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 |
|---|---|---|---|---|---|---|---|---|---|---|
48844769-4e13-48b2-a71f-dbe587f103f0 | easy | DDanlov | 2026-08-14 08:05:25.097943+00:00 | succeeded | 07bdd808d3da7c87f8f41193fb0b7eadc4e210b41784a54cc5ed67edeaad0ed7 | 21,977 | null | import math
from typing import Optional
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark import Submission, assert_model_state, OptimizerBundle
except ImportError:
try:
from client import Submission, assert_model_state, OptimizerBu... | {
"id": "48844769-4e13-48b2-a71f-dbe587f103f0",
"created_at": "2026-08-14 08:05:25.097943+00:00",
"db_md5": "cbd45f88286e55da918950a1685d7990",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "3d34aba5-648d-4f7d-b2dd-151d1b388bd4",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.8180792130075676,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.818438958426762,
"example_count": 600,
... |
4891dcec-2cf2-449f-b6b0-1bcbfd305f85 | easy | MarcoF1 | 2026-08-13 20:36:07.824612+00:00 | succeeded | 12776b46a674879babc17d0e47032397ff8071e46c9d192bcd7b1ceafbdf6abf | 14,397 | null | """Residue-clock model for fixed-N generalization.
Representation: parsed integer values of x and N feed (a) exact residue-phase
features over a bank of small integer periods, (b) learned double-hashed value
embeddings (per-residue identity), and T feeds cyclic phase features
(T mod 2..48) plus saturating channels — p... | {
"id": "4891dcec-2cf2-449f-b6b0-1bcbfd305f85",
"created_at": "2026-08-13 20:36:07.824612+00:00",
"db_md5": "6fa9500dea625f283a50d08a19dd4b84",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "010537a4-1dbc-4193-aba7-59698a28a471",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.289226717211393,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005729166666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3298678736099867,
"example_count": 120... |
48931077-5898-4f84-8355-69ebe63a05c4 | easy | AFeinfield | 2026-08-18 23:45:05.028257+00:00 | succeeded | dd6fa161e518fbc52f4ee06609ebf626a2e6547dc0d13279f3c9a2df6e87d74f | 13,788 | null | """083: conv kernel 5 -> 9 in the multiply sublayer.
Direct test of the mechanism story: the conv gathers neighbouring digits
so the bilinear form can multiply them, so a wider kernel should let more
digit PAIRS interact per iteration.
Original 073 notes follow.
073: BILINEAR multiply + K=5 agreement heads.
The multi... | {
"id": "48931077-5898-4f84-8355-69ebe63a05c4",
"created_at": "2026-08-18 23:45:05.028257+00:00",
"db_md5": "7f429cd8b404c5ed7219db333f350708",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "02d2cee2-cf73-410c-9980-c0bc7b0271d9",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 16.820249557495117,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009791667107492685
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 17.118196487426758,
"example_count": 30... |
489f53e6-6207-4e07-9d54-d85153fdfcc6 | easy | DDanlov | 2026-08-13 22:39:20.009073+00:00 | failed | 9c07177e2f874709e157bb02af1e8cf127a13d64096dfa9b2153f573544f68c4 | 22,143 | null | import torch
import torch.nn as nn
import torch.nn.functional as F
import math
from dataclasses import dataclass
from typing import Optional
from torch.optim import Optimizer
try:
from client import Submission, assert_model_state, OptimizerBundle
except ImportError:
Submission = None
assert_model_state = N... | {
"id": "489f53e6-6207-4e07-9d54-d85153fdfcc6",
"created_at": "2026-08-13 22:39:20.009073+00:00",
"db_md5": "af2c84ec212cff715dc26439fcf2b049",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "8e3e6a61-a1ba-4f93-a1bc-6b6e347b2a0d",
"tier": "easy",
"dataset_id": "e3",
"status": "failed",
... | null
|
48b2dbab-3e2b-4827-97df-47804b264abd | easy | karanganesan | 2026-08-08 04:09:39.996776+00:00 | succeeded | 8d9eecdc8607c91b190336e24b49597738decc69ce7d4e24ee5930ae97cfcad7 | 27,127 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "48b2dbab-3e2b-4827-97df-47804b264abd",
"created_at": "2026-08-08 04:09:39.996776+00:00",
"db_md5": "3e8439ba8ab561733e7dc883fab4ecb2",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "c891eb3f-7014-4de1-9a04-b76657524d45",
"tier": "easy",
"dataset_id": "e3",
"status": ... | {
"score": {
"mean_loss": 10.154979705810547,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003124999930150807
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.930216789245605,
"example_count": 800... |
48b45ef9-00b5-44b7-bd71-1da7c81b226c | easy | DDanlov | 2026-08-26 23:44:12.891587+00:00 | failed | b5d812a635c651efcc77d90b9e50117456dc30cacbf9d4b50d25e15dc893e9c6 | 35,196 | null | import math
import time
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from onelayerdeeper.api import Submission, OptimizerBundle, ModelSpec, OptimizerSpec
except ... | {
"id": "48b45ef9-00b5-44b7-bd71-1da7c81b226c",
"created_at": "2026-08-26 23:44:12.891587+00:00",
"db_md5": "b9eb8593bd3c217e362eb735a82fc880",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "be7078c2-b525-4ae5-b33a-2cf809443806",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
48b9925c-239d-46d7-a2aa-a035fc45c531 | easy | yashkant | 2026-08-10 21:30:42.094409+00:00 | succeeded | 5009134053ca85753760fce70a90333e1b374205628be5201f1e7dc4459fe898 | 22,265 | null | """Standalone compact-bigram hash, AdamW LR3.75e-3, EMA 0.997."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
from torch import ... | {
"id": "48b9925c-239d-46d7-a2aa-a035fc45c531",
"created_at": "2026-08-10 21:30:42.094409+00:00",
"db_md5": "d928b459701ca69468b300c96bb056f8",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "1998dbe4-6c8b-4319-ab03-ebb8263d524d",
"tier": "easy",
"dataset_id": "e4",
"status": "succeed... | {
"score": {
"mean_loss": 2.06392387164446,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009270833345750969
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.066083567737642,
"example_count": 1200,... |
48bae8f7-bfce-49ca-b861-c9a66ee64202 | easy | jayyala | 2026-08-16 14:25:37.915227+00:00 | succeeded | fdbface5dfdccbe87cd9f3d2dbae55b3a07f8c6bd7ae61ca96cb2cacc656a254 | 11,078 | null | """A recurrent machine over structurally encoded byte slots.
Integers use little-endian byte slots for compactness, but a byte is never an
atomic learned symbol: both encoding and decoding factor through its eight
bits. One tied transition cell is applied exactly T times. Larger integers
add slots and recurrent comp... | {
"id": "48bae8f7-bfce-49ca-b861-c9a66ee64202",
"created_at": "2026-08-16 14:25:37.915227+00:00",
"db_md5": "37e7a823641d4e5e3d415a40bb3e264e",
"submitter": "jayyala",
"github_login": "jayyala",
"run_id": "4966125f-6fdd-41c6-a2a7-fa35cbe19f50",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded"... | {
"score": {
"mean_loss": 33.89633560180664,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04031531699001789
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 35.135135650634766,
"example_count": 60,
... |
48d84fc4-f293-4be8-b5aa-db628431d4a0 | easy | yashkant | 2026-08-27 16:25:50.294010+00:00 | succeeded | 978391cef34c75f10d223afbbb6a15b516c4ef2c3f1b3c4b73bab61ffe753fd4 | 32,588 | null | """v18d = v18 + a label-free reducer N-diet (random semiprimes 10..24 bit each step) so OOD-N generalises on every
tier, including fixed-N Easy/Medium (where a single training modulus otherwise starves the reducer of N-variance).
One Layer Deeper HARD submission v18 -- RESIDUE-NUMBER-SYSTEM SQUARER + LEARNED REDUCER.
... | {
"id": "48d84fc4-f293-4be8-b5aa-db628431d4a0",
"created_at": "2026-08-27 16:25:50.294010+00:00",
"db_md5": "210313dc486261e336113804d68f60b1",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "947c83d2-73f3-4c0c-a930-e2b1b5f8fed0",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 100,
"exact_accuracy": 1.0,
"... |
48dcd9d4-0b37-4010-a9e5-d6a72c5440be | easy | adambrzosko | 2026-08-30 14:13:17.874676+00:00 | failed | 7b408ef4a5da2dfd1ef86effc2f86fc4d28f983001f9ce451a9e63e27a68851f | 9,373 | null | """One Layer Deeper — digit-slot recurrent machine ("dm").
The prompt is [N] digits(N) [X] digits(x) [T] digits(T) and the label is the
decimal expansion of x^(2^T) mod N, tail-aligned: the last prompt position
carries the units digit, the one before it the tens digit, and so on.
Architecture. The model parses the ... | {
"id": "48dcd9d4-0b37-4010-a9e5-d6a72c5440be",
"created_at": "2026-08-30 14:13:17.874676+00:00",
"db_md5": "29c5333c3432b0100eca52a5d398bf79",
"submitter": "Adam Brzosko",
"github_login": "adambrzosko",
"run_id": "d257bb66-91c7-4e07-9947-76bceb0d84c7",
"tier": "easy",
"dataset_id": "e6",
"status": "f... | null
|
48e836f5-e583-4495-aa9d-e89ea4a09446 | easy | AFeinfield | 2026-08-13 00:23:22.911875+00:00 | succeeded | f8a7bd403f425a6dac7ed45a8f50f9416367fa66cd86fa3f19bce64382812ef7 | 7,289 | null | """Looped (weight-tied) bidirectional Transformer, d=256, 8 iterations.
016: 013 + dropout 0.1 on attention-out and MLP (regularization diversity
for the grokking transition; complements/replaces pure wd pressure)
to amortize the evaluator's ~200ms/iteration overhead into 8x more steps.
prior. Same parameter count ord... | {
"id": "48e836f5-e583-4495-aa9d-e89ea4a09446",
"created_at": "2026-08-13 00:23:22.911875+00:00",
"db_md5": "e52f73ce8e951fd258eb9d88d8dae494",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "63993644-22d9-44f2-a5b1-46a22c6d9817",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 3.9753764867782593,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03666666615754366
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.912755250930786,
"example_count": 100,... |
48ecfa06-455f-4d5d-81e3-67affcbb25c6 | easy | oupadhyay | 2026-08-28 02:16:49.023110+00:00 | succeeded | 874dc534a6935a77a9cd094e6592c53b0fd3f12c1a46d38b72485a6149f9d31e | 4,275 | null | """Conditional structured-loss phase 5: focal075."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,TokenLossBatch,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,m... | {
"id": "48ecfa06-455f-4d5d-81e3-67affcbb25c6",
"created_at": "2026-08-28 02:16:49.023110+00:00",
"db_md5": "1a0da58f427ff03209a7199ad955f145",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "644a2796-a927-4b73-b32c-024181f732d7",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 1.9985162615776062,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08355856314301491
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9358478784561157,
"example_count": 60,... |
48f439aa-d786-4aaf-ba51-a09fad8d3a49 | easy | alirezashirvani-jr | 2026-08-31 00:32:27.576525+00:00 | succeeded | d9be1270c66f4cec678728ecb8d3888d74765a4fd36126c152a8eb335ac95f65 | 19,728 | null | from __future__ import annotations
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
PERIOD_MINIMUM = 3
PERIOD_MAXIMUM = 100
EX... | {
"id": "48f439aa-d786-4aaf-ba51-a09fad8d3a49",
"created_at": "2026-08-31 00:32:27.576525+00:00",
"db_md5": "d28e9bd3af3854ff3baff14eaf466501",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "6dc63854-33df-4805-ac70-fbac797989c1",
"tier": "easy",
"dataset_id": "e6",
... | {
"score": {
"mean_loss": 1.201117604970932,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6655405461788177
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8517959117889404,
"example_count": 60,
... |
48ff96b0-f66b-4a94-91f8-55963011ba47 | easy | liam-gb | 2026-08-14 00:04:38.118916+00:00 | succeeded | 1d746e3ae8d56e5df34ccf93f23d12fd3fcbc2567f189b9b9955d782eab23df6 | 15,758 | null | """rns_pm_gather_ct_widet_b46rnore: 45-prime bank + per-prime temperature
+ tables at 3e-2 + reuse OFF -- every lever that has won tonight, combined.
Base: 45-prime bank + per-prime temperature + tables at lr 3e-2.
Base: widet + per-prime temperature + 45-prime bank.
The wide readout is what makes long answers reach... | {
"id": "48ff96b0-f66b-4a94-91f8-55963011ba47",
"created_at": "2026-08-14 00:04:38.118916+00:00",
"db_md5": "b8e9c96ce32883e9498e263123a1efcd",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "f5301801-b5de-48eb-8818-8b4be9cd9acb",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.952395783552715,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06708333347613613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.586397171020508,
"example_count": 600,
... |
49026305-06a8-43ae-ba68-c34a9ad0fc9a | easy | Ranoobaba | 2026-08-21 04:09:32.771511+00:00 | succeeded | 7bdd1d7aefc37fd2b0b8bbecca76677e3c2532e9f2c0182d587595ce006c0a7d | 6,666 | 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": "49026305-06a8-43ae-ba68-c34a9ad0fc9a",
"created_at": "2026-08-21 04:09:32.771511+00:00",
"db_md5": "19c663271f8de5d1c23dc7a80d2697dc",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "8ec3c874-202a-426e-887b-011b5e6111b5",
"tier": "easy",
"dataset_id": "e5",
"status": "... | {
"score": {
"mean_loss": 2.355857486682088,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.409979154978098,
"example_count": 600,... |
49139fbc-0121-4189-b311-ced35e996e17 | easy | Ranoobaba | 2026-08-16 06:10:49.770899+00:00 | succeeded | a6940f6372845258ce3016d8f560344f18e7be865cc0214aa35b2d5611cd6d80 | 5,723 | 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": "49139fbc-0121-4189-b311-ced35e996e17",
"created_at": "2026-08-16 06:10:49.770899+00:00",
"db_md5": "f2dd087741b2bcf8811bb94acbaf5cdf",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "57669ccc-75ca-45ec-b0f1-a4ac87a4e012",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 7.232529886940174,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05166666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.689770221710205,
"example_count": 100,
... |
4915ae1a-7ebe-47a3-b554-7f5088594cbd | easy | jordanrubin | 2026-08-03 20:49:57.868885+00:00 | succeeded | 09586f90ffa296e3bdbb25e6a0b059c7579344900d2f7d9eb12b0e902c4848f6 | 15,390 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "4915ae1a-7ebe-47a3-b554-7f5088594cbd",
"created_at": "2026-08-03 20:49:57.868885+00:00",
"db_md5": "ef7d67d872a65aa96e81ca53991ab27d",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "0a8224e9-9f91-4e66-a2d5-4a403226ba99",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 3.266493217500707,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.20562499652927121
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.082639545790687,
"example_count": 300,
... |
4918c0c6-4e17-4cfa-8fb9-0d9bbcb8c9b9 | easy | DDanlov | 2026-08-20 13:14:09.103733+00:00 | succeeded | a51bfd5b3f73a376b8b87e9057350c7a039a6e4889a26b38a7fde677686306b5 | 20,958 | null | from __future__ import annotations
import math
import time
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
except Imp... | {
"id": "4918c0c6-4e17-4cfa-8fb9-0d9bbcb8c9b9",
"created_at": "2026-08-20 13:14:09.103733+00:00",
"db_md5": "22dabee10cfa54181bae781e83d445da",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "2a2479b0-4309-41e4-a91b-a98875fabc2e",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.5070596269473677,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.512000592834746,
"example_count": 600... |
491a11e0-eb58-4554-b8c9-2921c1fcd2e3 | easy | yashkant | 2026-08-17 00:20:08.708729+00:00 | succeeded | c4a7acfd48ee0c1fe7c7e734f3ff59e6a76a4c23f5da37663b620ce4ea37288b | 46,509 | null | """Round2202 factorizes one tiny recurrent correction into residue and carry.
The measured correction-reset official carrier stays intact. A pooled
four-value/four-carry state runs one shared transition twice per parsed-T
cycle and reinjects its immutable input halves on every call. Carry is
learned from signed resi... | {
"id": "491a11e0-eb58-4554-b8c9-2921c1fcd2e3",
"created_at": "2026-08-17 00:20:08.708729+00:00",
"db_md5": "e6a56ce878f0145792ec843649a315a9",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "91309d0c-bf0a-46d1-bb16-96ed30a1d426",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 2.233645796775818,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.024166667833924294
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.271254777908325,
"example_count": 300,... |
491d8c29-547f-4e37-8062-72819aba56be | easy | paulpister | 2026-08-04 17:21:28.795713+00:00 | succeeded | 617b5d3d1bf3c45844ecbcbbb35c14a15c5c9f5225c928b45cc229ce2b65e92a | 10,182 | null | """Loop-T tied recurrence v2: SwiGLU blocks, MICRO core steps per squaring.
Design:
- Structural features computed from input_ids (field segment, digit place
value from the little end, reverse position from sequence end). Sinusoidal
place/revpos features so longer OOD-N numbers extrapolate smoothly.
- T is parsed ... | {
"id": "491d8c29-547f-4e37-8062-72819aba56be",
"created_at": "2026-08-04 17:21:28.795713+00:00",
"db_md5": "c3eecd26d8c4f14428f736d6b60ffb19",
"submitter": "paulv",
"github_login": "paulpister",
"run_id": "c0f788d0-83b8-44ab-afa5-22e5065a7fe0",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded... | {
"score": {
"mean_loss": 2.128755569458008,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04333333298563957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3182268142700195,
"example_count": 100,... |
4923869b-2289-45bf-8e76-caff0f41f1a8 | easy | newjordan | 2026-08-09 20:31:36.127661+00:00 | succeeded | fa41fa7a6512e51b49452daf859bc2a0613cd139fb521c07f5b23a2fdba6a67b | 94,093 | null | """Helix-Crawler — a recursive-weight-shared "crawler" ALU for modular squaring.
The task is r_{i+1} = r_i^2 mod n, chained T times: a *recurrence* over a single
operator. This build embodies the crawler / Frugendorff lineage (spark:~/sota_crawler)
mapped onto that structure. n never enters as a table row (unseen modu... | {
"id": "4923869b-2289-45bf-8e76-caff0f41f1a8",
"created_at": "2026-08-09 20:31:36.127661+00:00",
"db_md5": "819b318faaf6fb8ce08db4f8cee528d0",
"submitter": "Frosty40",
"github_login": "newjordan",
"run_id": "35064350-c14f-4143-a16b-1933632d60b1",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 5.73977513392945e-06,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.740134054757992e-06,
"example_count": 800,
... |
49252ab4-81f3-4b2f-b0d3-d1408a1df109 | easy | sdrshn-nmbr | 2026-08-06 01:04:38.609035+00:00 | succeeded | 00e1b0ec51095afca533ddede35c0a94d55b88407ac492156b9a7cbf2de6f491 | 54,740 | null | """Typed model and optimizer factory for controlled architecture experiments."""
from dataclasses import dataclass, replace
from functools import partial
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
BatchReuseContext,
ModelSpec,
... | {
"id": "49252ab4-81f3-4b2f-b0d3-d1408a1df109",
"created_at": "2026-08-06 01:04:38.609035+00:00",
"db_md5": "ed757e9f14e52f66e824a47afecf8bb8",
"submitter": "Sudarshan Nambiar",
"github_login": "sdrshn-nmbr",
"run_id": "8280f0d4-b578-4d48-9508-e4b9e4c08b97",
"tier": "easy",
"dataset_id": "e5",
"status... | {
"score": {
"mean_loss": 3.395086910949166,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666679084302
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.284751934855508,
"example_count": 600,... |
492725dc-a6fe-404a-9ff5-1c3c4200c9fb | easy | erdavis0 | 2026-08-22 08:31:42.459087+00:00 | succeeded | 4052302344c2ee5544752d8dd2b682cc62f86ed61a7d43e055b398661db547e4 | 4,774 | null | """Masked butterfly-style global mixing over sequence positions."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
WIDTH = 64
HIDDEN... | {
"id": "492725dc-a6fe-404a-9ff5-1c3c4200c9fb",
"created_at": "2026-08-22 08:31:42.459087+00:00",
"db_md5": "7b255a5697e39c756d9c195ef0c4c794",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "bc1aeb1e-57a2-4a16-97bd-872744465f16",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 6.503941535949707,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08198198676109314
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.4105610847473145,
"example_count": 60,
... |
4928cf28-00a0-4796-9fc7-9e674108e6a7 | easy | GLauzza | 2026-08-20 21:24:25.570957+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": "4928cf28-00a0-4796-9fc7-9e674108e6a7",
"created_at": "2026-08-20 21:24:25.570957+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Gabriel",
"github_login": "GLauzza",
"run_id": "96350c66-1280-41ad-b4f0-e8d9ba26973f",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded"... | {
"score": {
"mean_loss": 4.71287214756012,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006756756920367479
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.502363681793213,
"example_count": 60,
... |
492dd54c-f56e-4c09-a59b-afcb5e2f854f | easy | gauravmishra | 2026-08-07 19:00:00.120469+00:00 | succeeded | c3b12ecbb4b36fb7cba102af079a325162580d028fd59093923e49f9fd52c824 | 3,724 | null | """Two-pass Transformer with two independently parameterized blocks."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128... | {
"id": "492dd54c-f56e-4c09-a59b-afcb5e2f854f",
"created_at": "2026-08-07 19:00:00.120469+00:00",
"db_md5": "847362712b6c84e8a3b8e935a647ea97",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "fe6fddb8-98dc-4e45-8690-09026e9aa405",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 3.6315720081329346,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333334140479565
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.7736945152282715,
"example_count": 1... |
49312bc9-4330-4ed0-9d03-ffd9cde88abc | easy | DDanlov | 2026-08-14 07:27:57.180996+00:00 | succeeded | 487aa0da03403574dc6ec80f73b42052299524c2adf7d680cc67d5dc2af59ca7 | 21,978 | null | import math
from typing import Optional
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark import Submission, assert_model_state, OptimizerBundle
except ImportError:
try:
from client import Submission, assert_model_state, OptimizerBu... | {
"id": "49312bc9-4330-4ed0-9d03-ffd9cde88abc",
"created_at": "2026-08-14 07:27:57.180996+00:00",
"db_md5": "695f11a7597eaac2123d88590f6dc738",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "78f756ab-1ca7-41f5-99cc-be64ea7f8beb",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.4581588662869436,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4645545433187697,
"example_count": 600,... |
493208b4-5df1-451f-a1c3-7878f09e1854 | easy | oupadhyay | 2026-08-15 21:56:58.752144+00:00 | succeeded | 5737bb4e824cc21780e315d02372894110c33f7f996e9339d07e242ad2b87d8a | 3,988 | null | """Conditional optimizer screen: wd10."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_size,s... | {
"id": "493208b4-5df1-451f-a1c3-7878f09e1854",
"created_at": "2026-08-15 21:56:58.752144+00:00",
"db_md5": "0b01606289c40609a41e96fa407a5e0f",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "0cc386c3-7ff3-406a-9950-baddd57dbfc6",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.174728536805667,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0020833333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.155344495858847,
"example_count": 600... |
4943514c-b56d-468a-9ede-b0381aaa5be7 | easy | anirudh-chakravarthy | 2026-08-12 04:21:03.911161+00:00 | succeeded | 78e7fac0dcbe708a41f9c489e5978832969b2d3d8cf4729545e04ef897ccb2d2 | 15,374 | 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": "4943514c-b56d-468a-9ede-b0381aaa5be7",
"created_at": "2026-08-12 04:21:03.911161+00:00",
"db_md5": "67c9c13a5c8444127b4abc24cb50d4e7",
"submitter": "Anirudh S Chakravarthy",
"github_login": "anirudh-chakravarthy",
"run_id": "df41481c-b8fa-44bd-805d-56bb9a265745",
"tier": "easy",
"dataset_id": "... | {
"score": {
"mean_loss": 3.852587859618448,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666741172473
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.676250825548386,
"example_count": 600,... |
49440111-e179-4cad-8fec-cccf89d018b3 | easy | rteehas | 2026-08-19 09:33:03.317444+00:00 | succeeded | 67a121bc5064417c595c65f4606dd9dc4cbe2a449dbffb056559cf8932241f1f | 50,908 | 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": "49440111-e179-4cad-8fec-cccf89d018b3",
"created_at": "2026-08-19 09:33:03.317444+00:00",
"db_md5": "15a833c238504b9d53e3ecf1bc8d9e51",
"submitter": "Ryan Teehan",
"github_login": "rteehas",
"run_id": "b1704ee3-5770-43fc-b941-a4c5a8a9d50b",
"tier": "easy",
"dataset_id": "e4",
"status": "succee... | {
"score": {
"mean_loss": 6.785299538281659,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003541666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.807546155038139,
"example_count": 1200... |
494c03d2-772b-467d-906d-84ec620ef76e | easy | 0Chris5R | 2026-08-08 09:27:58.509891+00:00 | succeeded | 485afd4496cb6a4249aaaf1402f99f66b05658589131bc454bd3bb351982725b | 22,913 | 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": "494c03d2-772b-467d-906d-84ec620ef76e",
"created_at": "2026-08-08 09:27:58.509891+00:00",
"db_md5": "d67a3f8d2303d1360c92d54d024dc909",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "c87a5fb1-0c2b-441b-8265-d75e054b1980",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.751465013878513,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04416666670391957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.665343549754053,
"example_count": 600,
... |
4952e54a-fc60-41c7-93c8-181e42bb1bdf | easy | KaustubhKumar05 | 2026-08-25 16:06:05.103923+00:00 | succeeded | b6d72aabc7da752cbf89d97df87ad90c9e9600f805ec3541ee7c84fab6ab2bb3 | 17,454 | null | """conv_sat: final_v1 + the DNGPU saturation cost (Freivalds & Liepins 2017).
Our hard gate clamp(1.2*sigmoid-0.1,0,1) has EXACTLY ZERO gradient outside
|x|>=2.398. The paper that introduced this gate says an explicit saturation
penalty is mandatory: without it units leave the linear band early and can
never return ("... | {
"id": "4952e54a-fc60-41c7-93c8-181e42bb1bdf",
"created_at": "2026-08-25 16:06:05.103923+00:00",
"db_md5": "5c8b9f4ebca3a2775b1bf2c2ddbd787e",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "8c12b96c-e4ba-4617-8608-6a7b3fa8cce5",
"tier": "easy",
"dataset_id": "e3",
"status": "succee... | {
"score": {
"mean_loss": 6.334872823282753,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011875
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.922012929711765,
"example_count": 800,
"... |
49535bf0-8808-41b2-98b7-58ec182a1c2a | easy | alirezashirvani-jr | 2026-08-30 08:40:30.771777+00:00 | succeeded | 1077d93bff0f88167a1af0f45e4abf432c45dabd2262a6e4423038ba0799d48b | 26,556 | null | from __future__ import annotations
import math
import time
from contextlib import nullcontext
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
MODEL_... | {
"id": "49535bf0-8808-41b2-98b7-58ec182a1c2a",
"created_at": "2026-08-30 08:40:30.771777+00:00",
"db_md5": "d3c25e5ed3fb41c685a17ce10e0fdbed",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "bc7fac6e-2fc1-476e-8275-290bd429d548",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 2.6840525893074973,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.029583333904544512
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.722308397293091,
"example_count": 600... |
495874eb-4497-47ae-be3c-60d8aa166e97 | easy | khushidahi | 2026-08-19 00:42:09.413655+00:00 | succeeded | 416deb7823b16a327a67857562144ff99997a9aa02889891e175302177deaf03 | 29,561 | null | """R252: deterministic late-hardening carry/reduce machine.
This candidate is deliberately diagnostic. It trains on evaluator-owned final
labels at several depths so that one shared transition must compose. The model
never converts N or X to native numbers and never implements multiplication or
modular reduction.
T... | {
"id": "495874eb-4497-47ae-be3c-60d8aa166e97",
"created_at": "2026-08-19 00:42:09.413655+00:00",
"db_md5": "969a54670ef11ba5c02bd3df83ccbc92",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "0c3d3efc-ff8a-412a-ae45-384aaca6dabc",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.731601789289539,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7608239362058082,
"example_count": 600... |
495942fa-05ca-4838-9d18-0af9f2e6699e | easy | Ranoobaba | 2026-08-19 07:49:20.262430+00:00 | succeeded | 86c29a833324f5c42c98d4936d2c087fc17112b2cb3b30d75234355f49654877 | 6,484 | 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": "495942fa-05ca-4838-9d18-0af9f2e6699e",
"created_at": "2026-08-19 07:49:20.262430+00:00",
"db_md5": "7d11e31f39169ce3db588bdccccca0af",
"submitter": "Syed Rayyan Ali",
"github_login": "Ranoobaba",
"run_id": "8bd5bdef-7ba5-4dbc-b434-918877c32b1b",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 6.449262172503959,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05166666701436043
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.466503620147705,
"example_count": 100,
... |
4959bcc7-9ac2-447c-a56c-6475b6912534 | easy | c-nnr | 2026-08-04 13:15:55.041754+00:00 | succeeded | 9c6ffb5b9b2d955a3dfc07f60bdd53537d8ac4d4aa643c59090af7cff9ecd4f3 | 6,440 | null | """One Layer Deeper E1 final-15 experiment: Embedding-scale sweep below 0.1 to test near-uniform initial predictions."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSp... | {
"id": "4959bcc7-9ac2-447c-a56c-6475b6912534",
"created_at": "2026-08-04 13:15:55.041754+00:00",
"db_md5": "8078127509518fc03352cac5550537f2",
"submitter": "c-nnr",
"github_login": "c-nnr",
"run_id": "f8e4f1ae-f8ec-415c-bbd0-363baea02831",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.513471245765686,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06833333149552345
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.578751802444458,
"example_count": 100,
... |
49616fc5-6baa-43bf-b1fb-da71acfdf341 | easy | Maxwells-Demons | 2026-08-07 02:31:03.422003+00:00 | succeeded | d480fff51bf952ba96ebb21d8bb075592c8a1797d11f8c0e26908edb96a329d0 | 10,996 | null | """v31: value-domain recurrence with a rational feature bank and gated readout.
The variable-modulus tiers evaluate x values never paired with that modulus in
training, so the transition has to be computed from the *values* rather than
looked up. Earlier digit-serial circuits were 16-250 soft layers deep and got
no g... | {
"id": "49616fc5-6baa-43bf-b1fb-da71acfdf341",
"created_at": "2026-08-07 02:31:03.422003+00:00",
"db_md5": "218bd601d6f397517fbaddbe3c840360",
"submitter": "Maxwells_Ayakashi",
"github_login": "Maxwells-Demons",
"run_id": "4f7f99e7-e878-45a5-b929-efdca61a9439",
"tier": "easy",
"dataset_id": "e4",
"st... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 1200,
"exact_accuracy": 1.0,
... |
4964a7c3-467d-40d2-ba48-7734a8c6d738 | easy | DDanlov | 2026-08-17 19:28:38.389335+00:00 | succeeded | 5f032de81856449f4c290cb8db4becd4eb06e0f3625a62b92bfe212d44b54a51 | 24,905 | null | import math
from typing import Optional
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark import Submission, assert_model_state, OptimizerBundle
except ImportError:
try:
from client import Submission, assert_model_state, OptimizerBu... | {
"id": "4964a7c3-467d-40d2-ba48-7734a8c6d738",
"created_at": "2026-08-17 19:28:38.389335+00:00",
"db_md5": "feba6d9a3c6050c96ad351429f15ffd8",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "2e7f2785-ac93-4951-83bd-3197c34a831d",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 236.7438901353256,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.09333333387970924
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 247.65904548986634,
"example_count": 100,... |
496aa240-6e08-4a9b-974d-9ece18b2056d | easy | chad-atexpedient | 2026-08-04 20:15:55.557765+00:00 | succeeded | db3dc5be024367fabb86e0367ad4d5b2d97c2e691f10691bbd6280ea8b9ad16b | 3,631 | null | """R000: immutable copy of the official AdamW baseline."""
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 = ... | {
"id": "496aa240-6e08-4a9b-974d-9ece18b2056d",
"created_at": "2026-08-04 20:15:55.557765+00:00",
"db_md5": "859b2a7c02ce6334691f84502bf9bb6e",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "bb17c519-a0ef-443d-b970-a95c675e50d3",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.9068320118241795,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.0207126044371737,
"example_count": 60... |
496e6181-dca8-4146-9497-3fe94e4a3eaf | easy | chad-atexpedient | 2026-08-18 01:28:46.232555+00:00 | succeeded | 836e2ce21cd40b7dfa8745f85eee86ec57ca345ec16969e16cf80b08a0c7d8d4 | 21,776 | null | """X94: supplied same-input prefix supervision.
When evaluator-supplied rows share exactly the same N and X but request different
T, the shorter row's final label is a genuine intermediate endpoint for the
longer row. X94 applies that supplied label directly to the longer row's matching
intermediate logits. It generat... | {
"id": "496e6181-dca8-4146-9497-3fe94e4a3eaf",
"created_at": "2026-08-18 01:28:46.232555+00:00",
"db_md5": "c51cd927de00ede295b1b1159dd81bbe",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "dfe5ed15-f45b-4132-a8bd-473e498f66dd",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.2866680820840166,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012083333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.275101092898792,
"example_count": 600... |
497b8bc5-3613-483e-a342-0eb03c3a57ba | easy | jordanrubin | 2026-08-09 02:41:51.226549+00:00 | succeeded | 23c96b25d2a5b82c71ed8ca6cf4fadcc3fc10ccfe43a16949682eade745a366c | 16,637 | 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": "497b8bc5-3613-483e-a342-0eb03c3a57ba",
"created_at": "2026-08-09 02:41:51.226549+00:00",
"db_md5": "550ddc9f9d2598d880f0a082c3722567",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "3e857072-b307-40e0-8d5e-3a1456261eac",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 3.2889664939088528,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007499999832361937
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.209855194583602,
"example_count": 600... |
497f815c-8140-4205-a5d2-b9321d59867b | easy | isaac0804 | 2026-08-04 16:10:38.333885+00:00 | succeeded | 1776c05a546b5c963e0e340093b78842300472392096a990504b719e39ad6620 | 3,740 | 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 = 12... | {
"id": "497f815c-8140-4205-a5d2-b9321d59867b",
"created_at": "2026-08-04 16:10:38.333885+00:00",
"db_md5": "32a6414edf492e9f1fa8fede4bc32722",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "0df4498b-3938-4ffa-aa3e-4b9b675d63d7",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 3.037359118461609,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018333333544433117
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.992464065551758,
"example_count": 100,... |
498c4199-c634-4526-be97-8fb8972be158 | easy | DDanlov | 2026-08-08 10:00:48.037039+00:00 | succeeded | 783ce94ea53716ce5aad3c4bacf25ccb017f7850d55b88b559657f1dac526575 | 13,315 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class RohanShampoo(Optimizer):
"""Self-contained RohanShampoo optimizer with eigenvalue ma... | {
"id": "498c4199-c634-4526-be97-8fb8972be158",
"created_at": "2026-08-08 10:00:48.037039+00:00",
"db_md5": "2ba8f88be8845d447ad26ddaba5fd4ed",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "811226c6-03a3-4c4e-828b-5331ee64fc84",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 104.5043116638488,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333333333333335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 112.12353515625,
"example_count": 100,
... |
499374fb-18e2-467e-a3c6-a005eaa1d38a | easy | DDanlov | 2026-08-17 22:07:00.705258+00:00 | failed | 1365b15b61a7760b944a409fe3b5b9513469b0a2938a444ba95a40960925c13f | 15,921 | null | """
Official Submission for One Layer Deeper Challenge
Architecture: Dual Continuous-Discrete Phase Space Port-Hamiltonian Torus (DCD-DEPH-Net)
- Continuous Momentum p_t in R^d (cotangent space smooth information waveguide)
- Discrete Balanced Ternary Register q_t in {-1, 0, +1}^d (discrete combinatorial state)
- Sympl... | {
"id": "499374fb-18e2-467e-a3c6-a005eaa1d38a",
"created_at": "2026-08-17 22:07:00.705258+00:00",
"db_md5": "a520b48b7ea230f47a1dfba834add8c7",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "f576bf3f-5581-4b8c-ab47-7dd03061be47",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
4995794d-aac0-4409-89e9-7af42d861608 | easy | sapient-sapiens | 2026-08-09 14:06:56.230123+00:00 | failed | f5fea4eac8239ff5b460e489d98f261db7c696db7af245677e34c5e18414e352 | 20,544 | null | """Near-500M learned adaptive recurrent model.
The input is treated only as categorical tokens. A one-time encoder creates
two equally-sized states; a tied reversible residual transition updates them.
A learned PonderNet hazard chooses the readout depth without reading a supplied
iteration count. Set OPTIMIZER_KIND ... | {
"id": "4995794d-aac0-4409-89e9-7af42d861608",
"created_at": "2026-08-09 14:06:56.230123+00:00",
"db_md5": "e6c796a6807b71e95d40256b3011971a",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "565c5367-32e7-412b-826b-e09f05708790",
"tier": "easy",
"dataset_id": "e1",
"status": "fa... | null
|
499d286f-f8d2-49f3-a318-ace5e37a304d | easy | TZZheng | 2026-08-06 18:00:31.424982+00:00 | succeeded | d02d8f9ae19672feba9dfad0e0d6a2e2433f7898250483fa352b7e4cb10a6e54 | 7,300 | 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": "499d286f-f8d2-49f3-a318-ace5e37a304d",
"created_at": "2026-08-06 18:00:31.424982+00:00",
"db_md5": "5b0ff109260af70cfad8542f8dce0cd4",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "951aa35d-dea6-4f35-ace0-74b46ac0b149",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.1436206698417664,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05333333043381572
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.381610870361328,
"example_count": 100,... |
499d61eb-46e2-4f7d-85d8-94b6b9712c6b | easy | AFeinfield | 2026-08-27 22:24:11.275439+00:00 | succeeded | 444fd2a000d681496ec4d509d38b8cba00dd5fde94ebd644e568ef3ced503182 | 12,166 | null | """157: 155 + comparator sharpness init (tau ~ |randn|*5+1 — random,
but in the regime where the ge gate starts decisive)
(randn*3.0 — still random; the decode no longer has to grow 100x).
151 notes: throughput/learnability pass.
Dynamic alignments (m <= batch N-width), carry passes 6 post-square / 3
in reduction, si... | {
"id": "499d61eb-46e2-4f7d-85d8-94b6b9712c6b",
"created_at": "2026-08-27 22:24:11.275439+00:00",
"db_md5": "7858042e75a990c0af771c93eb0e1bcf",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "f6944c27-25b6-42cc-a39c-13d2e756dd78",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.8263781666755676,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07999999821186066
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.7743152379989624,
"example_count": 100... |
49a2d380-5928-4c13-b693-5845b4d0dc52 | easy | yashkant | 2026-08-14 17:41:11.403126+00:00 | succeeded | 0c019fe8aee6a819d40fde1ecbf1a3f54feaa029b7d79eded7c5863e5c63f597 | 39,499 | null | """Round1786 evolutionary crossover of two controlled survivors.
The Round1780 W8/D2 carrier keeps its winning contrastive, exactly
antisymmetric learned set codes. The candidate additionally removes each
initial and recurrent Dykstra correction residual's coordinate mean before
the unchanged tanh bound, combining Ro... | {
"id": "49a2d380-5928-4c13-b693-5845b4d0dc52",
"created_at": "2026-08-14 17:41:11.403126+00:00",
"db_md5": "7948cc64a3f73c5906892895715d68a3",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "fec59b76-e115-4742-9682-6e618064d22d",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 3.002224087715149,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04437500238418579
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.517597198486328,
"example_count": 300,
... |
49a3d778-e485-44fc-80ba-c9c6743a6f9d | easy | DDanlov | 2026-08-08 10:26:29.273975+00:00 | succeeded | 8cc530aaa3cfd96917b619df350832d07b4d0a3e1aa4c6b3a74195c792f0eff0 | 15,561 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class RohanShampoo(Optimizer):
"""Self-contained RohanShampoo optimizer with eigenvalue ma... | {
"id": "49a3d778-e485-44fc-80ba-c9c6743a6f9d",
"created_at": "2026-08-08 10:26:29.273975+00:00",
"db_md5": "e8b5f06a26c5725fa3754b502958c7c2",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "27ab9842-a273-438e-b243-3ab6a59226b7",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 471.9893076724102,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 485.9944828358466,
"example_count": 600,
"e... |
49a9e08e-91bf-4acb-97b6-9415d873fc58 | easy | oupadhyay | 2026-08-15 22:49:17.013189+00:00 | succeeded | 4e30c03f3cc12749d4f5bb7efc79a87d2cac51472de678d064bf34c87567d0c5 | 4,317 | null | """Conditional structured-loss phase 5: hard75."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,TokenLossBatch,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,max... | {
"id": "49a9e08e-91bf-4acb-97b6-9415d873fc58",
"created_at": "2026-08-15 22:49:17.013189+00:00",
"db_md5": "043858e936845acfd3d4426475dfdaaa",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "fcc7d14c-b476-465c-8a72-ae25d50436b2",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.2968996927981626,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0020833333457509675
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.168151469508629,
"example_count": 60... |
49aee966-0edc-4cf4-b226-ea0996fe49bb | easy | shan23chen | 2026-08-08 02:05:48.435254+00:00 | succeeded | b64b888f6fe4119a2d6ff7f761549135e19c0b8252b324e4589c2d9d9fe86603 | 9,294 | null | """GPU-resident numeric-interface learner with trainable Fourier features.
This model hard-codes only the published decimal input/output interface. The
task mapping itself is represented by randomly initialized learned layers; no
recurrence, modular operation, or task-specific transition is supplied.
"""
from __futu... | {
"id": "49aee966-0edc-4cf4-b226-ea0996fe49bb",
"created_at": "2026-08-08 02:05:48.435254+00:00",
"db_md5": "ee52e4af9a51c15aafcda8e505e7dbe2",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "3fc69a6b-65bb-40c3-97ac-54a03d5899f5",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 4.3659446239471436,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004999999888241291
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.939016342163086,
"example_count": 600... |
49c24c6a-062f-49c0-a3c6-671648aabb81 | easy | sapient-sapiens | 2026-08-18 11:02:38.806217+00:00 | succeeded | ea8517f64930864b95a126616a0868c9a79b3fff02cd2d55f68747e681075651 | 26,966 | null | """v44 adaptive baseline: one encoder, four-layer T-free core, bounded halt.
Locked routing: field-local one-layer encoder; recurrent core sees only N/X;
T is pooled solely for the controller; decode once on a halt-weighted state
mix. Stopping support is 8 depths, not 64. T is never parsed in forward or
the loss.
"""
... | {
"id": "49c24c6a-062f-49c0-a3c6-671648aabb81",
"created_at": "2026-08-18 11:02:38.806217+00:00",
"db_md5": "c0466ed712e9b6a105b0ca7414938273",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "6e466f03-549e-499d-a9ab-4c426f19c7d8",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1508317720123165,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000037252904
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.154490658119655,
"example_count": 600... |
49c6694a-bf18-4940-a3f0-d20af9e6fc97 | easy | chuk-uzowihe | 2026-08-06 22:22:27.299790+00:00 | succeeded | e1a65d257d80b7bf0de5b6bfd77434d36f2d947299422df8224705c37fece35d | 31,441 | 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": "49c6694a-bf18-4940-a3f0-d20af9e6fc97",
"created_at": "2026-08-06 22:22:27.299790+00:00",
"db_md5": "bcd95166f10f9ee59cc10f2e3cf32311",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "6dcc0471-3be9-46fc-ac9d-057da9d4c449",
"tier": "easy",
"dataset_id": "e4",
"status": "... | {
"score": {
"mean_loss": 7.256207974623277,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.001875
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.394249519900296,
"example_count": 1200,
... |
49cc416c-82ba-4286-8404-05f4c9624ed1 | easy | erdavis0 | 2026-08-19 08:35:11.384727+00:00 | succeeded | 62bce0531fbe4b33cc9234df7ee22d96dc687d5abc88cc4b82262ee666d18e89 | 9,265 | null | """Raw-vocabulary recurrent operator with immutable deep supervision."""
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,
TokenLossBa... | {
"id": "49cc416c-82ba-4286-8404-05f4c9624ed1",
"created_at": "2026-08-19 08:35:11.384727+00:00",
"db_md5": "f912acb716edc73a43c9211f7154a6eb",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "fc3e37a5-390b-4bf8-8253-8d7899c3e25f",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.271021905624864,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.018125000000000002
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2705093574564046,
"example_count": 800... |
49cfd42c-4764-4683-8e84-eb1e290bd68f | easy | jordanrubin | 2026-08-06 19:40:57.802942+00:00 | succeeded | 09586f90ffa296e3bdbb25e6a0b059c7579344900d2f7d9eb12b0e902c4848f6 | 15,390 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "49cfd42c-4764-4683-8e84-eb1e290bd68f",
"created_at": "2026-08-06 19:40:57.802942+00:00",
"db_md5": "ef7d67d872a65aa96e81ca53991ab27d",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "9ee56c79-defd-45ae-bb61-e6580ad92244",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 3.144348778035648,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.27499999447415274
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.273178229071676,
"example_count": 300,
... |
49d81b08-8439-4f86-8260-01b6ec43d666 | easy | DDanlov | 2026-08-28 18:41:00.681950+00:00 | succeeded | f34590c11f9562d1cce74ef1e3142d9156a88fa22385b1af72d1d4cdc4cd78be | 14,969 | null | """
Model B Variant: Clean-Slate Bilinear Hadamard (WITHOUT z0 reinjection)
Config: D=384, H=6, T_macro=2, T_micro=2, lr=0.003
"""
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 b... | {
"id": "49d81b08-8439-4f86-8260-01b6ec43d666",
"created_at": "2026-08-28 18:41:00.681950+00:00",
"db_md5": "48c5d879ca78bb581c7f667ba40cf5ac",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "23b7a26e-9047-49b8-9126-2dd2de00dbbe",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.4218042174220473,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4551858074729216,
"example_count": 60... |
49d8761e-22a5-4850-9aca-df0eb3fcc40f | easy | oupadhyay | 2026-08-15 22:40:44.913766+00:00 | succeeded | 927e29bdb8bf02bf526ee0411b7fceb6218d6d87b4aacacd4810822f0e32c17d | 4,316 | null | """Conditional structured-loss screen: hard25."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,TokenLossBatch,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,max_... | {
"id": "49d8761e-22a5-4850-9aca-df0eb3fcc40f",
"created_at": "2026-08-15 22:40:44.913766+00:00",
"db_md5": "35fea2a9bbee43d0bda9b76800373a66",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "27177d49-0c37-4992-88a3-cd87f18cbf57",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.336502308635249,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2243349156572143,
"example_count": 600... |
49da13f4-6136-42f8-863f-5e6f1c73345a | easy | isaac0804 | 2026-08-06 16:00:32.313982+00:00 | failed | 29ddaf979268d32bab46c1d71a188811c5cfc704a39ce723f42623d2b379baca | 10,648 | null | """E3 candidate #8: same-modulus SupCon-style contrastive regularizer on
pooled row representations -- a soft, gradient-only push, not a structural
capacity/routing/attention change like every prior candidate this session.
Context (see docs/findings.md "Session log: subset-size sweep, real vs. fake
information bottlen... | {
"id": "49da13f4-6136-42f8-863f-5e6f1c73345a",
"created_at": "2026-08-06 16:00:32.313982+00:00",
"db_md5": "0dcd6c453a68f99c70d897a98ec9679e",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "94d88171-27e8-452e-8f3a-55bb829faa9a",
"tier": "easy",
"dataset_id": "e3",
"status": "faile... | null
|
49e20e51-3bee-47c3-ab92-d2a72a296390 | easy | sapient-sapiens | 2026-08-13 20:11:54.521727+00:00 | succeeded | 8df35e8a6be03e145bb2660555871b16c28d4c753891f99b03abf6c925492a74 | 5,972 | null | """Paper-style additive-input-injection loop for fixed-T=2 E3."""
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, assert_model_state
WIDTH = 1536
HEADS =... | {
"id": "49e20e51-3bee-47c3-ab92-d2a72a296390",
"created_at": "2026-08-13 20:11:54.521727+00:00",
"db_md5": "c4e3ea92808200c6cda4b41794f18138",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "a24fe071-d70a-4100-8e7a-82ec2367903b",
"tier": "easy",
"dataset_id": "e3",
"status": "su... | {
"score": {
"mean_loss": 2.131129382574258,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00875
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1264232170897768,
"example_count": 800,
"... |
49e94baa-e4c1-45f6-ab6f-19940982ba50 | easy | khushidahi | 2026-08-01 23:30:00.250232+00:00 | succeeded | 8445c527155424ec250693fd87953d6f94ec76e2fb7631b37d80f2e934ea0069 | 6,935 | null | """Prelude -> recalled shared core -> coda model for One Layer Deeper.
R2 architecture reset:
- use a real one-block prelude and one-block coda;
- apply one shared recurrent Transformer core four times;
- inject immutable prelude context at every recurrent update;
- use sandwich RMS normalization for stable repeated a... | {
"id": "49e94baa-e4c1-45f6-ab6f-19940982ba50",
"created_at": "2026-08-01 23:30:00.250232+00:00",
"db_md5": "ae2f0d97c3008969c47970d57082669b",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "a6ea5ea7-8eef-4339-aff4-759e16e03446",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.5913734436035156,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004999999888241291
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5975637435913086,
"example_count": 60... |
49ebbc06-6c5e-46c5-b2ec-593f38debeb8 | easy | khushidahi | 2026-08-06 04:40:34.746697+00:00 | succeeded | ac4e1b694a85da2453b25b7685cf0256ad335c23f2257db2610d4ac1836c3430 | 14,399 | null | """R6 structured decimal-workspace model for One Layer Deeper.
The model does not implement multiplication, modular reduction, or the public
recurrence. It uses the public tokenizer structure to build learned, right-
aligned decimal tapes for N, X, and T, then applies a shared neural transition
to a mutable work tape.... | {
"id": "49ebbc06-6c5e-46c5-b2ec-593f38debeb8",
"created_at": "2026-08-06 04:40:34.746697+00:00",
"db_md5": "bcc6b472a5b533ce756ec037f5aac1ad",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "a8298223-07db-4845-b142-1e407cee8bf3",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.230910539627075,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011250000447034836
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.23252010345459,
"example_count": 300,
... |
49f21704-5af5-4b8d-988c-ccf28dcd276f | easy | TZZheng | 2026-08-07 00:05:16.245520+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": "49f21704-5af5-4b8d-988c-ccf28dcd276f",
"created_at": "2026-08-07 00:05:16.245520+00:00",
"db_md5": "16ec378d6d82482653b34c09d2557d3f",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "eb6bfd6d-fff7-4d41-9de8-5755035e9693",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.91956627368927,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07499999925494194
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9264726638793945,
"example_count": 100,
... |
49f5a937-ae99-4a3a-a268-a6ee12520e69 | easy | richardcepka | 2026-08-23 20:59:37.120373+00:00 | failed | 7f8065fc7630ae1ba5d29069ec3cdc3e0dccfc97bb8f075f5a44a5fcaa201294 | 58,067 | null | """Recurrent digit-register model for One Layer Deeper.
Given a modulus N, a starting value x, and a step count T, predict the residue after squaring modulo N exactly T times.
repo/README.md
A looped transformer whose recurrent state is a decimal digit register. The loop is nested: an
outer step is one squaring, an... | {
"id": "49f5a937-ae99-4a3a-a268-a6ee12520e69",
"created_at": "2026-08-23 20:59:37.120373+00:00",
"db_md5": "f83c7a678398e24910f6fd38bfff9fec",
"submitter": "Richard Cepka",
"github_login": "richardcepka",
"run_id": "f7724e5f-3ce0-44c7-87f5-e2b7170a5716",
"tier": "easy",
"dataset_id": "e5",
"status": ... | null
|
49fee740-6731-496b-bfbb-a51dba3d0fd5 | easy | liam-gb | 2026-08-13 02:32:01.680892+00:00 | succeeded | f6087d31cb2d28e4fd9ea847cb46c143d974424bd942a493235224e324d83a76 | 14,650 | null | """rns_pm_gather_ct_widet_lrt100: tables at lr 1e-1.
Base: dataset-sized digit head, dead loop tail trimmed.
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 ... | {
"id": "49fee740-6731-496b-bfbb-a51dba3d0fd5",
"created_at": "2026-08-13 02:32:01.680892+00:00",
"db_md5": "50a13b7d6b4def5bc67d6d6befedc12c",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "8bc8ad04-1e7f-4fcf-a865-508d9e32815e",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.487042430083494,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.043333333420256775
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.860196113586426,
"example_count": 600,... |
4a096b54-ccb3-4b06-aca5-e70df23baca6 | easy | sapient-sapiens | 2026-08-09 13:56:49.987322+00:00 | failed | a798af32018ddae3d2e3ae94f304548273b3fea269b9cf7ba705dc08bd0168b7 | 20,698 | null | """Near-500M learned adaptive recurrent model.
The input is treated only as categorical tokens. A one-time encoder creates
two equally-sized states; a tied reversible residual transition updates them.
A learned PonderNet hazard chooses the readout depth without reading a supplied
iteration count. Set OPTIMIZER_KIND ... | {
"id": "4a096b54-ccb3-4b06-aca5-e70df23baca6",
"created_at": "2026-08-09 13:56:49.987322+00:00",
"db_md5": "c6cd4a01c03aeeb6324d8524ea8c67d8",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "0df298ad-7d22-476d-8528-8e0e3ad2ab62",
"tier": "easy",
"dataset_id": "e1",
"status": "fa... | null
|
4a0a16dd-ea7f-4b8a-98be-6f72d4517b8b | easy | shan23chen | 2026-08-08 02:10:49.720501+00:00 | succeeded | c8ee34a36b46e675bbd9ad19a6b98977fc0fa5d763d802d2528b0adcd913c6e6 | 20,379 | null | """Generic recurrent Transformer with normalized numeric and CRF objectives.
The model is deliberately task-generic: it learns a transition over prompt and
workspace tokens instead of embedding the public modular-squaring algorithm.
A learned linear-chain CRF models dependencies among adjacent output digits;
its endpo... | {
"id": "4a0a16dd-ea7f-4b8a-98be-6f72d4517b8b",
"created_at": "2026-08-08 02:10:49.720501+00:00",
"db_md5": "a0861bfdc517eb06be7ef7c36b91f05c",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "830fa2ee-af5c-4ce5-80d7-cc4cd899f545",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.9902079105377197,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012500000186264515
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.946104049682617,
"example_count": 600... |
4a0e1670-212b-4790-ae15-8045e9cb3cf1 | easy | erdavis0 | 2026-08-22 02:48:17.497376+00:00 | succeeded | 19038191486b389bca73abae38e37218f120f288e18de2ef93b525dab864dba8 | 5,372 | null | """Second-order positive-feature linear attention."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
WIDTH = 64
HEADS = 4
EXPANSION ... | {
"id": "4a0e1670-212b-4790-ae15-8045e9cb3cf1",
"created_at": "2026-08-22 02:48:17.497376+00:00",
"db_md5": "d0f6481a29fcf3f19730b9386572efad",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "7c743608-9e6d-4ae2-80db-6747daa61491",
"tier": "easy",
"dataset_id": "e7",
"status": "succeeded",... | {
"score": {
"mean_loss": 9.07229471206665,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.040374333038926125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.276894569396973,
"example_count": 85,
... |
4a142957-48b2-4f2a-ae3c-5166adea2cc1 | easy | isaac0804 | 2026-08-04 16:40:34.757738+00:00 | succeeded | 946161a2dd374e652b979936e9f9aba3d5cb0b8b1b6ae58f77775e2c5249978f | 6,910 | null | """Weight-tied looped Transformer that reads T from the prompt to set its own depth.
The task is x_t = x_{t-1}^2 mod N for T steps. A single fixed-depth block has no
mechanism to represent "apply this operation T times" and cannot extrapolate to
depths unseen in training. This submission instead applies one shared blo... | {
"id": "4a142957-48b2-4f2a-ae3c-5166adea2cc1",
"created_at": "2026-08-04 16:40:34.757738+00:00",
"db_md5": "d7a688e1999e1a65266208ed0c343b81",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "ab71d45a-3892-4cc7-8e87-26cf3c816d69",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.9641841053962708,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.026666666381061077
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.985734224319458,
"example_count": 100... |
4a14e1ec-0c2c-4419-83b9-06e90c289079 | easy | alirezashirvani-jr | 2026-08-29 12:26:50.988189+00:00 | succeeded | d0b5a4df581c6dbf80e071837f5ebba8307acb60e3f3a90aa679b9a403355bc9 | 32,601 | null | from __future__ import annotations
import math
import time
from contextlib import nullcontext
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
MODEL_... | {
"id": "4a14e1ec-0c2c-4419-83b9-06e90c289079",
"created_at": "2026-08-29 12:26:50.988189+00:00",
"db_md5": "18fbce8b1b23f3caadcb50ca8bbc70f0",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "96c17047-5e41-4471-ade1-84783c3cc054",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 2.978529497108481,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03083333378036817
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.0065066814422607,
"example_count": 600,... |
4a193466-c532-4a26-a655-4df6600d24f4 | easy | pepinu | 2026-08-26 17:41:50.403060+00:00 | succeeded | 74307ba76d035d71e8330e51db839434ca24ae8d3efbab3f4b12dd2ec11dd273 | 46,270 | null | """Squaring mod N as a learned permutation on CRT components.
WHY THIS SHAPE. (Z_N, x) is a commutative monoid, so every finite-dimensional
linear representation splits into multiplicative characters and squaring is
diagonal there -- the component basis is the only one in which this operation is
shallow. Two consequen... | {
"id": "4a193466-c532-4a26-a655-4df6600d24f4",
"created_at": "2026-08-26 17:41:50.403060+00:00",
"db_md5": "d7f34595d7f996ce3ab1db374501668a",
"submitter": "pepinu",
"github_login": "pepinu",
"run_id": "1d145c26-e7fe-44fa-bf97-43847fe1897d",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 0.009301922264287877,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.009315639901321565,
"example_count": 600,
... |
4a19620b-0635-45ed-8157-cc76f02484d5 | easy | khushidahi | 2026-08-23 21:36:48.630468+00:00 | succeeded | b5e8ffb97b8fbfd72128df2b0bb59a035bcc533286044a1225849354b952e5da | 6,240 | null | """Categorical, contest-compliant adaptation of official dt_net_recall_1d."""
from __future__ import annotations
from random import randrange
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
a... | {
"id": "4a19620b-0635-45ed-8157-cc76f02484d5",
"created_at": "2026-08-23 21:36:48.630468+00:00",
"db_md5": "5f6e9f075402c70d9065222043a4de50",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "aa8d8248-ca3e-4cc3-b107-758aed34df26",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 6.142688274383545,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.11554054543375969
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.232499122619629,
"example_count": 60,
... |
4a27179e-1413-4275-9392-a8dcb9518ec8 | easy | sapient-sapiens | 2026-08-19 11:22:12.610663+00:00 | succeeded | a49fb11a941ad5b7b199ba49b7305d04fb0838b3cabead963bc7b2a9fd6a31e6 | 33,317 | 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": "4a27179e-1413-4275-9392-a8dcb9518ec8",
"created_at": "2026-08-19 11:22:12.610663+00:00",
"db_md5": "c5798ff07416602bd68622d18fe67528",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "9676cce8-54e0-4ccc-b5ac-ed9dab418e26",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.70554689327092,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000037252903
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.708948810004333,
"example_count": 600,
... |
4a28779b-06bc-4948-b2db-6fa6eda6492b | easy | yashkant | 2026-09-01 00:35:55.299623+00:00 | succeeded | 6af2d1a14a490840ff32315852351e712c97abeaecc8d1a38d71bbf10015b736 | 111,812 | null | """One Layer Deeper submission -- a learned register machine.
Architecture (all outputs produced by learned parameters; no task arithmetic on the answering path):
a bank of R=6 digit registers (W=16 positions each) carried as DISCRETE tokens between micro-steps: five
data registers and one COUNTDOWN register holding t... | {
"id": "4a28779b-06bc-4948-b2db-6fa6eda6492b",
"created_at": "2026-09-01 00:35:55.299623+00:00",
"db_md5": "a8a79f8311a2da8d92742805615f437b",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "962804ca-5620-4526-b5d0-1e6f0f4586c4",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 4.224404811859131,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.319845676422119,
"example_count": 100,... |
4a297dd9-ed22-453d-b347-9b6dfc6f26ef | easy | karanganesan | 2026-08-04 01:28:04.475993+00:00 | succeeded | 974ec1eb359c34e313908a0d5394941de7eeb17b1078c99f1a45c9f5374e592e | 11,980 | 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": "4a297dd9-ed22-453d-b347-9b6dfc6f26ef",
"created_at": "2026-08-04 01:28:04.475993+00:00",
"db_md5": "bc7ef148a2426f1c6c90e28e9e05314c",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "2f8532f0-2fb8-45f3-a702-3a9ae90d4320",
"tier": "easy",
"dataset_id": "e2",
"status": ... | {
"score": {
"mean_loss": 2.3760253190994263,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012916667386889458
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.317185878753662,
"example_count": 300... |
4a2bf432-bea4-44c3-8f52-c2c01aa89de4 | easy | erdavis0 | 2026-08-27 01:01:48.476664+00:00 | succeeded | ca6f1b5c93401e93c705f4189e451ef833a9103266142696e3b482501ff8d97a | 31,305 | null | """Typed-prefix reversible transition B1: CE plus selector-oracle KL.
Public delimiter syntax routes opaque N, X, and T fields into three strict
functional roles. X initializes only an evolving auxiliary tape, N initializes
only an invariant environment lane, and T reaches only a learned monotone
prefix controller. ... | {
"id": "4a2bf432-bea4-44c3-8f52-c2c01aa89de4",
"created_at": "2026-08-27 01:01:48.476664+00:00",
"db_md5": "3f9846560514c463a6ae282585eac958",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "93d8cbd0-966e-4158-93a1-0dfdc7406cc2",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.5259793420809364,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0062500000372529025
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5351547788611444,
"example_count": 6... |
4a2d5562-3d7c-4f3b-8248-de8fe0531af6 | easy | 0xDEADD06 | 2026-08-12 14:23:20.503663+00:00 | succeeded | f01bb3b3d9d7eabc853117babf8eec794788ac101c6c632dfa1132c90bfce9b0 | 5,741 | null | """ParaLSTM: recurrent LSTM with weight-tied passes for algorithmic tasks."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn, Tensor
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL... | {
"id": "4a2d5562-3d7c-4f3b-8248-de8fe0531af6",
"created_at": "2026-08-12 14:23:20.503663+00:00",
"db_md5": "f9b99930c067c683803467c0b38c64ab",
"submitter": "0xDEADD06",
"github_login": "0xDEADD06",
"run_id": "e65e9501-6e2e-4b8e-9161-81f611780d3c",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 2.5180910057151396,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333469927311
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7015478623020273,
"example_count": 60... |
4a31fdac-e9ef-4f5f-b4b4-210be88a1a12 | easy | yashkant | 2026-08-23 02:28:27.674127+00:00 | succeeded | 4deffb934ba605f61e5c1da8a48f124d83d74b5a60d0e0c2516bb8bd3e2a777b | 6,937 | null | """One Layer Deeper — CRT residue-register machine.
State = per-modulus registers p_m over Z_m, init onehot(x mod m). Shared
transition p_m <- S_m p_m (S_m learned column-stochastic) applied T times;
since (y^2 mod m) depends only on (y mod m), each register exactly tracks
y mod m under repeated squaring -> exact and ... | {
"id": "4a31fdac-e9ef-4f5f-b4b4-210be88a1a12",
"created_at": "2026-08-23 02:28:27.674127+00:00",
"db_md5": "603eab355042488c60e2ded417016c92",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "c44e5704-0f7b-45b6-9790-77ead7c3b8e8",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 2.0571548342704773,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.09999999776482582
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9261702299118042,
"example_count": 100... |
4a3998ed-fda6-4e9b-9376-d852a5046da5 | easy | sapient-sapiens | 2026-08-31 01:46:46.967256+00:00 | succeeded | 38dc7f5c14d09ea479317a7dde6b4a72a91aa451ed933ff8bb8ed84984d74453 | 16,031 | null | """Eligible E5 live canonical-X Horner operator with strict S-to-A routing.
The tied S operator receives full immutable N, the live current X workspace,
current A/S, and the corresponding live X digit in one fixed role lane. The
initial X workspace is encoded in the same token-only digit chart used by A
and the readou... | {
"id": "4a3998ed-fda6-4e9b-9376-d852a5046da5",
"created_at": "2026-08-31 01:46:46.967256+00:00",
"db_md5": "f7ae4ad70b944faf8d23ed10a85fe84f",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "7f57c1e4-485e-4404-a325-1e5107024739",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1389297014259236,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1390893352405906,
"example_count": 60... |
4a3c5282-e7c6-4c15-8272-5cc5caeb2882 | easy | lzy54 | 2026-08-29 08:54:47.948749+00:00 | succeeded | f967c3bfea2806a17090394364d66f815812d86d38acc90cb7b0b8126f881c01 | 4,539 | null | """Phase 2 C2: shared block x4 with residual concat recall."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEAD... | {
"id": "4a3c5282-e7c6-4c15-8272-5cc5caeb2882",
"created_at": "2026-08-29 08:54:47.948749+00:00",
"db_md5": "649789da2b5f7837ff738282292ec92c",
"submitter": "lzy54",
"github_login": "lzy54",
"run_id": "cc4c44c8-5454-4fcd-9d89-cced56821d31",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.1630684534538593,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1668906436372763,
"example_count": 60... |
4a3ee3a6-ffbb-42fa-b540-0394100ece42 | easy | k-penchev | 2026-08-22 21:15:54.892902+00:00 | succeeded | f5404df3aa942d35a9427f79a19fafe9353da6a797b9e43b14531a38c6ae5c98 | 4,579 | null | """Two-loop tied Transformer with digit places and immutable input recall."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL... | {
"id": "4a3ee3a6-ffbb-42fa-b540-0394100ece42",
"created_at": "2026-08-22 21:15:54.892902+00:00",
"db_md5": "0b95a146e5464e4b69864fba7273c389",
"submitter": "Kaloyan Penchev",
"github_login": "k-penchev",
"run_id": "eeec81b4-3ab1-46cd-96df-796ad52177ec",
"tier": "easy",
"dataset_id": "e3",
"status": "... | {
"score": {
"mean_loss": 4.575880945712221,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000004656613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.115441287421458,
"example_count": 800,... |
4a48aa67-6cf5-4556-a5e3-bfb1ada8c496 | easy | Dandandan | 2026-08-05 22:22:46.506031+00:00 | succeeded | cf63a5cf6a2f46bb38abff9e855f809b23cb30f1ed38de4b537f9579a8903f40 | 41,151 | null | """End-to-end recursive function learner for One Layer Deeper.
A population of randomly initialized polynomial register networks is reused
exactly once per requested outer step. The evaluator trains every coefficient
and the evidence selector through its normal forward/backward/optimizer loop.
There is no recurrence-... | {
"id": "4a48aa67-6cf5-4556-a5e3-bfb1ada8c496",
"created_at": "2026-08-05 22:22:46.506031+00:00",
"db_md5": "67701dfb09c9f4d71ad540e59bfed9f1",
"submitter": "Daniël Heres",
"github_login": "Dandandan",
"run_id": "3b90cd65-e0f8-48ee-abb7-26c7058e5f3b",
"tier": "easy",
"dataset_id": "e2",
"status": "suc... | {
"score": {
"mean_loss": 1.1421220733609516e-05,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.1438210094638634e-05,
"example_count": 300,
... |
4a4a2da7-06ce-4eeb-97c8-541679cf54cd | easy | sankalp1999 | 2026-08-04 12:12:12.945543+00:00 | succeeded | 467a98e7aae237d08fe0f3eb068cd8bebdccb04e3c668ebbcd297c94ff4deb5b | 7,511 | null | """Latent recurrence: encode once, apply a shared squaring map T times, decode.
v4 put the loop inside a transformer and underfit badly (train acc 4% after
4,432 steps) because attention cost ~7ms/step at 13 tokens and bought nothing.
v2 fit the train set perfectly with a cheap flattened encoder but memorized,
because... | {
"id": "4a4a2da7-06ce-4eeb-97c8-541679cf54cd",
"created_at": "2026-08-04 12:12:12.945543+00:00",
"db_md5": "ffc5c3f22dbffb18f9c58cb8fb763d29",
"submitter": "Sankalp ",
"github_login": "sankalp1999",
"run_id": "0c114fa5-3351-4633-8a5e-7bc43c7d89a2",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 7.009697675704956,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07333333231508732
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.410889625549316,
"example_count": 100,
... |
4a4f65c2-7f87-4c73-b65b-34579dc5f39a | easy | DDanlov | 2026-08-18 09:08:14.706152+00:00 | succeeded | 2daa3dabbf47e1f004651c2e7895944312e775b363dd9a9f753435d4e887457f | 23,511 | null | """
Official Submission for One Layer Deeper Challenge
Architecture: 4-Layer Recurrent Core Pure No-Norm No-Skip DKS Reasoning Model:
- 4 Distinct Pure No-Norm No-Skip Transformer Layers in the Recurrent Core: L1 -> L2 -> L3 -> L4
- Unrolled T times for serial modular squaring
- dim=512, 16 heads, d_ff=1024, RohanShamp... | {
"id": "4a4f65c2-7f87-4c73-b65b-34579dc5f39a",
"created_at": "2026-08-18 09:08:14.706152+00:00",
"db_md5": "a45a27d6e56da3f75f753d1345b5cf43",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "4852b864-2b0f-4beb-baff-40b276b3ffa3",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.390997345915335,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05500000149011612
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3307539144558693,
"example_count": 100,... |
4a534ace-f5d9-454a-80b6-0720202be903 | easy | lpbb | 2026-08-21 12:59:50.135122+00:00 | succeeded | 3d458f461afc4618b0d07f83450f431cf11c581f52e2fa81dfd2537a28727949 | 12,545 | null | """S0 + RoPE + abacus + looped block, with *input-adaptive* depth: instead of
a fixed `num_repeats`, each sample's T is parsed from its own token stream
(the digits after the `T` marker) and the shared block of `num_layers` layers
is applied exactly T times for that sample. One block pass = one modular
squaring step, s... | {
"id": "4a534ace-f5d9-454a-80b6-0720202be903",
"created_at": "2026-08-21 12:59:50.135122+00:00",
"db_md5": "56c9cccf89cf4e4c0e87a1200916934f",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "2999c3c1-3658-4a2c-912d-7bbf14d34a8a",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 3.315385389506371,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.621983528137207,
"example_count": 100,... |
4a603ee8-2747-407e-a25e-89e6b615d12f | easy | viridale | 2026-08-21 22:23:41.989255+00:00 | succeeded | 9e00472fa1d55f3f31a78f3bec8cd9f07a3cc15aadd9742c3dcf262e6a5aaaad | 73,346 | null | """NV046: factored learned-table evaluation.
hypothesis: NV045's official E1 T64/OOD-N T32 result is evaluation truncation:
the artifact overrides the official 512-row evaluation batch with 128 and
repeatedly decodes each gathered categorical arithmetic row through tiny
matrix multiplies. Restore one-batc... | {
"id": "4a603ee8-2747-407e-a25e-89e6b615d12f",
"created_at": "2026-08-21 22:23:41.989255+00:00",
"db_md5": "eade56436e455c2b88a467a8b7c2b937",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "3e83ed80-ee88-4093-afb1-bf5183d32dc7",
"tier": "easy",
"dataset_id": "e4",
"status": "succe... | {
"score": {
"mean_loss": 2.384185637790465e-07,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.384185626673852e-07,
"example_count": 1200,
... |
4a616827-c164-4700-a4f5-c1174c228a0b | easy | ehonig | 2026-08-21 01:17:58.709993+00:00 | succeeded | 66f84915a4e39dce066e348521e26897daeb379b8c72116cfbb8ed3a25a44c7d | 31,878 | 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": "4a616827-c164-4700-a4f5-c1174c228a0b",
"created_at": "2026-08-21 01:17:58.709993+00:00",
"db_md5": "a47092414289ee3f0cbb9fe119e076e4",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "2fd166af-4657-4b6f-bfcb-2a3b32b02c72",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.265147566795349,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.020416666753590107
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.260310411453247,
"example_count": 300,... |
4a62e290-4a52-4b41-957f-c621c99f1dd9 | easy | KaustubhKumar05 | 2026-08-25 16:16:48.190358+00:00 | succeeded | 66c0d17ac95330bfa44c7d3f372988009e3ed4e7cbcca72adeca514afd3de49f | 14,717 | null | """Neural-GPU-style conv GRU, v6: v2s + relaxed weight sharing.
Four cell copies cycled by iteration index (Kaiser & Sutskever's "crucial"
trick for multiplication). A pull-together penalty on copy divergence is
returned as auxiliary and added to the loss, annealing the copies toward one
shared cell while keeping earl... | {
"id": "4a62e290-4a52-4b41-957f-c621c99f1dd9",
"created_at": "2026-08-25 16:16:48.190358+00:00",
"db_md5": "cebdbe1f9f283b2f115dc4641a0cdd29",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "257978a0-36cd-4af7-8534-ceca360ebb05",
"tier": "easy",
"dataset_id": "e8",
"status": "succee... | {
"score": {
"mean_loss": 2.641118288040161,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01724598975852132
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6695621013641357,
"example_count": 85,
... |
4a63c99c-e843-4ef4-b527-5ae30fa3b369 | easy | AdityaVG13 | 2026-08-15 11:36:14.506113+00:00 | succeeded | 9003a55c0e2c1c873e9b883a148f8c6dd671c37eed2343c98efaabafe340f4eb | 109,416 | null | """AEGIS V12 v12_fc_join06: frontier-carry with JOIN_WEIGHT=0.06.
One Layer Deeper submission candidate. All task transitions remain learned
end-to-end. The deterministic shell only identifies marker-delimited roles,
constructs a finite semantic tape, schedules learned interval programs, and
applies a trivalent Safe/U... | {
"id": "4a63c99c-e843-4ef4-b527-5ae30fa3b369",
"created_at": "2026-08-15 11:36:14.506113+00:00",
"db_md5": "8367f722e6d8cf618098d348caebe7bd",
"submitter": "AdityaG",
"github_login": "AdityaVG13",
"run_id": "044c72d1-2904-46bc-a724-b396ae89855f",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 1.9264036417007446,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6850000023841858
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.6606953144073486,
"example_count": 100,... |
4a6bc0ae-95ef-481f-be79-4febba2eef0b | easy | khushidahi | 2026-08-22 01:28:05.397667+00:00 | failed | 6f76e7df3b8017d37e4dabab1fea5ddc636585cf585476b5db920a65e62a4580 | 18,865 | null | """R395: minimal Universal-Transformer functional-root control for M5.
This candidate adapts the released Universal Transformer to the competition's
separate-output interface. Numeric prompt symbols are only rearranged onto
equal-length, right-aligned categorical N and X tapes. A third tape supplies
parallel learned... | {
"id": "4a6bc0ae-95ef-481f-be79-4febba2eef0b",
"created_at": "2026-08-22 01:28:05.397667+00:00",
"db_md5": "101ff2c6960b328e583c4dc5ad0f486c",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "f35cc7cc-b2df-4a81-a5fa-dcf4fc9458d2",
"tier": "easy",
"dataset_id": "e5",
"status": "fail... | null
|
4a6c30ae-b165-4f8b-b0cd-98bb97b6aaf6 | easy | repst | 2026-08-21 23:05:08.407262+00:00 | succeeded | 4326e40fd2e032d52b3229145a6a5951c542d3b505d22a14c5a3891fc5dc813d | 260,078 | null | # R489: unified learned mesocode with relation-free reduction.
#
# LEARNED: prompt-token/digit lexicon, semantic field route and depth address,
# successor geometry, evidence-grounded instruction meanings, variable-length
# MDL-selected program, and output lexicon. Medium and Hard ultimately execute
# a hardened progr... | {
"id": "4a6c30ae-b165-4f8b-b0cd-98bb97b6aaf6",
"created_at": "2026-08-21 23:05:08.407262+00:00",
"db_md5": "fae7f0028c8e85217c7e26fe4dfe1551",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "a729b683-68e0-4da0-9988-064e7e04069a",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 0.32189539074897766,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.3245851397514343,
"example_count": 100,
"ex... |
4a6cc6d8-d305-493e-9366-c319a98ed74a | easy | shreyash-chonkie | 2026-08-26 14:12:34.611509+00:00 | succeeded | 7b1cdc106e28b22155acdfaf10e17febfc968974750e49ffce238e597494ef4e | 12,097 | null | """Ternary register machine with hybrid Muon/AdamW optimization."""
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,
)
STAT... | {
"id": "4a6cc6d8-d305-493e-9366-c319a98ed74a",
"created_at": "2026-08-26 14:12:34.611509+00:00",
"db_md5": "40dda7eb9ab19af976e61791ffb0c6ac",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "163445ff-392b-4203-b019-cdc8d00cc894",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 1.8669543266296387,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03666666615754366
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.7867341041564941,
"example_count": 100... |
4a6fbf8d-3e76-4c8b-8403-2a432e7f0c91 | easy | LauraGomezjurado | 2026-08-30 22:26:36.548492+00:00 | succeeded | 3408701238f54d65930556a87eb96d126387c2e42db3b7fc2f6dee201a0a7d1c | 13,422 | null | """Looped depth-recurrent Transformer for repeated modular squaring.
Design notes
------------
The evaluator reads answers off the *last* ``target_len`` prompt positions, so
answers are right-aligned with the least-significant digit at the final valid
position. Digit significance is therefore a fixed offset from the e... | {
"id": "4a6fbf8d-3e76-4c8b-8403-2a432e7f0c91",
"created_at": "2026-08-30 22:26:36.548492+00:00",
"db_md5": "6dd9baf1b57fd37b2c3b5956af296902",
"submitter": "LauraGomezjurado",
"github_login": "LauraGomezjurado",
"run_id": "6a475079-ca6c-435d-8674-c0a7d1858899",
"tier": "easy",
"dataset_id": "e7",
"st... | {
"score": {
"mean_loss": 9.061687632563315,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04077540117908608
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.21795339351747,
"example_count": 85,
... |
4a75b9da-a3b8-4b31-98c7-26a8aea4dec3 | easy | Dandandan | 2026-08-07 02:59:40.403401+00:00 | failed | 6fb574fc71116a60fc0481542bc7607c6a7228ca089bfac7596dafdeba6c037c | 30,555 | null | """End-to-end recursive function learner for One Layer Deeper.
A population of randomly initialized polynomial register networks is reused
exactly once per requested outer step. The evaluator trains every coefficient
and the evidence selector through its normal forward/backward/optimizer loop.
There is no recurrence-... | {
"id": "4a75b9da-a3b8-4b31-98c7-26a8aea4dec3",
"created_at": "2026-08-07 02:59:40.403401+00:00",
"db_md5": "3eda3e258135eef299dc05e225955d03",
"submitter": "Daniël Heres",
"github_login": "Dandandan",
"run_id": "1bae99f6-9664-412a-9313-b2654083fc23",
"tier": "easy",
"dataset_id": "e1",
"status": "fai... | null
|
4a7ee94e-5add-4447-9442-a1271390a6f1 | easy | DDanlov | 2026-08-17 12:22:20.197111+00:00 | succeeded | 31f63be4f7d218b06cf23bfc287fd1afc19f89de34df87b97f13e0f76418ff0f | 32,944 | null | """
Scalable Recurrent Deliberation Transformer (ScalableDEPHNet)
============================================================
Architecture:
- Pure No-Norm: Single unit RMS normalization strictly isolated to the input token embedding layer.
Zero LayerNorm / RMSNorm inside recurrent deliberation blocks.
- Pure No-Skip... | {
"id": "4a7ee94e-5add-4447-9442-a1271390a6f1",
"created_at": "2026-08-17 12:22:20.197111+00:00",
"db_md5": "1c3f1b4541343eefff2b9e7b1ca29b29",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "e927cf9d-f87a-4558-ae25-1f37587c5444",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.0641026577152815,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.051666666815678275
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0227038860321045,
"example_count": 10... |
4a7fee67-f96c-4bb8-94e8-352e9c9b6750 | easy | shirvani-jr | 2026-08-30 23:31:03.121613+00:00 | succeeded | 5e04560fff6d76c6350ed134dce34bbeb5ddf03992d210d9cdc6dda6ff09212c | 17,544 | null | from __future__ import annotations
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
PERIOD_MINIMUM = 11
PERIOD_MAXIMUM = 100
E... | {
"id": "4a7fee67-f96c-4bb8-94e8-352e9c9b6750",
"created_at": "2026-08-30 23:31:03.121613+00:00",
"db_md5": "cfec548a5587d3a2942b9c10220452bc",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "bdf71ad0-352e-4ff7-bc56-d677e0bc4005",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded"... | {
"score": {
"mean_loss": 8.93152904510498,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013958333991467953
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 10.738463401794434,
"example_count": 300,... |
4a80d3d6-7c77-4bd9-aa11-a084b22c1cb7 | easy | isaac0804 | 2026-08-06 15:03:22.688910+00:00 | succeeded | 1d0ac2d625a850be3fc57b08dbd9e2c9525947398f059ca987f7183b0f1c27b9 | 9,322 | null | """E3 candidate #3: N-curriculum -- warm-start on one confirmed-learnable
modulus, then progressively widen to more.
Three things this session established, in order:
1. submissions/e3_single_n_isolation: a single E3 modulus, trained alone with
its real ~290 natural examples, reaches ~7-9% held-out accuracy (N=713
... | {
"id": "4a80d3d6-7c77-4bd9-aa11-a084b22c1cb7",
"created_at": "2026-08-06 15:03:22.688910+00:00",
"db_md5": "f287668217b5a77022d571a899e0b48f",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "239ccdaf-bb1a-46b2-a2f0-18a2bff4cf82",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 11.95942581344012,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000009313225
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 12.196451688244105,
"example_count": 800... |
4a846316-2474-4b5e-b9ec-26d9d217cacb | easy | yashkant | 2026-08-14 16:34:49.821927+00:00 | succeeded | ed518136bca52d911a494de3facd677c02f6bf33c754143bcea75961ecde77bb | 35,927 | null | """Round1803 ReLIT residual-state refinement inside W12 Dykstra.
The completed Round1765 W12/1T carrier remains intact. Every active public-T
cycle first executes its learned Dykstra projection, then updates separate
answer y and latent z states with one shared transition. The immutable input
anchor is reinjected in... | {
"id": "4a846316-2474-4b5e-b9ec-26d9d217cacb",
"created_at": "2026-08-14 16:34:49.821927+00:00",
"db_md5": "578fc5f93ead6dc959f9e1f68784e804",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "50ad9598-fe57-4c82-992c-f80700c9b640",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 4.0795928835868835,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.3329166807234287
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.472601413726807,
"example_count": 300,
... |
4a8566f5-0f6d-4465-9a1d-be0a3db9a7a8 | easy | oupadhyay | 2026-08-15 21:52:38.823573+00:00 | succeeded | ac990a8c6c23bd6f297e4dc30a65d4e18a68ae08f66a480ca8f70184600df301 | 3,991 | null | """Conditional optimizer screen: lr15."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import ModelSpec,OptimizerSpec,OptimizerBundle,Submission,assert_model_state
W,D,H=4,64,32
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_size,s... | {
"id": "4a8566f5-0f6d-4465-9a1d-be0a3db9a7a8",
"created_at": "2026-08-15 21:52:38.823573+00:00",
"db_md5": "967aba6008b829b65f37ee4449c22fea",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "2974f0e4-f2ac-449a-af44-892a578d2a0d",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.4053405273255883,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.263085500007253,
"example_count": 600... |
4a85d426-2794-4ab4-a325-9dea0a413a48 | easy | shirvani-jr | 2026-08-29 00:39:05.091921+00:00 | succeeded | 3685b381d99816499015803de8f43607e22525db5bf77bc5d2df649ca184d39f | 49,713 | null | """Adaptive learned discrete recurrent fabric for One Layer Deeper.
A gradient-trained, operation-free recurrent computational machine. The
model learns, from the competition's endpoint supervision: how to parse the
prompt into a bank of categorical digit registers, a reusable transition
circuit applied serially, a l... | {
"id": "4a85d426-2794-4ab4-a325-9dea0a413a48",
"created_at": "2026-08-29 00:39:05.091921+00:00",
"db_md5": "88345330be208273f90bac5157686628",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "ab900560-4614-4319-8ced-9ba28fb25ff5",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.447811484336853,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.34833334293216467
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5945675373077393,
"example_count": 100,... |
4a8f8383-15e1-4284-bc97-2b6bb58c3395 | easy | chad-atexpedient | 2026-08-06 02:57:19.581023+00:00 | succeeded | 3a371e423653134d2a4d0a77cadb3052181929e96143026079e806f57328ba76 | 10,383 | null | """B02: endpoint-aligned serial transducer with straight-through commits."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
WIDTH = 320
HEADS = 8
FFN_WIDTH =... | {
"id": "4a8f8383-15e1-4284-bc97-2b6bb58c3395",
"created_at": "2026-08-06 02:57:19.581023+00:00",
"db_md5": "efd6556e6820f17805b708132115f247",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "f56177da-9851-4dce-ad1b-01cc2cca2c55",
"tier": "easy",
"dataset_id": "e3",
"st... | {
"score": {
"mean_loss": 2.225455794963757,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0006250000046566129
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.220534183620955,
"example_count": 800... |
4a9390fc-d15e-4f79-a5f3-7f5092b92121 | easy | viridale | 2026-08-24 20:35:52.263670+00:00 | succeeded | 91cd5950f43540f7f622552857426de0cd8e06f6ac3b94c1a7db4cce8fa0fed0 | 16,513 | null | """Scale-covariant vertex bilinear Transformer program.
hypothesis: v855's shared scalar readout collapsed all interaction queries onto the same integer vertex; independent random-initialized readouts and sub-vertex optimizer steps should let the unchanged endpoint objective select interaction coordinates separately.
... | {
"id": "4a9390fc-d15e-4f79-a5f3-7f5092b92121",
"created_at": "2026-08-24 20:35:52.263670+00:00",
"db_md5": "59cd9e74200e7468d696c72a65a6d160",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "63d3f128-c087-499c-8e46-988c4b6d20d7",
"tier": "easy",
"dataset_id": "e2",
"status": "succe... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 300,
"exact_accuracy": 1.0,
"... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.