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 |
|---|---|---|---|---|---|---|---|---|---|---|
3548f3a4-ad71-4765-83bb-8e7a0f73b303 | easy | briheet | 2026-08-07 03:12:16.230322+00:00 | succeeded | 3fdfe3f5e6c8ca92438abcb19de02ff9f4e88a7c45c039a35c771cf4f1883d39 | 6,896 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 64
NUM_HEADS = 4
NU... | {
"id": "3548f3a4-ad71-4765-83bb-8e7a0f73b303",
"created_at": "2026-08-07 03:12:16.230322+00:00",
"db_md5": "7709eb0981cbc75ba965dfb09146132c",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "a7ba65a3-532c-44a5-9463-e02c85de7dde",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 4.14744758605957,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0283333333209157
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.17537784576416,
"example_count": 100,
... |
354a32b2-d595-4a97-9031-92e404f517e7 | easy | oupadhyay | 2026-08-28 02:26:37.828478+00:00 | succeeded | 79694030b38024b641bdc729c71b2922a9a008a3bca5f9766fc017c0d920a3ca | 6,747 | null | """Easy E5 Neural-GPU with one homogeneous recurrent grid cell."""
import torch
from torch import nn
from benchmark import OptimizerBundle, Submission, assert_model_state
PAD, N, X, T, ANS, DIGIT = 0, 2, 3, 4, 5, 7
CHANNELS, MAX_PLACES, MAX_STEPS = 48, 16, 64
STATE_ELEMENTS = 64_330
class Config:
def __init__(s... | {
"id": "354a32b2-d595-4a97-9031-92e404f517e7",
"created_at": "2026-08-28 02:26:37.828478+00:00",
"db_md5": "98192a45597baf6fe89b1f88d847e528",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "ca00b96c-8b3d-43f6-b43e-e77026ead4cd",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 2.3051750659942627,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010135135613381863
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3579845428466797,
"example_count": 60... |
354aeba2-6d8b-4b81-8043-b7f21cd1372a | easy | ehonig | 2026-08-14 01:23:52.908698+00:00 | succeeded | d4863f4de678d842869b640fbf6ce59ff6dec0b2ff6dff8337c93025539210f1 | 54,753 | null | """DigitLoop — a place-value recurrent model for repeated modular squaring.
The task is ``y = x^(2^T) mod N`` with the answer read back tail aligned, one
decimal place value per token position from the right. A stack of attention and
MLP layers has no primitive for the two operations this needs — a partial-product
co... | {
"id": "354aeba2-6d8b-4b81-8043-b7f21cd1372a",
"created_at": "2026-08-14 01:23:52.908698+00:00",
"db_md5": "f62c4fc3d75fd4327b86be1aebf443ce",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "d8151956-2fc2-4fd4-b0bf-80406ffc70bd",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 5.788520336151123,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.1966666728258133
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.131183624267578,
"example_count": 100,
... |
35532e42-10a7-4bde-8a2e-b8622470f7af | easy | liam-gb | 2026-08-13 00:58:15.022924+00:00 | succeeded | 9db49abf7e909af0722b300a0081859b559063890d7ec3fa0650817e9e1d46d6 | 14,529 | null | """rns_pm_gather_ct_widet_nc: widet with N-conditioning re-injected each loop.
The wide readout is what makes long answers reachable; the trim is what pays
for it, by not running loop iterations whose update is multiplied by zero.
Base docstring:
Single change vs rns_pm_gather_ct: the readout carries one
distance-fr... | {
"id": "35532e42-10a7-4bde-8a2e-b8622470f7af",
"created_at": "2026-08-13 00:58:15.022924+00:00",
"db_md5": "e8ff57bb0d37941db3f6534309a0d17b",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "68d56c7f-39e7-4b05-8bbc-c63f8f3b78bf",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.3828736322487263,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014583333289871613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.7726025581359863,
"example_count": 60... |
355967ac-b7a5-40f4-9b31-f8931b9f2cf2 | easy | sapient-sapiens | 2026-08-09 15:13:47.082913+00:00 | succeeded | cea92a1f7ddc2fbcda9957fc43eb31e0810d8f594ad956514ccb9c2af278d54d | 13,059 | null | """Near-500M adaptive recurrence grown from the hosted-working baseline."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
... | {
"id": "355967ac-b7a5-40f4-9b31-f8931b9f2cf2",
"created_at": "2026-08-09 15:13:47.082913+00:00",
"db_md5": "24504537fd1e6d1ad3c971c202f4bd3b",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "71918d15-91b4-44b9-9320-29a018c15f9a",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 1.9571857100376424,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.940808930948599,
"example_count": 100,
"exac... |
355e2a96-54a1-4d1d-83a9-a4705e4295b8 | easy | erdavis0 | 2026-08-25 01:51:51.221648+00:00 | succeeded | 441829d198f853ac2efa0065a0a953215d79f7cb90ee2b49280f23481a092acf | 24,085 | null | """Matched categorical-boundary panel: C1 continuous endpoint auxiliary.
Public delimiter syntax is used only to route opaque token identities onto
separate right-aligned tapes. One T-blind learned macro maps a categorical X
tape and a digitwise N tape to a new categorical tape through freshly reset
interaction, carr... | {
"id": "355e2a96-54a1-4d1d-83a9-a4705e4295b8",
"created_at": "2026-08-25 01:51:51.221648+00:00",
"db_md5": "18bcf63268fc44cb8c6a3e621ae926dd",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "c0dd7189-165a-4c64-a052-c76aa94e6600",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 1.8754958510398865,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06036036275327206
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8522833585739136,
"example_count": 60,... |
3566bcfa-9b2c-46ac-bb2c-37ebe1432169 | easy | chad-atexpedient | 2026-08-09 02:46:52.050450+00:00 | succeeded | 64abca8f0d46ee77c4069a2192654c4932ec9552a94e8ed3af14b3e5429f6c4e | 17,960 | null | """X04C: categorical pair/feedback factorial arm.
The mutable register is initialized from x once. Every subsequent application
of the tied cell receives only the current LSD-first digit state and immutable
N digits. Requested T controls only the number of applications. The state
logits are also the answer logits,... | {
"id": "3566bcfa-9b2c-46ac-bb2c-37ebe1432169",
"created_at": "2026-08-09 02:46:52.050450+00:00",
"db_md5": "fc75e306c93179e5ca6dec0121d99fcc",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "8519b829-d866-4604-b2a5-36f0967c581c",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.9225667327776685,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4389336109161377,
"example_count": 60... |
357cd01e-1b5c-4868-af9a-a418a09e6c81 | easy | DDanlov | 2026-08-12 21:12:36.990395+00:00 | failed | 87bfc5acf173028e021ff8092a7f4be99862be0ee1b48b0c0e044c5b8ee5dd2a | 25,445 | null | from __future__ import annotations
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
try:
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
except ... | {
"id": "357cd01e-1b5c-4868-af9a-a418a09e6c81",
"created_at": "2026-08-12 21:12:36.990395+00:00",
"db_md5": "91e418234ea1c9741ea6fa4abe658d34",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "b95fb2ed-cc60-4de4-8ccc-9fbcc4dfc091",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
357dc41a-5bd3-46fe-b156-634fa1681f99 | easy | Ranoobaba | 2026-08-24 02:06:11.127051+00:00 | succeeded | f0e3e99775298be1bb90ea41c475f64588472b94992a771f9d8775ea92a06abc | 7,813 | 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": "357dc41a-5bd3-46fe-b156-634fa1681f99",
"created_at": "2026-08-24 02:06:11.127051+00:00",
"db_md5": "733097a163e399a5c441da55453e58d0",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "9ba9aaf8-f2c2-42a0-97d0-d644e935e859",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 1.977348192671588,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666554907959
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9642366170883179,
"example_count": 100... |
35807940-98ec-4c9c-bea6-f8fa6cb55e8c | easy | 0Chris5R | 2026-08-08 09:12:47.880941+00:00 | succeeded | 69de3cc7622cb8ba3b3e21fc4560e48ba8dfcfc2969985aa13a1e9a39354f79c | 22,931 | null | """Shared-digit Neural GPU with a compute-matched operator curriculum.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the ... | {
"id": "35807940-98ec-4c9c-bea6-f8fa6cb55e8c",
"created_at": "2026-08-08 09:12:47.880941+00:00",
"db_md5": "c28de8efe3e8978ca7ee479d408f70a7",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "abb018c9-def8-42cb-bfde-ae294dc12758",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.874349117874726,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04291666670391957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.798029784129874,
"example_count": 600,
... |
3583fda4-741e-4217-8805-94fcd0b35d8d | easy | shreyash-chonkie | 2026-08-26 11:59:14.170131+00:00 | succeeded | 45f7bef2c8ed84bc01101b24eae132a0c40281f3fc0992f4c7119404f4be8b09 | 13,895 | null | """Compiler-emitted soft addresses executed by an executor-owned interpreter."""
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_st... | {
"id": "3583fda4-741e-4217-8805-94fcd0b35d8d",
"created_at": "2026-08-26 11:59:14.170131+00:00",
"db_md5": "9fb023d4cd6882bda7d373cda256cd2c",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "0d5cb433-1d72-4223-a7de-5244b2ff6509",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 3.7343000173568726,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04333333298563957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.690607786178589,
"example_count": 100,... |
3586ff10-75cd-4e01-96e2-2b4da453011d | easy | erdavis0 | 2026-08-24 12:48:18.296200+00:00 | succeeded | 9229eee5ea80a887dc8b3571f5a1e135b489a11fbe792f049ffca8d8cd05da45 | 26,559 | 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": "3586ff10-75cd-4e01-96e2-2b4da453011d",
"created_at": "2026-08-24 12:48:18.296200+00:00",
"db_md5": "f76733d8084fd6c5a4c949912cf678b8",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "0dfdcb42-2ad2-4b25-8dab-0a3b6519ff36",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",... | {
"score": {
"mean_loss": 6.711994739363796,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.020625000065192578
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.737563373564069,
"example_count": 800,... |
358bd9bd-f5f9-4224-9345-11126cf0d2df | easy | oupadhyay | 2026-08-29 18:28:47.526674+00:00 | succeeded | 26c7ce4cd082b6cd8943b311d55f890626e97ff14e03d73fb50c516397dfcb10 | 3,993 | null | """Conditional optimizer phase 2: lr10_wd0."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_s... | {
"id": "358bd9bd-f5f9-4224-9345-11126cf0d2df",
"created_at": "2026-08-29 18:28:47.526674+00:00",
"db_md5": "1b21538318993fe935eaf9f8d26646db",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "a83cfa23-4f06-4728-8a7b-ed246096201b",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 1.9696871638298035,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03018018137663603
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.943730115890503,
"example_count": 60,
... |
358c233a-a0f0-4f03-ad4f-13a31570f741 | easy | chad-atexpedient | 2026-08-10 02:11:16.721906+00:00 | succeeded | 1962885169b0df037aec991a2603b4f9015c833ef624db01019dd64a6af9118c | 10,938 | null | """BL8A: width-32 final-readout control."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
LANES = 6
IMMUTABLE_LAN... | {
"id": "358c233a-a0f0-4f03-ad4f-13a31570f741",
"created_at": "2026-08-10 02:11:16.721906+00:00",
"db_md5": "3934aca23e32dcc233c608e888578329",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "b1242da6-07d3-4752-a0b6-86c8c6c3ac67",
"tier": "easy",
"dataset_id": "e4",
"st... | {
"score": {
"mean_loss": 3.884224859456321,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.8680432374404923,
"example_count": 1200,
... |
3597c4d9-fd5b-43af-beed-d383bd4b790a | easy | DDanlov | 2026-08-10 08:13:25.709916+00:00 | succeeded | 3b387d3fa859002269d7f2be12e4fa2466a81a8743cbcc103fb8e0deca7421de | 16,673 | 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
def init_cutting_the_skip(module):
"""
SVD-based initialization to ensure initial Jaco... | {
"id": "3597c4d9-fd5b-43af-beed-d383bd4b790a",
"created_at": "2026-08-10 08:13:25.709916+00:00",
"db_md5": "96bc5d9feac205c8b2b421842c64131b",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "9faf2585-344e-4ba8-98e4-7b09c8df45d1",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 4.647991126909518,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.3502197265625,
"example_count": 100,
... |
359a2c3b-dcdb-4762-a5e5-b74427b010b5 | easy | repst | 2026-08-05 20:52:02.588598+00:00 | succeeded | f3bc52c851886f7d46f8c9ad3e0c08c16d70c4ce67d552c7712d6002f62f848d | 6,270 | null | """Tied-recurrent Transformer with a 600-step early-stop checkpoint."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_stat... | {
"id": "359a2c3b-dcdb-4762-a5e5-b74427b010b5",
"created_at": "2026-08-05 20:52:02.588598+00:00",
"db_md5": "4c0e38ae0a5aa51d99f5b6e60a1ae842",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "52c9578a-aaf5-4cf1-9058-2f4f724a6659",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.2672034850394405,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0058333332774539795
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3066439628601074,
"example_count": 6... |
35a24c7f-cb4f-47d9-a0fc-5d3fcb510191 | easy | erdavis0 | 2026-08-29 01:02:39.506476+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": "35a24c7f-cb4f-47d9-a0fc-5d3fcb510191",
"created_at": "2026-08-29 01:02:39.506476+00:00",
"db_md5": "dac67f1d3596515bff6b89e826ff98e0",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "d625f4a0-0a87-4af7-88fd-36141921f363",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 0.326664287596941,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9716666638851166
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.11813978105783463,
"example_count": 100,... |
35a3c7db-8431-476d-935c-2da625be3e20 | easy | DDanlov | 2026-08-28 22:13:04.270841+00:00 | succeeded | 20c68b785bfdf2f8c6080c358242901fac9ff86eee37116fa51c80ba6e45e179 | 15,434 | null | """
Trial 37: Model D: D=256, H=4, M=16, u=2 (T=32)
"""
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": "35a3c7db-8431-476d-935c-2da625be3e20",
"created_at": "2026-08-28 22:13:04.270841+00:00",
"db_md5": "60f64b9d8a3c86bf7b4f0ddc64c28758",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "c108f8cd-7240-409e-9c49-91b95c35420e",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.6017235259705362,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.620176421687207,
"example_count": 600,
"exa... |
35b21254-8b1f-4151-88cb-8b864250fdb1 | easy | DDanlov | 2026-08-19 19:43:46.422343+00:00 | succeeded | ccdb4e4972cc692c6f9bd357841b44dadb43fbc8b1b811f4362ced635de8997a | 17,727 | null | """
Dynamic Submission for One Layer Deeper API Benchmark
Config: dim=1024, num_heads=32, d_ff_mult=4, fixed_trec=112
"""
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": "35b21254-8b1f-4151-88cb-8b864250fdb1",
"created_at": "2026-08-19 19:43:46.422343+00:00",
"db_md5": "db3941d3b869809da07df6368b37256b",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "ff987d97-9230-4274-b332-c089043e50c0",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.364289009373892,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000024835268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.427680112053995,
"example_count": 600,... |
35b37a27-3147-4aec-b283-6528a942dd5a | easy | adamsoliev | 2026-08-01 22:50:10.691682+00:00 | succeeded | 4e5e218ebc7924a0b4573afddcc33a08452821c3b304702453f04e37d0f71054 | 35,723 | null | """Dual-factor semigroup-block automaton for One Layer Deeper."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
from torch import Tensor, nn
WIDTH = ... | {
"id": "35b37a27-3147-4aec-b283-6528a942dd5a",
"created_at": "2026-08-01 22:50:10.691682+00:00",
"db_md5": "6522259187ade0b39126e1487cbbcf74",
"submitter": "Adam Soliev",
"github_login": "adamsoliev",
"run_id": "a46fe580-f620-40bb-906a-2a6019efd047",
"tier": "easy",
"dataset_id": "e3",
"status": "suc... | {
"score": {
"mean_loss": 2.249503288140835e-05,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2440531446345168e-05,
"example_count": 800,
... |
35b4fe2e-67ca-4b16-8396-b8baa839ef59 | easy | DDanlov | 2026-08-11 19:35:41.197622+00:00 | succeeded | 9b459e5a4067369a3b4ef937d7dcc5770af4f7304dda3b2b9105fb4a9fca2cd0 | 20,034 | 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
def init_cutting_the_skip(module):
"""SVD-based orthogonal initialization to ensure initia... | {
"id": "35b4fe2e-67ca-4b16-8396-b8baa839ef59",
"created_at": "2026-08-11 19:35:41.197622+00:00",
"db_md5": "f056b0d2509ae968c2479cd4a4a8c63b",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "245797be-2270-4a14-a13c-b07cf4436324",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.782462947089178,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333221574625
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.705864429473877,
"example_count": 100,... |
35baef03-e7f0-4a08-aa5b-27103031a827 | easy | chad-atexpedient | 2026-08-20 16:07:55.508804+00:00 | succeeded | 70235ed99bd1bcf028f58cfd3b3f1e98fee7ab2e358cf06156ab46dcf7bff5e6 | 22,735 | null | """X121: training-only modulus digit embedding dropout.
X120 showed that additional modulus identity raises seen-N while erasing OOD-N.
X121 tests the inverse: during training only, independently replace ten percent of
N digit token vectors with their mean embedding, forcing contextual reconstruction
and discouraging ... | {
"id": "35baef03-e7f0-4a08-aa5b-27103031a827",
"created_at": "2026-08-20 16:07:55.508804+00:00",
"db_md5": "ca52131dd695fdff6b36d72b9e37e6bb",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "d94e90b0-f9d6-42a2-80c7-04376a077c71",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.3910567904798636,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.45564562002105,
"example_count": 600,... |
35bfa77b-4b5d-4935-9426-dd474dc6182f | easy | sapient-sapiens | 2026-08-13 19:52:26.236633+00:00 | succeeded | 7c591b2007c16b2ea44eca5a61f2181b4c7aaeee888f5d71a9b79a03b6f9c724 | 8,946 | null | """E3 fixed-T=2: differentiable, field-aligned digit-state recurrence.
The model reads marker positions to extract and right-align fields, but never
converts digit tokens into numbers. A learned four-block operator is applied
twice. The first application's categorical distribution is softly re-embedded
so final-answ... | {
"id": "35bfa77b-4b5d-4935-9426-dd474dc6182f",
"created_at": "2026-08-13 19:52:26.236633+00:00",
"db_md5": "a3d07634e05090ad5315b98fc943f30a",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "c1a076ab-98cd-46da-b69e-bba7c19d6025",
"tier": "easy",
"dataset_id": "e3",
"status": "su... | {
"score": {
"mean_loss": 2.1308581661167043,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.138787930183266,
"example_count": 800,
"... |
35c28133-b805-47e0-94df-db43942db4f2 | easy | sapient-sapiens | 2026-08-16 18:15:14.693734+00:00 | succeeded | f4bebea918730523d3db9c91121f53c85890725956d9f7e1c77c72da04bd65c5 | 21,530 | null | """E6 WD ablation: frozen v55 with WEIGHT_DECAY=0.
The only change relative to act64_e6_w1024_dh4_L6_v55 is turning off
weight decay. Residual dropout stays 0. T is diagnostic-only.
"""
from __future__ import annotations
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, ... | {
"id": "35c28133-b805-47e0-94df-db43942db4f2",
"created_at": "2026-08-16 18:15:14.693734+00:00",
"db_md5": "eed83005b1e7eb792fb997292c28927b",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "572a4017-e03d-4064-a5f6-26d0503c3655",
"tier": "easy",
"dataset_id": "e6",
"status": "su... | {
"score": {
"mean_loss": 6.447892698523116,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.09864865167922265
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.637289047241211,
"example_count": 60,
... |
35c68216-6641-45f1-ba29-68d2d986a42c | easy | sapient-sapiens | 2026-08-14 18:22:20.117903+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": "35c68216-6641-45f1-ba29-68d2d986a42c",
"created_at": "2026-08-14 18:22:20.117903+00:00",
"db_md5": "1efadda9e01ae84ec2857432a666c542",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "acc0c2ac-fbda-4dc1-b1da-8b02a6e9da5b",
"tier": "easy",
"dataset_id": "e4",
"status": "su... | {
"score": {
"mean_loss": 2.061752355006632,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008229166666666668
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0657801847880792,
"example_count": 120... |
35cb6610-7be5-42e2-b990-5bea56b42560 | easy | sapient-sapiens | 2026-08-19 19:32:19.094734+00:00 | succeeded | daa4fc1f1821b95b049e62d1602f10121592d66b9efbbb3e78e141bca4403b2b | 33,252 | null | """Depth-quantized recurrence with a mixture-over-depth objective.
The recurrent state is a bank of right-aligned digit slots. Every operator
application is followed by a soft quantization back onto the token simplex, and
the digit logits produced by that quantization are the answer logits at that
depth. There is no s... | {
"id": "35cb6610-7be5-42e2-b990-5bea56b42560",
"created_at": "2026-08-19 19:32:19.094734+00:00",
"db_md5": "2553f964fd1803486430c2690edf5c49",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "afee60e1-94fd-414b-b69e-fdaee2c01764",
"tier": "easy",
"dataset_id": "e8",
"status": "su... | {
"score": {
"mean_loss": 4.827942908756317,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08533868114605095
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.673460960388184,
"example_count": 85,
... |
35cd5afd-280e-481d-9b54-c85ab0f4ac94 | easy | TZZheng | 2026-08-07 00:01:39.794730+00:00 | succeeded | b8301dddda760e50719bf9604f3c986ebab327ae9bb1c3e9bbcca5a5b32c39f5 | 7,342 | null | """Self-contained standard Transformer scaling-sweep submission."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
VARIANT = {'block... | {
"id": "35cd5afd-280e-481d-9b54-c85ab0f4ac94",
"created_at": "2026-08-07 00:01:39.794730+00:00",
"db_md5": "16ec378d6d82482653b34c09d2557d3f",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "a4da283d-f3dc-4f97-a259-dbb449df0d74",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.1703339219093323,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.061666667461395264
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.424264669418335,
"example_count": 100... |
35decc87-86c8-4b63-802f-c881e6607709 | easy | AFeinfield | 2026-08-18 23:22:59.193258+00:00 | succeeded | 765132c3584a008dfe8bb3b7f798f1040fcafd742b77196742afa8d7cd40f1b0 | 17,117 | null | """081: 077 minus the carry GRU sublayer.
The GRU is the most optimization-hostile component in the block
(sequential, gated). Ablation asks what it costs in fit speed and
whether the carry mechanism is load-bearing for generalization.
Original 077 notes follow.
077: ACTIVE-ROW trajectory loop (compute optimization of... | {
"id": "35decc87-86c8-4b63-802f-c881e6607709",
"created_at": "2026-08-18 23:22:59.193258+00:00",
"db_md5": "fe5968d4746a53e2cddf666951850014",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "2ed66995-e51b-4de3-8d39-52e95d591b54",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 3.6311434507369995,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666772216558
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.906703233718872,
"example_count": 300... |
35e28957-a89f-45a3-85fd-cbf31abf1a11 | easy | Alan-Qin | 2026-08-26 18:44:43.180959+00:00 | succeeded | 10a1368f86d9c13a38fa64de93029aac846af4571b91dd2c8aeed26cac66b886 | 26,476 | null | """v91h17s = v91h17 + file-hash init seeding (m6 full-ladder lottery tickets).
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-quoti... | {
"id": "35e28957-a89f-45a3-85fd-cbf31abf1a11",
"created_at": "2026-08-26 18:44:43.180959+00:00",
"db_md5": "19a5c64917593459d6270900a45835e1",
"submitter": "Zeyu Qin",
"github_login": "Alan-Qin",
"run_id": "ae5c4870-5af2-47a0-8347-bc05c3e33114",
"tier": "easy",
"dataset_id": "e6",
"status": "succeede... | {
"score": {
"mean_loss": 0.05862808786332607,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.06470267474651337,
"example_count": 60,
"ex... |
35e6c24d-d8b8-4bae-9c9d-ab6a3c24aaf5 | easy | liam-gb | 2026-08-16 19:46:28.857591+00:00 | succeeded | 08861a83832278cc0578fac15abf7aa0b72524ff644357684630ce2e20c6c3db | 15,720 | null | """rns_pm_gather_ct_widet_l32nore: norelrt with LOOPS = 32.
SINGLE CHANGE vs rns_pm_gather_ct_widet_norelrt: LOOPS 8 -> 32. A counted
loop capped at 8 cannot certify past T=8 by construction while the ladder
runs to 64; this is the corrected-recipe body that can. The loop trim keeps
the extra depth free at train time ... | {
"id": "35e6c24d-d8b8-4bae-9c9d-ab6a3c24aaf5",
"created_at": "2026-08-16 19:46:28.857591+00:00",
"db_md5": "c5770d075888ed9d73d2bd866ce45fca",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "a2c9d27e-9c72-47a9-9312-47388f8fa772",
"tier": "easy",
"dataset_id": "e7",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.1692281365394592,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4065062440931797
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.078921794891357,
"example_count": 85,
... |
35fef872-ee6e-448c-9eb4-3734e89ca4ec | easy | khushidahi | 2026-08-15 22:50:08.206664+00:00 | succeeded | 3a94ccd0603b3628f3eebcfbdf5a09fe3034ca7a54f58f2811327bf50ad0c5b8 | 22,153 | null | """R76: composition-trained Neural GPU with a well-defined state tape.
This candidate is deliberately diagnostic. It trains on evaluator-owned final
labels at several depths so that one shared transition must compose. The model
never converts N or X to native numbers and never implements multiplication or
modular re... | {
"id": "35fef872-ee6e-448c-9eb4-3734e89ca4ec",
"created_at": "2026-08-15 22:50:08.206664+00:00",
"db_md5": "97492f9b654d0e596fc23d53a7db8728",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "c97ade31-3743-4d8d-8a7d-3f58288acd4d",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 0.7058894783258438,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.8450450897216797
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.3501425087451935,
"example_count": 60,
... |
3601ce66-3421-4a67-ab3d-6c843e799e16 | easy | isaac0804 | 2026-08-05 08:29:22.051499+00:00 | succeeded | 0ab40eb7cfa42b9700f7e4a4f170eba07ae49a93a935f124b68d54c84d706635 | 9,365 | null | """Weight-tied looped Transformer (v6 core: damped gate, 2 sublayers/iteration)
plus explicit place-value embeddings for N's and x's digit tokens.
Motivation (see docs/findings.md, N-conditioning section): every architecture
tried so far treats N's (and x's) digits as an opaque span for generic
attention to notice imp... | {
"id": "3601ce66-3421-4a67-ab3d-6c843e799e16",
"created_at": "2026-08-05 08:29:22.051499+00:00",
"db_md5": "6c332bc632593f83d44af5125cddee51",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "b9279107-a7f0-4c43-b5e8-36ef898881c1",
"tier": "easy",
"dataset_id": "e4",
"status": "succe... | {
"score": {
"mean_loss": 2.1457029045729077,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006979166679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2079818927059733,
"example_count": 12... |
3602fcbb-09fb-4e92-ba33-58625e4739c2 | easy | 0Chris5R | 2026-08-07 09:16:14.797934+00:00 | succeeded | bdd6a030ca5d54a82fb719e60ce0ce1e6526144a24278a4de9abf6b7f82c2acf | 31,952 | 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": "3602fcbb-09fb-4e92-ba33-58625e4739c2",
"created_at": "2026-08-07 09:16:14.797934+00:00",
"db_md5": "ce3fced9241df33cba6b46cc34b68335",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "9db7b91a-d30f-44f8-aedd-af823fe8f128",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.487631545216959,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.015833333333333335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.178269800049307,
"example_count": 600,... |
36037248-92d4-42f9-bcc3-b9c0742d6d9c | easy | pepinu | 2026-08-21 22:56:48.389424+00:00 | succeeded | 613f7882ef36fc8855ca6a5351506070fef00c191bdfe2d75e5077504d9e3e44 | 41,384 | 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": "36037248-92d4-42f9-bcc3-b9c0742d6d9c",
"created_at": "2026-08-21 22:56:48.389424+00:00",
"db_md5": "0cc26802042f1a1f77154f64ae8140f2",
"submitter": "pepinu",
"github_login": "pepinu",
"run_id": "56d524ae-b642-424d-bf94-a5c399db1228",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
... | {
"score": {
"mean_loss": 0.12113953991410419,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9937500059604645
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.2270158921463629,
"example_count": 800... |
3603976b-35d8-4787-9787-eeb01782554a | easy | EyimofeA | 2026-08-16 23:46:53.109097+00:00 | succeeded | 69a2be7d91d3dafba20144a3db0281b4d61b0933bd10d303faa8b70d10cbb4c8 | 10,906 | null | """Binary fused work-state recurrence: width 256 + tuned flattened Muon.
The model receives exact binary representations of evaluator-provided x and N,
then learns the complete transition from final labels only. It has no explicit
squarer, reducer, arithmetic trace, or task oracle.
"""
from __future__ import annotati... | {
"id": "3603976b-35d8-4787-9787-eeb01782554a",
"created_at": "2026-08-16 23:46:53.109097+00:00",
"db_md5": "952b18f2cb73b7293c0fff54bb5ffeb3",
"submitter": "mof",
"github_login": "EyimofeA",
"run_id": "8114c6da-4098-4629-a854-4c26e73deb7e",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 36.257373305177204,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0016666666666666668
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 36.726458733391866,
"example_count": 6... |
36082eb4-49e9-4b96-9c0b-a4f70fdd2923 | easy | jordanrubin | 2026-08-07 06:06:42.104537+00:00 | succeeded | 6f17eb27b2bb1fbb5891f0392535694f326c447b00cd9f37136d01d78d2e30ec | 16,908 | 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": "36082eb4-49e9-4b96-9c0b-a4f70fdd2923",
"created_at": "2026-08-07 06:06:42.104537+00:00",
"db_md5": "378594dd4459a9c7efa73ec8d9805eae",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "7d4ad0f2-abac-4e4f-a21a-159f11c1eea4",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 4.2276107252380974,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333147068819
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.2414662987662,
"example_count": 600,
... |
3608a456-cbed-47bd-99c2-2113e02dc214 | easy | velocizapkar | 2026-08-16 10:22:20.288831+00:00 | succeeded | e57e8abe8734e266a4320aa2c05e320baa90237312391d13e42d2cfcd2c6b8bc | 15,781 | null | """Fixed-width right-aligned decimal-field Transformer.
The public prompt consists of three marker-delimited decimal fields. Every
contiguous digit run is placed in the same fixed-width field tape with its
least-significant digit at the right edge. This is deterministic input
padding/positioning only: all content fe... | {
"id": "3608a456-cbed-47bd-99c2-2113e02dc214",
"created_at": "2026-08-16 10:22:20.288831+00:00",
"db_md5": "10bda0239a67df8598bb3a3305698611",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "a5e7a344-088a-424c-a261-20e269bc671c",
"tier": "easy",
"dataset_id": "e6",
"statu... | {
"score": {
"mean_loss": 2.073455572128296,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05180180352181196
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9735136032104492,
"example_count": 60,
... |
36112399-0246-4093-a5f8-608a919b664a | easy | deepakvijaykee | 2026-08-10 14:25:13.580868+00:00 | succeeded | 6c446d398bd406c2cd4deb3a533fc3bd8bc18d41dfadbb5694de90c02df39617 | 21,083 | null | """One Layer Deeper candidate: tied digit-tape recurrence + verified deep supervision.
DESIGN
The task is T applications of an unknown map on Z_N, read and written as
decimal digit tokens. Five stages with explicit contracts; every free choice
lives in ExperimentConfig with its rationale.
parse input_ids -> Par... | {
"id": "36112399-0246-4093-a5f8-608a919b664a",
"created_at": "2026-08-10 14:25:13.580868+00:00",
"db_md5": "ba693bc578f8157dbd2bc2988753775d",
"submitter": "Deepak Vijaykeerthy",
"github_login": "deepakvijaykee",
"run_id": "5d795f22-a8f6-4e64-b07b-62b69fa1de83",
"tier": "easy",
"dataset_id": "e3",
"s... | {
"score": {
"mean_loss": 2.1948812649505864,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0062500000558793544
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.189087501633403,
"example_count": 80... |
36128a6c-22fa-41b8-b36c-ce65ce87ed27 | easy | velocizapkar | 2026-08-06 22:18:56.981118+00:00 | succeeded | 7c30e5a31446543f6df260ee653c7e53ed3b5c5446000d863d94f7eee3563207 | 55,676 | null | """Fast packed residual-free Attention-Residual transformer with Muon.
The model is a generic learned sequence model. It contains no parser,
arithmetic primitive, task constant, generated data, label feedback, or answer
construction. Every macro-state is produced by a serial pack of attention and
MLP transformations... | {
"id": "36128a6c-22fa-41b8-b36c-ce65ce87ed27",
"created_at": "2026-08-06 22:18:56.981118+00:00",
"db_md5": "7c66af1511a277bc24bc0fa2577307b6",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "c3f6fb2e-84f7-4394-8b46-515f98d7afeb",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 3.201090931892395,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.016874999273568392
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1558454036712646,
"example_count": 800... |
36166b28-993b-4773-a338-70e2134cc066 | easy | liam-gb | 2026-08-17 09:34:16.229196+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": "36166b28-993b-4773-a338-70e2134cc066",
"created_at": "2026-08-17 09:34:16.229196+00:00",
"db_md5": "46c5decfff7d97a850fb297981e9d310",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "0a06205c-a9a1-4de2-ba61-c6fbd535dc01",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded"... | {
"score": {
"mean_loss": 0.6712471842765808,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.8527027368545532
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.9732085466384888,
"example_count": 60,
... |
361dd3f9-2cca-495d-a5af-3972a4c40a34 | easy | viridale | 2026-08-05 01:00:17.978488+00:00 | succeeded | be1529b0fd9deeaa40bf8c215c6153b9fd84cfd7c44cc581b661efb5fa67f2df | 9,059 | null | """Clean fixed-runtime adaptive latent-depth recurrence.
hypothesis: the same fixed terminal recurrent state is wrong for prompts needing
different amounts of learned computation; a differentiable hazard mixture over
shared states can adapt representation depth while executing every step.
axis: arch
target: of... | {
"id": "361dd3f9-2cca-495d-a5af-3972a4c40a34",
"created_at": "2026-08-05 01:00:17.978488+00:00",
"db_md5": "c003df09a9522e4ffb3947542c0cd88b",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "d200de82-ed34-4fa5-b886-6f2d4e24c2f2",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.8589626550674438,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666716337204
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8169450759887695,
"example_count": 10... |
361f49e5-ca90-4a2a-99e6-53913e497ac0 | easy | karanganesan | 2026-08-07 08:56:14.390866+00:00 | succeeded | 78d822b7d1f9bb5b46f977e12ceb542f196ddd50f2f7a90f51b4f9012f406971 | 25,547 | 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": "361f49e5-ca90-4a2a-99e6-53913e497ac0",
"created_at": "2026-08-07 08:56:14.390866+00:00",
"db_md5": "c6174b599773d987b1c6e74c1c2f0956",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "e50bbf2c-a77c-4e23-8c22-2c334d4abc0c",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.2813035249710083,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000121071935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3114748001098633,
"example_count": 60... |
36264ea3-9f07-4009-8c53-c50fb92f4497 | easy | sapient-sapiens | 2026-08-29 21:40:11.713456+00:00 | failed | 25f7ca63235eae30e0f59dd0ab7dfa8c615607e7c0ce545e33d26d544a01dc86 | 15,772 | null | """Three-loop one-layer Transformer for iterated modular squaring.
The learned arithmetic operator is one tied pre-norm Transformer block. Every
square makes four Horner-digit calls and every digit makes nine micro-calls.
N and the complete operand are immutable during a square; scratch resets at a
digit boundary; an... | {
"id": "36264ea3-9f07-4009-8c53-c50fb92f4497",
"created_at": "2026-08-29 21:40:11.713456+00:00",
"db_md5": "5d4fec0dd4f64fba5eff27d230c981de",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "f9298c3e-92cf-4ea6-9145-bd14e60254e0",
"tier": "easy",
"dataset_id": "e9",
"status": "fa... | null
|
362930ea-a19e-4874-8804-08da84a45df8 | easy | gauravmishra | 2026-08-09 04:56:42.220395+00:00 | succeeded | 683a05c8a1ac6d366537e80f594bd115ba87b23b1c3a863d84971818185a8ed8 | 5,040 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
D_MODEL = ... | {
"id": "362930ea-a19e-4874-8804-08da84a45df8",
"created_at": "2026-08-09 04:56:42.220395+00:00",
"db_md5": "e3d5156a58ee304d8694783f31aacf87",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "091350b8-7013-4d9c-877f-eb17951d57c2",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.9265595478496444,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1327817151364723,
"example_count": 60... |
362cf20d-2332-4299-bdff-82cc1dc8c25f | easy | ipsa-loquitur | 2026-08-26 04:57:16.737580+00:00 | succeeded | 09ffd239340965ea76d9bb2854f377748262776b557b07b90a97b4ae2ea2bc63 | 15,929 | null | """Standalone NetF-V1 submission candidate for One Layer Deeper.
The learned rule is shared across bit positions, modular-square iterations,
moduli, and tape widths. During training, exact double-and-add states provide
dense one-step supervision. During evaluation, the model receives no state
corrections and rolls out... | {
"id": "362cf20d-2332-4299-bdff-82cc1dc8c25f",
"created_at": "2026-08-26 04:57:16.737580+00:00",
"db_md5": "2a8d287e6210327252633797ebc0f1f3",
"submitter": "ipsa-loquitur",
"github_login": "ipsa-loquitur",
"run_id": "e48a2cb5-ea86-496f-9b81-a5f8739547a6",
"tier": "easy",
"dataset_id": "e4",
"status":... | {
"score": {
"mean_loss": 16.29397181107096,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.2085416669646899
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 18.55306406738631,
"example_count": 1200,
... |
362cf289-dfd5-4a5b-aa19-a1087ed317b4 | easy | 0Chris5R | 2026-08-07 12:48:36.233241+00:00 | succeeded | f3df2e00f64862186bcf04ec3fa53d9e674c9e9557be1a6ca7e4527cdbee31d5 | 21,318 | null | """Shared-digit Neural GPU with a compute-matched operator curriculum.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the ... | {
"id": "362cf289-dfd5-4a5b-aa19-a1087ed317b4",
"created_at": "2026-08-07 12:48:36.233241+00:00",
"db_md5": "ca53940ce894f77658ef816c175cbbcc",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "e757891c-615e-4257-8e57-03ad4dc495cf",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.8240439401546635,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03375000002483527
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7650073739979835,
"example_count": 600... |
3630aa50-a300-45b9-9a94-40c855af1f76 | easy | khushidahi | 2026-08-21 22:23:28.802923+00:00 | succeeded | ff854d6035e34ec1bff62a75b5ac36cec5e8efb137c094cbfbb91d002010b4fa | 29,467 | null | """R389: categorical pair workspace plus late shallow-root polishing.
The first 75% of the clock fits the supplied endpoint distribution. AdamW
state is then cleared and the final 25% emphasizes short X fields, which are
the least ambiguous final-label evidence for one application of the shared
transition. This is the... | {
"id": "3630aa50-a300-45b9-9a94-40c855af1f76",
"created_at": "2026-08-21 22:23:28.802923+00:00",
"db_md5": "a2c8473401362aeed2ed2246afc8ab3c",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "cd6439ce-0bc8-416a-b064-634b94516053",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 3.3085371149036282,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000055879355
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.3136903222589167,
"example_count": 80... |
363254b1-8c9e-4c0b-829b-cece78820f5a | easy | yashkant | 2026-08-09 02:36:39.285756+00:00 | succeeded | 8934eafe72f1ad37979c30151536931be76393afcb74b62af7fa0a24dcee7cf1 | 22,835 | 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": "363254b1-8c9e-4c0b-829b-cece78820f5a",
"created_at": "2026-08-09 02:36:39.285756+00:00",
"db_md5": "0a8833dab91885b9b75504da4e1e6537",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "b877e818-2cda-4ec7-b4fd-373e90fbb25a",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 2.119230038380499,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009375000018626452
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.117102818073823,
"example_count": 800,... |
3640fb17-bc53-41ed-be52-b357e38793da | easy | DDanlov | 2026-08-11 22:19:03.709125+00:00 | succeeded | 2992a0e866968efa8f82065c1b5d3f545d008ada27d800a8213c4cfc5796fd2d | 20,572 | 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
def init_cutting_the_skip(module):
"""SVD-based orthogonal initialization to ensure initia... | {
"id": "3640fb17-bc53-41ed-be52-b357e38793da",
"created_at": "2026-08-11 22:19:03.709125+00:00",
"db_md5": "10a4f6d9df659bb4ae6e721697941f15",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "07db2efb-d509-4c2d-9980-40939ba6ca4d",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2487.5836987721773,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2547.9472422236286,
"example_count": 600,... |
36466b76-f771-4b8f-93ee-274ffb3a308c | easy | khushidahi | 2026-08-23 18:10:01.344060+00:00 | succeeded | c626a1e1a705408f1bab369561566a27864019d22128d56694f704bce72ca2a6 | 11,941 | null | """ART-DT recall candidate.
A learned recurrent sequence model with right-to-left Abacus positions and
global-sequence rotary attention. This variant changes only optimizer LR and
warmup schedule from its frozen architectural parent. One immutable prompt embedding is recalled at the entry to each recurrent visit of
... | {
"id": "36466b76-f771-4b8f-93ee-274ffb3a308c",
"created_at": "2026-08-23 18:10:01.344060+00:00",
"db_md5": "2f3b053f5d13b29c433fa41642ef4eb7",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "1a47720b-8950-4876-ae38-ca841cb74da2",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 2.684691905975342,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.043468470685184
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6542866230010986,
"example_count": 60,
... |
36497520-7fcc-4188-b0e2-5760e8acddd6 | easy | christopherhwood | 2026-08-08 20:47:56.229661+00:00 | succeeded | 87f8beed1c44ae168c30c41e1745c1adec02840b5dbc280aaa93c8b0a82f7636 | 12,015 | null | """Endpoint-trained posterior over a generic integer accumulator ISA.
Five instruction slots select from sixteen ordinary calculator operations.
Modulo is an explicit learned instruction; the architecture does not reduce
automatically after arithmetic and does not encode a recurrence formula.
One factorized global pos... | {
"id": "36497520-7fcc-4188-b0e2-5760e8acddd6",
"created_at": "2026-08-08 20:47:56.229661+00:00",
"db_md5": "973f8886e5e5559e3303150d118394f6",
"submitter": "Chris Wood",
"github_login": "christopherhwood",
"run_id": "3161d1d3-5582-4525-80e4-126478f89adc",
"tier": "easy",
"dataset_id": "e5",
"status":... | {
"score": {
"mean_loss": 1.8505724706819945e-06,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8502813487642183e-06,
"example_count": 600,
... |
36504a2e-cbfb-45b3-b0c3-fd5051eea323 | easy | TZZheng | 2026-08-06 17:44:58.086557+00:00 | succeeded | b8dc53f104d69604b36f42185ae8fbb02259589b2d9b67c3e1a132a77c5ab222 | 7,296 | null | """Self-contained standard Transformer scaling-sweep submission."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
VARIANT = {'block... | {
"id": "36504a2e-cbfb-45b3-b0c3-fd5051eea323",
"created_at": "2026-08-06 17:44:58.086557+00:00",
"db_md5": "fd4b1be0fc99f5163f2e42ebdf6cde77",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "e9d91f36-af82-4b5f-81ee-d95fc632ff2e",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.138053596019745,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.056666663847863674
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.354467391967773,
"example_count": 100,... |
3654a34b-a9b6-45a5-b7c2-a78d82b744e2 | easy | ehonig | 2026-08-20 04:06:25.820738+00:00 | succeeded | db163760775ef375a0d6714b307e93343411ad5bd0dcdb66c63af1b6ef9a0778 | 22,765 | 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": "3654a34b-a9b6-45a5-b7c2-a78d82b744e2",
"created_at": "2026-08-20 04:06:25.820738+00:00",
"db_md5": "17c3ac8dd759485979784e818e3823c7",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "33059159-6cf5-4d1e-8b47-6e833a700a75",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.8735913634300232,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666716337204
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8493784666061401,
"example_count": 10... |
3655c2a2-d103-4a08-a39d-73008077d335 | easy | rougee | 2026-08-22 08:36:53.228804+00:00 | succeeded | fd6ffece74563edd0931e55e61943ccf256247c2475f7c1d9109dc626b09cee1 | 11,527 | null | """Official Submittable Champion for One Layer Deeper (Input-Reinjected Cayley Recurrent Transformer).
Architecture:
- Input Reinjection: Injects the prompt context (N, X, T) into every step of the recurrent loop.
- Core: Cayley Orthogonal Recurrence (W = (I - S)(I + S)^-1) with Modulus-conditioned skew generators.
- ... | {
"id": "3655c2a2-d103-4a08-a39d-73008077d335",
"created_at": "2026-08-22 08:36:53.228804+00:00",
"db_md5": "46fdfffa231f0a38e58634496b34a924",
"submitter": "Rougee",
"github_login": "rougee",
"run_id": "eeecf8c1-8b58-4997-a2aa-6a02cb7d2627",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 6.235580921173096,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05499999783933163
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.24018669128418,
"example_count": 100,
... |
36579743-f96c-4674-ae77-b28a3bed8422 | easy | ahuhn | 2026-08-03 14:16:36.317198+00:00 | succeeded | a7d6749c00f7b4c169907a9c89556748833b610a8ac24e751274758d053417ce | 10,014 | null | """A compact, gated recurrent Transformer for One Layer Deeper.
The model keeps approximately the arithmetic cost of the first baseline while
reorganizing it into more recurrent depth. One encoder block reads the prompt,
then one shared transition block refines the state six times with learned input
reinjection and up... | {
"id": "36579743-f96c-4674-ae77-b28a3bed8422",
"created_at": "2026-08-03 14:16:36.317198+00:00",
"db_md5": "3cd563f55a046e0fb8a722e4615ba6f0",
"submitter": "Niki Huhn",
"github_login": "ahuhn",
"run_id": "e6491199-01d0-4579-8df4-54e5fecac5a4",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 8.323994636535645,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03666666615754366
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.903376579284668,
"example_count": 100,
... |
365c1ea0-0700-4974-ae67-933d64c4723b | easy | ehonig | 2026-08-14 00:51:01.708097+00:00 | succeeded | a10733f80d0103c5be14b8cf10185c026b723357de617c63865413855357537a | 48,079 | null | """DigitLoop — a place-value recurrent model for repeated modular squaring.
The task is ``y = x^(2^T) mod N`` with the answer read back tail aligned, one
decimal place value per token position from the right. A stack of attention and
MLP layers has no primitive for the two operations this needs — a partial-product
co... | {
"id": "365c1ea0-0700-4974-ae67-933d64c4723b",
"created_at": "2026-08-14 00:51:01.708097+00:00",
"db_md5": "e0533d47111a48836c3700b73c9bbf98",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "a53e632d-12c6-42e2-a129-30f35be55174",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 5.372967481613159,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.12333333492279053
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.872854709625244,
"example_count": 100,
... |
365c5815-cace-49d6-8829-606aee225ad2 | easy | KaustubhKumar05 | 2026-08-25 15:43:06.848832+00:00 | succeeded | 1c97e6ade7b7c4704016b0ee590b894d107a7dd895518e2d223d26813d2c9766 | 13,655 | null | """Neural-GPU-style conv GRU, v2s: v2 + speedrun pack.
- LR: warmup, hold max to 85% of wall-clock, steep late cliff (2026 schedule theory)
- identity-at-init: cell output projection zero-initialized
- batch reuse x6 (multi-pass theory)
Base (v2):
Changes over v1 (each backed by measured results in the literature):
... | {
"id": "365c5815-cace-49d6-8829-606aee225ad2",
"created_at": "2026-08-25 15:43:06.848832+00:00",
"db_md5": "f9a601f96038214772004fc5576b07aa",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "48ad0078-b575-4165-8fc0-5b76698519cc",
"tier": "easy",
"dataset_id": "e6",
"status": "succee... | {
"score": {
"mean_loss": 4.99031925201416,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.23896396905183792
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.550942897796631,
"example_count": 60,
... |
366169f1-4587-4456-8214-cfde4d70f279 | easy | sapient-sapiens | 2026-08-25 07:48:38.920171+00:00 | succeeded | 9a785ce04f9cd9a405cce09cf2a0185a062e3d9ceec717660b1421b5072e9350 | 18,821 | 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": "366169f1-4587-4456-8214-cfde4d70f279",
"created_at": "2026-08-25 07:48:38.920171+00:00",
"db_md5": "79f614a950b6f63dd9392dbbddb6f0c3",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "c9828b04-0212-4b12-83ac-81d0f129eb9c",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.3377886693523586,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00625
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.402839238868166,
"example_count": 600,
"... |
36667377-96e3-47a5-88f0-eb1f4bfb0d57 | easy | khushidahi | 2026-08-12 20:28:12.621627+00:00 | succeeded | c9f9b53cb30aa2425a60e728f98f25b21ba10904221fba77f8e0a48d68c255ef | 24,153 | null | """R53 learned digit-pair table with directional carry/reduction scans.
The model learns a full 10x10 pair embedding table from final-label training.
For the diagonal variants, pair feature (i,j) is routed to decimal place i+j.
No multiplication values, carry rules, modular-reduction rules, generated
examples, or inte... | {
"id": "36667377-96e3-47a5-88f0-eb1f4bfb0d57",
"created_at": "2026-08-12 20:28:12.621627+00:00",
"db_md5": "b0d6d53220d9550848a54ae1efc1352d",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "6c7ae685-af5b-40fc-9ab2-6971ba5a7c77",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 3.669165849685669,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010416666977107525
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.6630866527557373,
"example_count": 600... |
366750c2-8467-42ac-85ac-b7df3d48eff4 | easy | DDanlov | 2026-08-10 08:45:02.331617+00:00 | succeeded | 319c88ea8754dd23e1462e54838c25927a6f8aa7f4a011555b9941e9d6a991d8 | 16,674 | 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
def init_cutting_the_skip(module):
"""
SVD-based initialization to ensure initial Jaco... | {
"id": "366750c2-8467-42ac-85ac-b7df3d48eff4",
"created_at": "2026-08-10 08:45:02.331617+00:00",
"db_md5": "b9cb46cb09ab581b69c4387b27d30ef8",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "532313f1-47c2-40af-896a-1969080804fd",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.537806468354073,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002500000024835269
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.994523114954944,
"example_count": 600,... |
36682c7d-a10a-448a-8ea4-cd5abf226c39 | easy | rteehas | 2026-08-19 11:35:48.917680+00:00 | succeeded | eee3fad8c29f2b6a4a650814765829e8f3b149afc707f0bd418d8583dbd0785e | 53,669 | null | """Anonymous marginal-program submission.
Mechanism
---------
The model is a differentiable probabilistic program. Its forward pass
computes, exactly, the answer-token predictions of a small structured latent
hypothesis space:
(prompt segmentation = which token types head the three fields)
x (field-role permutat... | {
"id": "36682c7d-a10a-448a-8ea4-cd5abf226c39",
"created_at": "2026-08-19 11:35:48.917680+00:00",
"db_md5": "ff7cd2b619c0a06a243f35734a6884e2",
"submitter": "Ryan Teehan",
"github_login": "rteehas",
"run_id": "f830a71a-7885-4fcd-8f79-77b45f28128f",
"tier": "easy",
"dataset_id": "e4",
"status": "succee... | {
"score": {
"mean_loss": 6.5643357172534476,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0021875
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.530533807575851,
"example_count": 1200,
... |
36720cf4-bb23-418f-aedd-c82dd1a8dcc2 | easy | chuk-uzowihe | 2026-08-06 23:21:07.337525+00:00 | succeeded | 532a9f763d5487e4144483edb0001007c6df4fffbd30dac60a3ba733ef3c0d87 | 33,677 | null | """One Layer Deeper submission: FPRM over a digit lattice with a Fourier T latent.
Structure (v4)
--------------
1. N and x are *digit slots*: each digit is gathered into a dense per-number
grid (column j = place value 10^j, short rows left-padded with explicit
zero digits) and embedded as digit token + learned ... | {
"id": "36720cf4-bb23-418f-aedd-c82dd1a8dcc2",
"created_at": "2026-08-06 23:21:07.337525+00:00",
"db_md5": "6b7aafc82434c8051892bb1c92617b0b",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "b50197d2-373a-4995-9c29-1f905c200845",
"tier": "easy",
"dataset_id": "e4",
"status": "... | {
"score": {
"mean_loss": 6.754112842922966,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006562500000000001
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.7068537982708785,
"example_count": 120... |
36747d66-f59a-44de-8731-e282bb1e504d | easy | jordanrubin | 2026-08-12 05:39:38.688311+00:00 | succeeded | 6072ddd859baaaadd512d39619839cdbe7ff402f38c47bd8ba7941d112bebe6f | 24,918 | null | """Weight-tied MLP-cell loop over soft digit states (mlploop).
Cell = MLP over [expected digits, pairwise digit products, RNS residue
simplices (fixed differentiable mixing over Z_p), N digits]. The cell that
learns one-step modular squaring from direct pairs (52% unseen at 14k rows,
day-20 screen) inside the exact-T ... | {
"id": "36747d66-f59a-44de-8731-e282bb1e504d",
"created_at": "2026-08-12 05:39:38.688311+00:00",
"db_md5": "77ff777c577d8e7bf8ce9b05ac674fb7",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "1f97a97d-7638-460e-a589-94a07286cabd",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.803465703518777,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.028749998969336353
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.201620340480933,
"example_count": 600,... |
367992be-5072-4ccd-9714-52459b2e1a68 | easy | viridale | 2026-08-06 10:31:38.791929+00:00 | succeeded | 88d98712a2545db55770d97375c9e3a0b2374d9a80b80ef50dcff6fd2345a491 | 8,880 | null | """Decoupled content-position Multi-head Latent Attention recurrence.
hypothesis: ordinary attention entangles token content with absolute position before
transport; a shared per-token content K/V bottleneck plus a separate relative
rotary channel should encourage reusable content circuits without losing geome... | {
"id": "367992be-5072-4ccd-9714-52459b2e1a68",
"created_at": "2026-08-06 10:31:38.791929+00:00",
"db_md5": "ba217c34f6e84f005a041bace433515f",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "b1252a27-52ff-437f-a7c7-0bbb4eebc81d",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 2.1832209825515747,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.180426836013794,
"example_count": 100,
... |
368cbacd-0fd0-4b63-bf40-054fd2712b0d | easy | sapient-sapiens | 2026-08-26 11:13:53.863762+00:00 | succeeded | 78fde246acf90f6cf73a1e35f8739db8bc634caf349ff2c1c1001cabe9c885ab | 20,855 | null | """Standalone nested inner/outer recurrence candidate.
Representation is categorical only: ``E_token + E_group + E_leftrel``. No
field is decoded into a number, no field value sets depth, halting, or any
mixing weight, and the only supervised target is the final answer.
Structure
---------
One prompt encoder produce... | {
"id": "368cbacd-0fd0-4b63-bf40-054fd2712b0d",
"created_at": "2026-08-26 11:13:53.863762+00:00",
"db_md5": "c8d3dbdf43c5a03f3be35db84af5d77a",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "5f96c1c7-98c1-4c3e-a539-cf8b03833f95",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 5.200687484374379,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00625
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.078132415566209,
"example_count": 600,
"e... |
368ed7dd-4b7e-479c-a2b3-9065bd8d3475 | easy | sapient-sapiens | 2026-08-25 08:59:09.892290+00:00 | succeeded | 6a145bdffd93fc6129b60ada16183c5d7787ef44f37606498c42a4daaac95225 | 18,291 | null | """Standalone nested inner/outer recurrence candidate.
Representation is categorical only: ``E_token + E_group + E_leftrel``. No
field is decoded into a number, no field value sets depth, halting, or any
mixing weight, and the only supervised target is the final answer.
Structure
---------
One prompt encoder produce... | {
"id": "368ed7dd-4b7e-479c-a2b3-9065bd8d3475",
"created_at": "2026-08-25 08:59:09.892290+00:00",
"db_md5": "9854b76ce7a0a80490a17b05f40aa93e",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "edb35e1d-8568-4203-bb7d-434c116d1468",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.3874611324521453,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.39192371518088,
"example_count": 600,... |
36a5aa1c-fb48-4d87-b1a6-faf293e5bb1e | easy | sapient-sapiens | 2026-08-25 09:06:41.091050+00:00 | succeeded | 72ad950f4a619c4e4f2672563c5fd0b8f49e07e3d0fa0e1ed516438aa1c87a1f | 18,757 | null | """Standalone nested inner/outer recurrence candidate.
Representation is categorical only: ``E_token + E_group + E_leftrel``. No
field is decoded into a number, no field value sets depth, halting, or any
mixing weight, and the only supervised target is the final answer.
Structure
---------
One prompt encoder produce... | {
"id": "36a5aa1c-fb48-4d87-b1a6-faf293e5bb1e",
"created_at": "2026-08-25 09:06:41.091050+00:00",
"db_md5": "2149e13b314854e1ad36cc8978afecd6",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "52f38e56-bffd-41df-9b47-e8814c885c12",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 3.4997939644232785,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.4931554077986644,
"example_count": 600,
... |
36a5db7f-cff1-4c78-8d82-ec7321c87956 | easy | jordanrubin | 2026-08-06 08:10:22.902111+00:00 | succeeded | bb47f9232dc848559bbf44a22b58c3ab4520dd1e57ce1cb71ddcda2d953a9be9 | 25,385 | null | """Rank-one product lift with a local categorical quotient/remainder stream.
One tied, T-blind outer transition has the following learned dataflow:
categorical STATE
-> gauge-normalized monotone scalar digit law
-> symmetric rank-one ordered-pair interactions
-> fixed right-origin place-sum aggr... | {
"id": "36a5db7f-cff1-4c78-8d82-ec7321c87956",
"created_at": "2026-08-06 08:10:22.902111+00:00",
"db_md5": "a2f2f2a4f13a20bd26d7c88c82041604",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "386f2937-ff8f-4079-8e2a-6657b0f00d5f",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 1.897740662097931,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8664783239364624,
"example_count": 100,
"exac... |
36af4a72-c47c-4548-8e66-e7d93b85dfc7 | easy | erdavis0 | 2026-08-25 02:02:57.883642+00:00 | succeeded | ae412f48bd9234888fd7bb68dc2dc649ac54245d371d4ab6f9d7476293bb40ba | 24,120 | null | """Matched categorical-boundary panel: C2 straight-through endpoint.
Public delimiter syntax is used only to route opaque token identities onto
separate right-aligned tapes. One T-blind learned macro maps a categorical X
tape and a digitwise N tape to a new categorical tape through freshly reset
interaction, carry, a... | {
"id": "36af4a72-c47c-4548-8e66-e7d93b85dfc7",
"created_at": "2026-08-25 02:02:57.883642+00:00",
"db_md5": "ea525cb9db225168d76b8b0dc9c9bc4a",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "2b395bdb-fefc-4774-b5b0-f2f7332ca966",
"tier": "easy",
"dataset_id": "e9",
"status": "succeeded",... | {
"score": {
"mean_loss": 1.8660330176353455,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333656191826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.873134732246399,
"example_count": 90,... |
36af91bb-85cb-414e-b90a-0f3c574f6b27 | easy | erdavis0 | 2026-08-26 04:27:10.000222+00:00 | succeeded | 34958bd1fecf39e0cd177cec37f4a5b763f56ecda181924fc21973bf094acd2c | 30,362 | null | """Latent-role continuous REV C3: sequence mixture plus role balance.
Public delimiter syntax only routes opaque token categories into distinct,
right-aligned N, X, and T tapes. Learned local cells compile separate
interaction, carry, and modulus-conditioned workspaces. N is isolated from
interaction and carry and f... | {
"id": "36af91bb-85cb-414e-b90a-0f3c574f6b27",
"created_at": "2026-08-26 04:27:10.000222+00:00",
"db_md5": "63a88d8095c75c1574375cc20d89bc96",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "f1c93562-123a-4db7-be52-e8c3fd02c804",
"tier": "easy",
"dataset_id": "e8",
"status": "succeeded",... | {
"score": {
"mean_loss": 5.221473932266235,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.27326203882694244
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.014470100402832,
"example_count": 85,
... |
36bfee3c-6187-480b-ba1a-799538027f8a | easy | sapient-sapiens | 2026-08-22 18:37:59.461339+00:00 | succeeded | bd0f787a0b5b59b0ba8c2751b73ff6335d80025764df8955b68a20d5788b8268 | 19,806 | null | """Batch-64 LR-6e-4 eight-layer T-private Q+V TRM for E5.
Research hypothesis: doubling peak LR to 6e-4 converts batch 64's greater
example throughput into stronger fixed-clock learning.
Primary experimental variable: peak learning rate, 3e-4 versus 6e-4, at fixed
batch 64 with model, optimizer, SAM, and loss held fi... | {
"id": "36bfee3c-6187-480b-ba1a-799538027f8a",
"created_at": "2026-08-22 18:37:59.461339+00:00",
"db_md5": "c0c0a301be7b7af2ed895826ce87ca81",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "039631ef-3b08-41d2-959c-28f439643e32",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.286812063952058,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2889291907372495,
"example_count": 600,
"ex... |
36c0202c-a693-4848-a9c5-c1b7963792b6 | easy | G-AshwinKumar | 2026-08-13 09:31:24.592997+00:00 | succeeded | 1aa3b158ef579b646ed6d026d05958ca5bd7aaeeff76fa22f988cdc06fff078a | 7,332 | null | """Multi-pass Gradient Accumulation and Refinement Optimizer with SAM perturbation and batch reuse."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from torch.optim.lr_scheduler import CosineAnnealingLR
from benchmark import (
BackwardPassContext,
... | {
"id": "36c0202c-a693-4848-a9c5-c1b7963792b6",
"created_at": "2026-08-13 09:31:24.592997+00:00",
"db_md5": "1cd225555897e11dfd196d52c8574c3d",
"submitter": "Ashwin Kumar",
"github_login": "G-AshwinKumar",
"run_id": "1a0eb133-cd39-4759-a0d6-7f8b3515f4cc",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.488658905029297,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01666666753590107
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.796750783920288,
"example_count": 100,
... |
36d11ff4-0145-41fb-bacf-65f6b72ac41e | easy | viridale | 2026-08-02 22:07:52.432607+00:00 | succeeded | ec34594f78e06a7daf014cd602590767aebfd23ba70f7ed9d77d6b832ece90b2 | 68,369 | null | """Collision-safe complete degree-six evidence search.
hypothesis: v115's seven minimum-depth rows can repeat the same learned field
input (for example one x at several depths), destroying the Vandermonde
certificate and allowing an alias to be absorbed permanently. Select the
cheapest seven rows that sha... | {
"id": "36d11ff4-0145-41fb-bacf-65f6b72ac41e",
"created_at": "2026-08-02 22:07:52.432607+00:00",
"db_md5": "f44f410a59f0463a8224d228170e164e",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "56d314cf-d36d-4389-83be-8c9f9b0d90b3",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 0.003014546214626436,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.00301454615599774,
"example_count": 800,
"... |
36e5f087-550b-43aa-a4d8-bced9f83b92d | easy | lzy54 | 2026-08-27 15:06:16.767433+00:00 | succeeded | 87e739e96effb7f4089165eb194e15a4ec46791e71d6012dbb1f9f36114bd2b3 | 3,989 | null | """Phase 1 A4 additive-recall control capped at 200 optimizer updates."""
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 = 1... | {
"id": "36e5f087-550b-43aa-a4d8-bced9f83b92d",
"created_at": "2026-08-27 15:06:16.767433+00:00",
"db_md5": "e00b838e713567ceba94204b17c936f8",
"submitter": "lzy54",
"github_login": "lzy54",
"run_id": "ed20ab1c-6445-41b6-b24b-3315c7df4820",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",
... | {
"score": {
"mean_loss": 4.2630854845047,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03513513784855604
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.8618948459625244,
"example_count": 60,
... |
36eb58e0-e86f-4794-b00e-84cf803bb5f3 | easy | gauravmishra | 2026-08-08 21:20:26.641638+00:00 | succeeded | b0b55fee7e56e6941e5a1bc53d69113790983b1be1b71bbe629ff7460b3da524 | 4,643 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
D_MODEL = ... | {
"id": "36eb58e0-e86f-4794-b00e-84cf803bb5f3",
"created_at": "2026-08-08 21:20:26.641638+00:00",
"db_md5": "161f77f076521ee63b89c3e3b8b443d3",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "0cb93f48-b09c-42b9-82e6-6667a9ed3d92",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.217065238742059,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.238906441248051,
"example_count": 600,... |
36ffe322-1998-45f7-a341-415a0d86d4ff | easy | alirezashirvani-jr | 2026-08-30 21:44:20.679241+00:00 | succeeded | 3fa6488bbbb697ef739fb15cafcce44b37ecb85477b348af20ab3ea038d5545c | 24,483 | 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
DIGIT_CARDINALITY = 10
STEP_FIELD_W... | {
"id": "36ffe322-1998-45f7-a341-415a0d86d4ff",
"created_at": "2026-08-30 21:44:20.679241+00:00",
"db_md5": "41fd7a05980039664511de3256ab2ebb",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "9299c43f-a9f7-4843-9cef-d976657da8c0",
"tier": "easy",
"dataset_id": "e2",
... | {
"score": {
"mean_loss": 8.746095514489127e-11,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 300,
"exact_accuracy"... |
3702fc25-67aa-4c06-bbad-580719a24ffd | easy | mnida | 2026-08-25 15:35:53.190149+00:00 | succeeded | faacb2c5aa341f3def1fb1ff2e60f9462475f55a37e47d390fcfc29bf40406c4 | 12,816 | null | """Digit-simplex shared quadratic attention with PyTorch AdamW."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MOD... | {
"id": "3702fc25-67aa-4c06-bbad-580719a24ffd",
"created_at": "2026-08-25 15:35:53.190149+00:00",
"db_md5": "368667c23a244da40a2ba16819d25a93",
"submitter": "mnida",
"github_login": "mnida",
"run_id": "7e4a350a-c295-40e4-b318-3f7dbcd2831a",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.1428078186173733,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1428888870461638,
"example_count": 600,
... |
37106984-88c7-4094-9114-00cbe79c59b5 | easy | AFeinfield | 2026-08-12 23:35:56.364380+00:00 | succeeded | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "37106984-88c7-4094-9114-00cbe79c59b5",
"created_at": "2026-08-12 23:35:56.364380+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "14c27f5b-23db-41eb-ae18-be0e68a8f995",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 3.0448198318481445,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.026800632476807,
"example_count": 100,
"exac... |
3712f440-8338-4716-a04b-17c2ba812424 | easy | erdavis0 | 2026-08-27 01:25:43.591789+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": "3712f440-8338-4716-a04b-17c2ba812424",
"created_at": "2026-08-27 01:25:43.591789+00:00",
"db_md5": "aa01dfb95948365c687b75f02f03cd86",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "eca6e8d5-2211-4982-95d3-d4610d8dd306",
"tier": "easy",
"dataset_id": "e10",
"status": "succeeded"... | {
"score": {
"mean_loss": 5.766772985458374,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.16862626560032368
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.251376152038574,
"example_count": 125,
... |
3716fa4b-5a14-4709-8d7c-1eee107ff49d | easy | lzy54 | 2026-08-29 15:21:53.122048+00:00 | succeeded | 04dc20a287ccf7506d4a60295243b72ba73c943238b6b9d0ddc83a68a37f810b | 17,703 | null | """R7 Time-Aligned Register Machine.
The shared learned transition never sees T. T only routes which recurrent
state is decoded, aligning supervision at T=1/2/3 with one/two/three calls of
the same transition. The transition contains no hard-coded task algorithm.
"""
from __future__ import annotations
import math
i... | {
"id": "3716fa4b-5a14-4709-8d7c-1eee107ff49d",
"created_at": "2026-08-29 15:21:53.122048+00:00",
"db_md5": "16a400706e11138e86074125bec9cf0c",
"submitter": "lzy54",
"github_login": "lzy54",
"run_id": "5fa9c23f-a6fc-467e-8e92-779533a2abdc",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.1416515746718527,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.144988679030551,
"example_count": 600... |
371869b3-7cff-485d-bc9e-6963f4217772 | easy | Yalyenea | 2026-08-12 08:38:06.473748+00:00 | succeeded | 1e9f9c637aebb441d6f62cef7d40d53e1a339de76d336faa39081f178af17bc6 | 14,476 | null | """Compact decimal recurrence with conservative hyperspherical Muon updates."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
... | {
"id": "371869b3-7cff-485d-bc9e-6963f4217772",
"created_at": "2026-08-12 08:38:06.473748+00:00",
"db_md5": "2573f615ee4d719451e99575376e591c",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "11a2c34e-6acb-4ef2-a627-86eade71f563",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 5.342150792184237,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0129166666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.682876390192006,
"example_count": 600,
... |
371df0a1-e501-416a-a49f-c5b013fe4e35 | easy | jordanrubin | 2026-08-07 05:50:01.730415+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": "371df0a1-e501-416a-a49f-c5b013fe4e35",
"created_at": "2026-08-07 05:50:01.730415+00:00",
"db_md5": "671c15c2db39c821999bae96a7defa51",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "32e0d890-c05c-4260-af3a-985917a13232",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 4.29659728690472,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012083333063249786
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.314423540114287,
"example_count": 600,
... |
371fa5d5-05fe-4acc-9a47-e7629a75baf4 | easy | erdavis0 | 2026-08-27 01:23:23.786431+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": "371fa5d5-05fe-4acc-9a47-e7629a75baf4",
"created_at": "2026-08-27 01:23:23.786431+00:00",
"db_md5": "2981fa4ef2c40f38bd0d3090c6d64a88",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "0f99e831-e39f-4e88-a0b3-7bbcbfafe807",
"tier": "easy",
"dataset_id": "e10",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.6650567054748535,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.12062626518309116
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.532597064971924,
"example_count": 125,... |
37202c5e-0e6f-4f97-afd1-cf6b586b99cf | easy | sapient-sapiens | 2026-08-15 08:15:20.119134+00:00 | succeeded | 9a6bb33a0f49a739edf3b75729668c4b5b374def290afa9f97801f5d937ab6c1 | 10,840 | null | """Narrow/deep looped Transformer with optional latent workspace tokens.
Research candidate for sampled-modulus fixed-T=2. The encoder builds immutable
prompt memory. A stack of learned operator blocks updates a mutable latent
stream, and the whole stack is reused for each task recurrence. Optional blank
workspace ... | {
"id": "37202c5e-0e6f-4f97-afd1-cf6b586b99cf",
"created_at": "2026-08-15 08:15:20.119134+00:00",
"db_md5": "bc8125d74bbb741a77e52f91071bec13",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "c7475943-abea-4e9c-97af-f1b1d5e575bf",
"tier": "easy",
"dataset_id": "e4",
"status": "su... | {
"score": {
"mean_loss": 2.0622201308658594,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005104166666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.064897511215666,
"example_count": 120... |
3721acb7-f17e-4afc-b5f0-b3b20466c67f | easy | ehonig | 2026-08-13 21:43:56.729932+00:00 | succeeded | d9da35246e4a487b5f97a438e83528f06666ddc62f088f96e3bf4e422a49f0a2 | 42,207 | null | """DigitLoop — a place-value recurrent model for repeated modular squaring.
The task is ``y = x^(2^T) mod N`` with the answer read back tail aligned, one
decimal place value per token position from the right. A stack of attention and
MLP layers has no primitive for the two operations this needs — a partial-product
co... | {
"id": "3721acb7-f17e-4afc-b5f0-b3b20466c67f",
"created_at": "2026-08-13 21:43:56.729932+00:00",
"db_md5": "57f4bf0aeed6257aecb6a3aeeadf5a0c",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "b6443a81-5fe5-4b6b-a27f-98acb980e522",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 5.308244943618774,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07333333045244217
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.509286403656006,
"example_count": 100,
... |
372850f8-6430-4b72-8577-7e5fdb7d93b5 | easy | chad-atexpedient | 2026-08-16 05:29:26.230886+00:00 | succeeded | db0cb9870aab6533e749dbb7e87e47ccc7bc12c28423cba55990915b45789aeb | 21,213 | null | """X55: Hybrid continuation-label + state consistency.
Combines X54's continuation-label supervision with X18's state consistency loss.
- Continuation: supervise A at step a+b with B's label z when A.output == B.input.
- Consistency: match T=2 intermediate state (step 1) with T=1 final state when
T=2.output == T=1.i... | {
"id": "372850f8-6430-4b72-8577-7e5fdb7d93b5",
"created_at": "2026-08-16 05:29:26.230886+00:00",
"db_md5": "c43eb3e3c460793e88aa17761944bdba",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "70556929-8052-4733-b9d4-7a5e437345cc",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.190140991109044,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666666666666
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.200386234463063,
"example_count": 600,... |
372a3c1d-37bb-4516-9087-ffb827b03529 | easy | anirudh-chakravarthy | 2026-08-12 03:09:13.480750+00:00 | succeeded | ce1e327e47472a6a2ae24148fc8218cc97871281a67d2fca6ec57f5e948ae0f2 | 14,963 | null | """A learned state transition applied once per requested time step."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
asser... | {
"id": "372a3c1d-37bb-4516-9087-ffb827b03529",
"created_at": "2026-08-12 03:09:13.480750+00:00",
"db_md5": "4ab4ff6430b752d5c26535ac7e575fa4",
"submitter": "Anirudh S Chakravarthy",
"github_login": "anirudh-chakravarthy",
"run_id": "5450dd62-f2df-404e-9a89-a2a50583ba24",
"tier": "easy",
"dataset_id": "... | {
"score": {
"mean_loss": 2.5382425573712073,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010833333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7729822509491924,
"example_count": 60... |
372a4aa9-bbd9-418e-bed8-8cc878b0e601 | easy | yashkant | 2026-08-17 00:08:48.008614+00:00 | succeeded | 352f6264e253326f7174283934eb9a30e0e70e3176ddf70b39a8726f5a824bd6 | 43,287 | null | """Round2230 masked mean/max MLP carrier for correction reset.
Each of the three width-320 self-attention sublayers is replaced by a
position-shared two-layer token MLP. Its three inherited input projections
produce local, masked-mean, and masked-max features from the current token
states; the inherited learned gate ... | {
"id": "372a4aa9-bbd9-418e-bed8-8cc878b0e601",
"created_at": "2026-08-17 00:08:48.008614+00:00",
"db_md5": "ef14e7b97eae0481c5a9dd54dfd5dcc3",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "0b3e2b4f-65d4-4d4a-9172-5139c6e1220b",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 5.803138256072998,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06333333440124989
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.750213146209717,
"example_count": 100,
... |
372fcde3-d958-45de-9a9e-420af1b87e5c | easy | alirezashirvani-jr | 2026-08-31 23:31:19.589757+00:00 | succeeded | 9a5422979edba9390ca1c79e01f9d569eeffe25dcd46455628e8560e27bd8420 | 21,426 | null | from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
HP = {
"width": 0,
"maximum_outer_steps": 64,... | {
"id": "372fcde3-d958-45de-9a9e-420af1b87e5c",
"created_at": "2026-08-31 23:31:19.589757+00:00",
"db_md5": "67c9ca914641ec708c2fc9aef7904b23",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "6a431ce0-75ac-4aba-85b9-dba2488a1bed",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 2.4268432228325167,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666679084302
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4291527132282345,
"example_count": 60... |
373ddaad-d8c3-49b0-a273-3d905ddd3512 | easy | 0Chris5R | 2026-08-02 15:38:41.930166+00:00 | failed | d8958ff1ff5735dddc18f4a69ec9a63070d248efce4655fe42d50a87727b4c63 | 29,702 | null | """Shared-digit Neural GPU with a K=4/order-2 completion-volume loss.
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 s... | {
"id": "373ddaad-d8c3-49b0-a273-3d905ddd3512",
"created_at": "2026-08-02 15:38:41.930166+00:00",
"db_md5": "e5e5fc276a8da30236360a1fe0c90a76",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "ab60fbf5-3e54-48ea-8dd7-0b1c2e5e24b4",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
373ef505-c47d-4d10-beb5-92be57f68724 | easy | chuk-uzowihe | 2026-08-05 22:11:16.291390+00:00 | succeeded | b9e39e4e25e0c68d03a369e9e1d636470be002ea8540518abf9b05de9fcd530e | 20,701 | null | """One Layer Deeper submission: FPRM over a compressed (N, x, T) lattice.
Structure
---------
1. The digit-token prompt is parsed (exactly, in int64, on GPU) into the three
integers N, x, T. Each becomes one latent vector via a fixed Fourier
featurization (phases of v mod P over a bank of periods P) followed by ... | {
"id": "373ef505-c47d-4d10-beb5-92be57f68724",
"created_at": "2026-08-05 22:11:16.291390+00:00",
"db_md5": "4104ac83dfdd0cf666ef572f708cef0c",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "92a75eef-1e05-4d55-84ca-41897f98e22b",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 13.835140228271484,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03333333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 13.549721717834473,
"example_count": 100... |
3741bf70-e12e-49dc-a51e-5c8ba5483feb | easy | chuk-uzowihe | 2026-08-08 02:46:03.384934+00:00 | succeeded | dbfd30b13a43f2f90abbd907a9a39e8f8b95a57cb01866a4eb60990fe0501807 | 27,744 | null | """One Layer Deeper submission: FPRM over a compressed (N, x, T) lattice.
v6 = v2's architecture (the only version that has generalized on m1: 15.18%
test, 11.6% at unseen T=32) rebuilt on the v5 chassis, with its one fatal
flaw fixed: input numbers are no longer parsed into int64 (silent garbage
past 19 digits) — ins... | {
"id": "3741bf70-e12e-49dc-a51e-5c8ba5483feb",
"created_at": "2026-08-08 02:46:03.384934+00:00",
"db_md5": "7ca7736996f7bac98da5cd782390e26f",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "51de70ce-4ec1-4aac-9b42-20d6efb7efb4",
"tier": "easy",
"dataset_id": "e2",
"status": "... | {
"score": {
"mean_loss": 25.359965324401855,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02250000135973096
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 25.845705032348633,
"example_count": 300... |
374319a8-dd71-4fae-950c-712b44c9bb63 | easy | sapient-sapiens | 2026-08-26 11:30:35.920474+00:00 | succeeded | 3affd8e7aa9c0ca1eb2ef540002c3153f1d09640e6d2c62d6405744f503c7f4b | 20,461 | null | """Standalone nested inner/outer recurrence candidate.
Representation is categorical only: ``E_token + E_group + E_leftrel``. No
field is decoded into a number, no field value sets depth, halting, or any
mixing weight, and the only supervised target is the final answer.
Structure
---------
One prompt encoder produce... | {
"id": "374319a8-dd71-4fae-950c-712b44c9bb63",
"created_at": "2026-08-26 11:30:35.920474+00:00",
"db_md5": "e47d0a8b410efdc3aa37920d7a87b3bc",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "fe38c9e3-f854-403b-875e-49186b7c624a",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 9.17438702937731,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.127888286060282,
"example_count": 600,
... |
37439b48-df76-4366-8ffe-82715eb8aef3 | easy | thedhruvrawat | 2026-08-19 15:28:17.816433+00:00 | succeeded | e8e1bfe3f3464f20813a2d9b76c63c8a9fc2be6e23efa451e964b6127de9a69e | 24,978 | null | """One Layer Deeper submission: a weight-tied recurrent refinement transformer.
Architecture
Token + learned absolute position (small random offset while training) are
fused into an injection tensor ``x``. A two-layer pre-norm block with RoPE,
QK-norm and SwiGLU is applied L times to a latent state ``z``,... | {
"id": "37439b48-df76-4366-8ffe-82715eb8aef3",
"created_at": "2026-08-19 15:28:17.816433+00:00",
"db_md5": "058331bf67f9cc5393aef68d4bf7e4be",
"submitter": "Dhruv Rawat",
"github_login": "thedhruvrawat",
"run_id": "47700215-951f-4ece-bc82-cc50a0a6cf91",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 5.652373552322388,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06833333149552345
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.794947147369385,
"example_count": 100,
... |
374b9638-2918-4131-8c09-b2cd78b4ac42 | easy | alexkreimer | 2026-08-22 21:15:56.932422+00:00 | succeeded | d4d6ef0986d871383c73d63942a038be58e4781d6d66c23541d8b412692c875e | 10,559 | null | """Tied-recursive latent-depth model, iterated exactly T times per example.
Starting point for One Layer Deeper: one small attention+MLP cell is
applied repeatedly to a shared latent state (weight-tied recursion), with
the original token/position embeddings re-injected at every iteration so
the operands (N, x, T) don'... | {
"id": "374b9638-2918-4131-8c09-b2cd78b4ac42",
"created_at": "2026-08-22 21:15:56.932422+00:00",
"db_md5": "6c7bb6b2f2cd5884cae024c79ed78565",
"submitter": "Alex Kreimer",
"github_login": "alexkreimer",
"run_id": "fffa6221-8c5b-4095-b736-7f9948d585b5",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 4.256081335331698,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.016666666865348817
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.344243711500026,
"example_count": 100,... |
374d5a71-5a25-4e57-8cd7-e2f47e62cffe | easy | im-a-king-but-im-far-from-a-saint | 2026-08-10 15:35:43.254324+00:00 | succeeded | 7f0b68e22e19b808175ae97106794527cce91810c353b5d1b4278a9b5f40bcbb | 7,507 | null | """Weight-tied recurrent (Universal-Transformer-style) baseline.
Applies ONE shared transformer block NUM_LOOPS times instead of stacking
distinct layers. Re-injecting the original embedding at every iteration
keeps the recurrence anchored to the problem statement, which is what lets
this kind of network extrapol... | {
"id": "374d5a71-5a25-4e57-8cd7-e2f47e62cffe",
"created_at": "2026-08-10 15:35:43.254324+00:00",
"db_md5": "066266388acf70f1c1af3b93aa608a35",
"submitter": "im-a-king-but-im-far-from-a-saint",
"github_login": "im-a-king-but-im-far-from-a-saint",
"run_id": "a2586958-6db3-4719-8193-8dec78363c5a",
"tier": "... | {
"score": {
"mean_loss": 2.293760595787255,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0031250000093132257
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.335229836545374,
"example_count": 800... |
3758c2dc-3f86-47e2-811c-6b4e541fc33f | easy | sapient-sapiens | 2026-08-25 07:46:14.483574+00:00 | succeeded | 3453c72f2c6682035376d9f1f75e05803fcca61889f0476aa12ddc2c6915d3f6 | 18,825 | 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": "3758c2dc-3f86-47e2-811c-6b4e541fc33f",
"created_at": "2026-08-25 07:46:14.483574+00:00",
"db_md5": "6ca80f6e968013566d56805fdaf2ffdc",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "23a5211c-0fae-4139-a81a-759c0423b53e",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.173329333833327,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1794589907599136,
"example_count": 600... |
37601145-216f-4c1e-b960-1052266cfd37 | easy | arthurfeeney | 2026-08-07 06:39:36.251139+00:00 | succeeded | 619bb389240d9185e5146d1aa7452b3dde173fca19e1f6159b89fd9a516fdded | 8,387 | null | r"""
Notes
1. goal is basically learn y = G(x_0, T, N).
G is always a recurrence `x_t+1 = f(x_t, t) mod N`.
Shouldn't be possible in general since f isn't determinable from a dataset...
I.e., multiple f can generate the same training dataset.
2. can't really use any info on structure of f...
"""
from __future... | {
"id": "37601145-216f-4c1e-b960-1052266cfd37",
"created_at": "2026-08-07 06:39:36.251139+00:00",
"db_md5": "0caaadc8f5ff5f97f0dfe1c7bc5ac884",
"submitter": "Arthur",
"github_login": "arthurfeeney",
"run_id": "a39d3df4-19eb-4733-ac0d-87e9dfe2a020",
"tier": "easy",
"dataset_id": "e1",
"status": "succee... | {
"score": {
"mean_loss": 8.999359607696533,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014999999664723873
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.143357276916504,
"example_count": 100,... |
3760b068-ff3e-4069-8c67-6075abe7f9b5 | easy | liam-gb | 2026-08-16 20:32:16.220431+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": "3760b068-ff3e-4069-8c67-6075abe7f9b5",
"created_at": "2026-08-16 20:32:16.220431+00:00",
"db_md5": "46c5decfff7d97a850fb297981e9d310",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "aa116731-b810-4805-b60f-0f1f86dc860f",
"tier": "easy",
"dataset_id": "e7",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.6636443734169006,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4106951989233494
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.064560413360596,
"example_count": 85,
... |
37612a20-f016-44a1-ad36-fa8fb8beb1b7 | easy | oupadhyay | 2026-08-10 19:22:33.647150+00:00 | succeeded | 64ff4a06a02ff905200257d0ff0a8d8c103116cb2e091d2eceeade2ab354c3a7 | 10,720 | null | """Persistent learned edge-state processor for Easy E5."""
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, RANK, HEADS, ROUNDS, MAX_PLACES, MAX_STEPS = 64, 32, 4, 4, 16, 64
class ... | {
"id": "37612a20-f016-44a1-ad36-fa8fb8beb1b7",
"created_at": "2026-08-10 19:22:33.647150+00:00",
"db_md5": "c90b2856492fa80735e68ed713b8ff94",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "716d6cbe-9bb4-4782-a487-6797ee2829f9",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 5.344611626291947,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000037252903
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.455306914889759,
"example_count": 600,... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.