fc3a0c02fc5a",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "54799674-5aaf-4db3-b003-c54271025905",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.3093083596175856,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014166666977107525
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3165316581726074,
"example_count": 60... |
5563ae21-707b-46b7-a522-4c7e7933c60d | easy | DDanlov | 2026-08-28 21:59:15.671622+00:00 | succeeded | bebe41f967ea9a98b50f1a86cb23afa07bba3415a831dd93dc811683911d0829 | 15,069 | null | """
Trial 26: Model C: D=256, H=4, M=32, u=2 (T=64)
"""
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, ... | {
"id": "5563ae21-707b-46b7-a522-4c7e7933c60d",
"created_at": "2026-08-28 21:59:15.671622+00:00",
"db_md5": "2b1975a428579947f024687e578f75b0",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "a8d56cf7-9890-4c2c-b325-87d9acf48629",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.58311701200736,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6032760117887914,
"example_count": 600,... |
556bef10-0077-4a36-a439-bb69181bd2b8 | easy | jordanrubin | 2026-08-02 18:11:37.578336+00:00 | succeeded | 2dde7cb00ecbada10f77e9d3d68006074e49407eed42ab1f3187d20ffea8fbcb | 13,993 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "556bef10-0077-4a36-a439-bb69181bd2b8",
"created_at": "2026-08-02 18:11:37.578336+00:00",
"db_md5": "300c787dfb590f35607a4c406c26c779",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "152ac114-4241-4a29-a6c1-aae5f3d136ae",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 5.609078503260238,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0899999967465798
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.208071986241127,
"example_count": 100,
... |
5581f388-7f85-4123-b006-47c3cb30bae9 | easy | sapient-sapiens | 2026-08-25 08:07:33.305204+00:00 | succeeded | fc7bc8bfc9fb2ed79003f4009b6d577a90e039c539c9f00e87218cde15903ae0 | 18,868 | null | """Standalone fixed-depth small-model E5 scaling candidate.
Generated variants differ only in the CONFIG literal. Inputs remain
categorical token/group/left-relative embeddings. No N/X/T field is decoded
numerically, fixed depth is independent of T, and only the final answer is a
training target.
"""
from __future__ ... | {
"id": "5581f388-7f85-4123-b006-47c3cb30bae9",
"created_at": "2026-08-25 08:07:33.305204+00:00",
"db_md5": "ba6e5994cf86f62e30dc9a3f6612ad21",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "fc297dfd-3a90-427f-ab8c-8e497ecead24",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 6.069450082560419,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010416666666666668
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.157842300397933,
"example_count": 600,... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.