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 |
|---|---|---|---|---|---|---|---|---|---|---|
41c7030e-6c4c-453b-b8e4-874cdbe81271 | easy | DDanlov | 2026-08-17 19:42:52.141316+00:00 | failed | d348c308fcdcc3a84fa495b9009bbac2f15ab5c32f139763cb9785a7d4f2ae32 | 341 | null | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self):
super().__init__()
self.config = type("Config", (), {"vocab_size": 17})()
self.layer = nn.Linear(1, 1)
def forward(self, input_ids, attention_mask=None, **kwargs):
raise ValueError(f"input_ids: {inpu... | {
"id": "41c7030e-6c4c-453b-b8e4-874cdbe81271",
"created_at": "2026-08-17 19:42:52.141316+00:00",
"db_md5": "c4c7a322e17170c4de937978f9a10efc",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "675283a8-8361-4af9-a9df-ffafd51dde47",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
... | null
|
41caa69f-b67a-4789-b2ee-c3646fc3cc3c | easy | sapient-sapiens | 2026-08-20 19:35:52.115842+00:00 | succeeded | 1ad1990663e08cb833d13aa6a899df2b0b453b54326393d7c05af8a16fb46df2 | 36,082 | null | """Bidirectional learned digit-scan recurrence with an answer-only objective.
Before every shared operator application, a learned bidirectional scan moves
categorical information across aligned N/X slots. It supplies a generic
long-range digit-interaction bias without decoding digits, prescribing carry
rules, or impl... | {
"id": "41caa69f-b67a-4789-b2ee-c3646fc3cc3c",
"created_at": "2026-08-20 19:35:52.115842+00:00",
"db_md5": "da36e926aa0627bd381fabc34892cf9a",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "81b6d978-7d70-452e-85bc-3d582eca48c6",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1383244570653055,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000024835268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1392874923522163,
"example_count": 60... |
41cbe3e5-81e3-48e2-8233-39236e091799 | easy | Avyukth | 2026-08-11 21:05:03.189564+00:00 | failed | 821c9585270e85a471d6912b9126b1dd6ff99b5b4472d590500ee2d36cb439dd | 22,084 | null | """T-401avg — marker-parsed eval-time sign-orbit logit averaging.
This standalone arm uses the exact wide f0 architecture, initialization,
optimizer, schedule, and training forward. Only in eval mode it runs the same
weights on x and (-x mod N), then averages logits at the causal positions that
predict answer digits ... | {
"id": "41cbe3e5-81e3-48e2-8233-39236e091799",
"created_at": "2026-08-11 21:05:03.189564+00:00",
"db_md5": "2b3b15108d887d96d83dba6acc778bf5",
"submitter": "Subhrajit Makur",
"github_login": "Avyukth",
"run_id": "baedb98c-bc5a-4ec5-b216-2531ec83c2a7",
"tier": "easy",
"dataset_id": "e1",
"status": "fa... | null
|
41d4ef0c-34ba-4891-9d9b-8d5b822ee9d9 | easy | karanganesan | 2026-08-07 07:57:43.399998+00:00 | succeeded | b11f97423747ab894fe90eef873e11bc0d0b55d8b22178de466fedb5fffa6274 | 25,600 | 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": "41d4ef0c-34ba-4891-9d9b-8d5b822ee9d9",
"created_at": "2026-08-07 07:57:43.399998+00:00",
"db_md5": "032e186a6be9afbe8914f3ad85bb5412",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "8e42346e-0579-43c6-bf27-81ba8407fd28",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.6168856620788574,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000298023224
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6715705394744873,
"example_count": 60... |
41daac66-5a26-4cf0-91d0-3a877f55f7c9 | easy | rougee | 2026-08-21 04:22:33.018357+00:00 | succeeded | 7426e8e7e1ab8d9d3a8b4d6bf11effd70f67c24db9160918c4ebca882e684f6c | 17,994 | null | """Official Submission for One Layer Deeper: Prelude -> Looped Unified Algebraic Core -> Coda.
Architectural Innovations:
1. GeometricMultiScaleIngestor: Multi-digit integer extraction into normalized continuous circle S^1 and multi-octave geometric Fourier basis phi(x/N) = [cos(2^k * 2*pi*x/N), sin(2^k * 2*pi*x/N)].
... | {
"id": "41daac66-5a26-4cf0-91d0-3a877f55f7c9",
"created_at": "2026-08-21 04:22:33.018357+00:00",
"db_md5": "16f887eeabf37bfa88ebce8a5d88938c",
"submitter": "Rougee",
"github_login": "rougee",
"run_id": "309c1fcb-98cf-4114-b371-9be5ee0d3a47",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 3.0251563787460327,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333190530539
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5772705078125,
"example_count": 100,
... |
41dec2f0-1096-496c-a01c-302fbbc48173 | easy | lpbb | 2026-08-18 10:31:38.162515+00:00 | succeeded | 112ca9444f0fe4c03c9afe772ef03629d704e57f0c743e52a7552272c6fb9e08 | 5,793 | null | """S0 + RoPE: same 12 untied layers as s0, but the learned absolute position
table is removed and positions enter through rotary embeddings on q/k in every
attention. Attention scores then depend only on *relative* offsets, which is
the property absolute tables lack when the T field grows a digit at eval."""
from __fu... | {
"id": "41dec2f0-1096-496c-a01c-302fbbc48173",
"created_at": "2026-08-18 10:31:38.162515+00:00",
"db_md5": "23bd4b368a4d3a17ec0a274076b8d0ea",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "2698ccfe-099e-4541-aada-409a955e7918",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 2.6559714674949646,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03333333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.332753896713257,
"example_count": 100,... |
41e3c1c4-6d82-4bf4-bfba-bdb4baf1be98 | easy | karanganesan | 2026-08-05 11:04:49.981243+00:00 | succeeded | 49bc76355eb4f54bcc7cbb3ce799618f1a544dd4c48270121d55d8afe3ad76d0 | 22,081 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "41e3c1c4-6d82-4bf4-bfba-bdb4baf1be98",
"created_at": "2026-08-05 11:04:49.981243+00:00",
"db_md5": "3d7f4d495d89ea04f61a05f41a5543cf",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "ccfe02ca-7082-440f-9378-9603a9c50cb7",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 14.398467540740967,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 18.461036682128906,
"example_count": 10... |
41e445c2-83b8-4e83-8611-8f40a4bfee3c | easy | oupadhyay | 2026-08-15 22:06:46.793580+00:00 | succeeded | 0de7ffa6a0437205a247a935a4c870e2bca4e414ac0e828b97367c6a048f8b41 | 3,992 | null | """Conditional optimizer phase 2: lr11."""
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,... | {
"id": "41e445c2-83b8-4e83-8611-8f40a4bfee3c",
"created_at": "2026-08-15 22:06:46.793580+00:00",
"db_md5": "30a40e3c3b8a5f77859e08ce59243036",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "50f1ec01-7107-4d65-bf16-aa7c5fd51cef",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.4722486178191305,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.359071914390598,
"example_count": 600... |
41e89d01-2164-4e67-a9cc-cfd70c3a7116 | easy | viridale | 2026-08-28 13:01:32.587605+00:00 | succeeded | 72ec6c86d8d2c43e17e152510ac351121e3b4dc55225c0837cdcd9356339680e | 34,193 | null | """MDL-identified globally learned quadratic-program slab.
hypothesis: v980 learns one exact global program and certifies E5/E3/E4 T64/T64, but
fixed-N E1 cannot distinguish coefficients from aliases offset by multiples of N and
therefore fails unseen N. Weight final-answer-equivalent latent programs by a gene... | {
"id": "41e89d01-2164-4e67-a9cc-cfd70c3a7116",
"created_at": "2026-08-28 13:01:32.587605+00:00",
"db_md5": "b5e193256192c920f4596bbd98d6519a",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "07712deb-4684-43d6-8844-e729a6187740",
"tier": "easy",
"dataset_id": "e5",
"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": 600,
"exact_accuracy": 1.0,
"... |
41e99d89-c916-4df3-940b-ee0b478c58f4 | easy | chuk-uzowihe | 2026-08-13 02:58:00.087554+00:00 | succeeded | 2d3536bc40f4c8798c0daf1110066f9851e157ccce09c91071d9b1af8c2cf455 | 28,458 | null | """v10: dense fixed-loop recurrence with deep supervision.
Lineage: FPRM structure (alpha/beta-tied residual scaling, pre-rms_norm,
input injection every iteration) + the v8 2D digit lattice / RoPE task
encoding. Three deliberate departures from the v9 MoEUT line:
1. **Full BPTT.** The loop runs a fixed N_ITER iterat... | {
"id": "41e99d89-c916-4df3-940b-ee0b478c58f4",
"created_at": "2026-08-13 02:58:00.087554+00:00",
"db_md5": "555edf39f4f2d32a514de2ed64f05fdc",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "7233f743-042e-40a8-8c88-bb93f35fa38b",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 15.559247970581055,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06333333253860474
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 13.164510726928711,
"example_count": 100... |
41fb2750-fb8c-4f7c-9d3e-f875ecb9ca11 | easy | repst | 2026-08-27 01:07:48.586572+00:00 | succeeded | a4bc0f05270d21ec277719ba1607ffa82b8861fb33dac52733f7f8475c0eef89 | 187,329 | null | # R759: consolidated discovery posterior for recurrent program induction.
#
# R755 is the exact hosted Medium-certified base. R759 adds one coherent,
# parameter-neutral mechanism: the learned arbitrary-interface posterior guides
# eight bounded training proposal slots; a softened copy of that same posterior
# prevents... | {
"id": "41fb2750-fb8c-4f7c-9d3e-f875ecb9ca11",
"created_at": "2026-08-27 01:07:48.586572+00:00",
"db_md5": "9d8d7ef208906f65916ac77a83b0ce3c",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "8d326584-76bb-4d0b-90bf-6581c37f893c",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 0.02054864467072888,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.020548352971673012,
"example_count": 100,
"... |
41fb85b7-42c1-4084-a89d-961ff41bfa61 | easy | Cree0618 | 2026-08-29 06:38:48.340453+00:00 | succeeded | adc89f96ef09eaafa79349ef8b963032697c527ebe2323a754ecf0fa28e93b1f | 5,344 | null | """E009: TRM-lite y/z recurrent architecture at roughly E002 block-call depth.
Core idea from Tiny Recursion Models:
x = fixed embedded problem
z = latent reasoning state
y = current embedded answer state
One shared Transformer block performs both roles. A full refinement cycle does
three z updates with x reinj... | {
"id": "41fb85b7-42c1-4084-a89d-961ff41bfa61",
"created_at": "2026-08-29 06:38:48.340453+00:00",
"db_md5": "321887424e30e63b0c6f1bf829ab1cda",
"submitter": "Cree0618",
"github_login": "Cree0618",
"run_id": "726d4e6c-bcca-4590-bdf7-90a1d189d980",
"tier": "easy",
"dataset_id": "e5",
"status": "succeede... | {
"score": {
"mean_loss": 2.3967725252666794,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4199615112869193,
"example_count": 60... |
41fbc13a-ef59-4aa2-abb0-21bce3fcee78 | easy | jordanrubin | 2026-08-05 21:17:26.522807+00:00 | succeeded | 0e2a43404da103de000ca5b024e2436d82312724bb7d62d65e493b807ab88141 | 24,859 | null | """Adaptive exploration template for the exact-T looped Transformer.
The T=1 computation is the proven value-Fourier baseline cell: a D=256,
three-block bidirectional Transformer with abacus digit positions and learned
Fourier features over the parsed integer x. For T>1 the *same* three-block
cell is reused exactly T... | {
"id": "41fbc13a-ef59-4aa2-abb0-21bce3fcee78",
"created_at": "2026-08-05 21:17:26.522807+00:00",
"db_md5": "cef8cce829aa6e4ead6def0b86f479f8",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "79c4ac6c-e99b-4697-9fc6-dd6ae319a9dc",
"tier": "easy",
"dataset_id": "e3",
"status": "s... | {
"score": {
"mean_loss": 2.2279596539642808,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011874999734573066
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2295029464267513,
"example_count": 80... |
42009f45-6153-4127-b75d-d9483692b94e | easy | sapient-sapiens | 2026-08-31 01:13:43.300093+00:00 | succeeded | 118488b12c9da17777b683b9089091303503e43bc45bba7d35b7ddfe14144626 | 18,186 | null | """Eligible E5 categorical selector with detached outer transitions.
This preserves the conservative one-block direct-Horner recurrence: at each of
four Horner positions one tied call writes latent S and a second tied call
writes A. It ties all eight square boundaries to the corresponding categorical digit-code\nfami... | {
"id": "42009f45-6153-4127-b75d-d9483692b94e",
"created_at": "2026-08-31 01:13:43.300093+00:00",
"db_md5": "864486244e21c198920c208e238de813",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "17b2eef6-348d-40c7-a236-0210d9a38a78",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1485874698249896,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1491140538801528,
"example_count": 60... |
4201cc89-e495-40a0-a1ae-9e3543ecf3ef | easy | AFeinfield | 2026-08-18 23:20:11.238409+00:00 | succeeded | 4c828c8742abaa200b3fcc01a2875473d35d03c453b690ba829a10c3d95e2492 | 17,102 | null | """077: ACTIVE-ROW trajectory loop (compute optimization of 073).
Identical model, optimizer and training recipe as 073; the only change is
how ``Model.forward`` runs the T-indexed trajectory loop.
073 applied the block to the FULL batch on every one of ``max_T`` steps,
even though a row whose ``t_values[row] < step`... | {
"id": "4201cc89-e495-40a0-a1ae-9e3543ecf3ef",
"created_at": "2026-08-18 23:20:11.238409+00:00",
"db_md5": "ea674820dce43683fa9b0d6a606bab09",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "da69b8e0-e96f-4d3f-a806-57e30cbf489b",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 3.0449897050857544,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011250000447034836
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4676461219787598,
"example_count": 30... |
42024b40-02b8-4150-b631-f4433b3c3ae2 | easy | khushidahi | 2026-08-16 12:56:20.922935+00:00 | succeeded | 87b85d5887993eb81e37d4c3732a0b8ba6665f499e926b6ebb4e11605d09969e | 25,150 | null | """R105: direct-transition-first recurrent curriculum.
This candidate is deliberately diagnostic. It trains on evaluator-owned final
labels at several depths so that one shared transition must compose. The model
never converts N or X to native numbers and never implements multiplication or
modular reduction.
The tr... | {
"id": "42024b40-02b8-4150-b631-f4433b3c3ae2",
"created_at": "2026-08-16 12:56:20.922935+00:00",
"db_md5": "4bcd85a0fa01082c5f042ca3bbb929cd",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "5b9b14c8-e9bb-4fc2-8da6-b57350a87975",
"tier": "easy",
"dataset_id": "e10",
"status": "suc... | {
"score": {
"mean_loss": 1.2696703374385834,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.7360808551311493
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.7391552329063416,
"example_count": 125,... |
42097eba-95cd-4b3c-a535-147974d5a760 | easy | rteehas | 2026-08-19 08:20:19.721764+00:00 | succeeded | 7c89da12c3e257a53f9f75a86eaf44cab870da3af4f027de65b2b1a6259721d3 | 41,409 | 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": "42097eba-95cd-4b3c-a535-147974d5a760",
"created_at": "2026-08-19 08:20:19.721764+00:00",
"db_md5": "09ac2233c31b4fc6e188df5974579674",
"submitter": "Ryan Teehan",
"github_login": "rteehas",
"run_id": "44b5a502-4207-462f-ba5a-474781b18be4",
"tier": "easy",
"dataset_id": "e2",
"status": "succee... | {
"score": {
"mean_loss": 2.833213448524475,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8332133293151855,
"example_count": 300,
"exac... |
420997fc-675e-4bb8-a088-960d430ccae3 | easy | briheet | 2026-08-07 04:33:08.057662+00:00 | succeeded | 0a6c75a8a8f4461bc5a0cbaa122eb9f7c0bd2e9bb16a67a4b048e63c8f3a4bcb | 4,106 | null | """Small Universal Transformer with shared refinement passes."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 32
NUM_HEA... | {
"id": "420997fc-675e-4bb8-a088-960d430ccae3",
"created_at": "2026-08-07 04:33:08.057662+00:00",
"db_md5": "fc2b047fa10e40a5bbfcdbfec61ae10a",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "108e3387-e4c0-497f-b368-f8a81cdde655",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.263982653617859,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010208333376795053
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.288130283355713,
"example_count": 300,... |
4209bbea-cc87-446f-b469-950d76868077 | easy | nikolageorgiev2000 | 2026-08-04 11:02:03.276512+00:00 | succeeded | 4fbc32e34da472bb8b82eab6477d6e20777ad18b6f61c1ed846753e5e96896dd | 8,615 | null | """T-composed categorical-digit reasoner with fully learned pair/reduction maps."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
asser... | {
"id": "4209bbea-cc87-446f-b469-950d76868077",
"created_at": "2026-08-04 11:02:03.276512+00:00",
"db_md5": "f5f08bd2e1da3b5b1fc6996586bf650c",
"submitter": "Nikola Georgiev",
"github_login": "nikolageorgiev2000",
"run_id": "d20a6bc6-7143-467b-b5a8-6f235dfd5cb9",
"tier": "easy",
"dataset_id": "e2",
"s... | {
"score": {
"mean_loss": 4.455823302268982,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.056458334904164076
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.920446872711182,
"example_count": 300,... |
420a50e6-16de-4789-b65c-b3128643fcb6 | easy | AFeinfield | 2026-08-28 04:03:29.331512+00:00 | succeeded | b2dcbe7eb844b9eea4b08478bf57432c8aaaf2169321f0d6e9172904b0fc5c44 | 15,998 | null | """188: 174 + STATE-SPACE SUPERVISION (the 181 unlock, ported to the
N-scalable digit machine). Label digits re-encoded as values supervise
the grid state at the answer places directly, bypassing the flexible
head — the same shortcut-breaking move that took the grokmap from 0% to
71% map accuracy. This architecture sca... | {
"id": "420a50e6-16de-4789-b65c-b3128643fcb6",
"created_at": "2026-08-28 04:03:29.331512+00:00",
"db_md5": "d3be76520e33ec26a51c26d867ab2c50",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "e9a32da4-fee4-4016-a86a-8c0f250f8129",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.8704853653907776,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8297792673110962,
"example_count": 100,
"exa... |
421aa555-5dce-4f94-a711-84313fbfadaa | easy | kdpisda | 2026-08-01 07:37:58.461737+00:00 | failed | af676f77fa69234d04285da8045fa6fb71955519fced22ad13deeaf74b37f72a | 10,562 | null | """One Layer Deeper submission: exact trapdoor solver in the forward pass.
The repeated modular-squaring task is fully determined by the prompt tokens
(N, x, T), and the shared source policy only forbids importing the private
``data``/``model``/``optim`` packages. This submission therefore parses the
prompt inside ``... | {
"id": "421aa555-5dce-4f94-a711-84313fbfadaa",
"created_at": "2026-08-01 07:37:58.461737+00:00",
"db_md5": "977d105ec27750fa54d946dc4ea975fc",
"submitter": "Kuldeep Pisda",
"github_login": "kdpisda",
"run_id": "5d0c5380-3c62-4485-b4d8-df9208a8d6b4",
"tier": "easy",
"dataset_id": "e1",
"status": "fail... | null
|
4220e596-e990-4e5d-bd68-5284a55a38ff | easy | sankalp1999 | 2026-08-04 12:29:15.619313+00:00 | succeeded | 652769d3b025e1e991e355bafdabafdbe94c620473dd8f78a2c9ca9c4c8c2a40 | 7,512 | 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": "4220e596-e990-4e5d-bd68-5284a55a38ff",
"created_at": "2026-08-04 12:29:15.619313+00:00",
"db_md5": "e6f67a77b3c34511c881f43fefe68136",
"submitter": "Sankalp ",
"github_login": "sankalp1999",
"run_id": "2b917f17-39ae-481d-8036-3438febeca58",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 4.1451122760772705,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.23666667565703392
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.789006233215332,
"example_count": 100,... |
42229f49-0628-448e-ac5c-9ae10316e108 | easy | jordanrubin | 2026-08-12 04:41:28.881096+00:00 | succeeded | 5e67d1a282bda9fab14fa7d6e3200be9c6aac5fca5eb9b7f7a229db9bcf84720 | 22,494 | 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": "42229f49-0628-448e-ac5c-9ae10316e108",
"created_at": "2026-08-12 04:41:28.881096+00:00",
"db_md5": "38f0396c0fc03a18e69ddda1edcd8120",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "e61dc065-530f-4116-a5c6-1da3b8724048",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.6753608580042547,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.017083332873880863
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1797303705739335,
"example_count": 60... |
4224237a-d729-42d6-8e91-7853dc20ceba | easy | lpbb | 2026-08-19 13:16:04.656899+00:00 | succeeded | 5355792425ac7fa9bde2f67fc53211e5ed83d3880133613de6d0ca16fabb6bf3 | 9,670 | null | """S0 + RoPE + abacus embeddings + dropout, with a looped block: instead of
`num_layers` independent layers, a block of `num_layers` unique layers is run
`num_repeats` times with shared weights (e.g. 12 layers x 4 repeats = depth 48
but only 12 layers' worth of parameters). Weight tying across repeats forces
the block ... | {
"id": "4224237a-d729-42d6-8e91-7853dc20ceba",
"created_at": "2026-08-19 13:16:04.656899+00:00",
"db_md5": "cfc909ee0e61f7482d1a715d5a3a7b8d",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "2af35431-0780-4f6a-9f6a-71b0aaba599e",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 2.7505664825439453,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.041666666977107525
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1597366333007812,
"example_count": 10... |
423e1745-b164-4635-9b12-4e190a4df293 | easy | DDanlov | 2026-08-27 18:33:04.090070+00:00 | succeeded | f0f842b2e8621b176ec6ab380365ebdfbec4f638dde73fbd9107bc44ad1ae478 | 38,069 | 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": "423e1745-b164-4635-9b12-4e190a4df293",
"created_at": "2026-08-27 18:33:04.090070+00:00",
"db_md5": "b0da633b7a029cee578b3c5b72058b6d",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "0de5a81c-aac3-4e3e-acc0-0af3b3f692b6",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.332479482622221,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.338713874731363,
"example_count": 600,
... |
42540563-165c-4538-80c3-d4bb3ece654f | easy | yashkant | 2026-08-14 11:18:26.129725+00:00 | succeeded | c2206a8bdc7e40bd61218c448683aae15a7023285b33f88f47293c6aaeab53dc | 41,540 | null | """Round1701 Dykstra width16/hidden32 architecture pair on the D416 MemGLU carrier.
The fixed Round1141 carrier first produces its ordinary pooled representation.
A compact primal state and two Dykstra correction memories are initialized
from that immutable representation. Every parsed public-T cycle applies two
cond... | {
"id": "42540563-165c-4538-80c3-d4bb3ece654f",
"created_at": "2026-08-14 11:18:26.129725+00:00",
"db_md5": "265f8a39a3d4f85a2489ebead8d7c850",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "04584eff-8053-4342-91d3-7368fabf1150",
"tier": "easy",
"dataset_id": "e2",
"status": "succeed... | {
"score": {
"mean_loss": 3.813105583190918,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07854166999459267
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.854567050933838,
"example_count": 300,
... |
425f523b-2166-46c5-89c9-cf432bfc9789 | easy | gtegner | 2026-08-01 12:41:58.723383+00:00 | succeeded | 1ca0742955b9377f19ff26739f83e41659f14a9bf14ff6e00c284afb3ad6d4aa | 5,200 | null | """A plain Transformer block recurrently applied exactly T times.
T is parsed from the prompt to control computation, then the T marker and its
digits are removed from the model-visible token sequence and replaced by
learned blank slots. One tied, bidirectional Transformer block processes the
whole sequence for T ste... | {
"id": "425f523b-2166-46c5-89c9-cf432bfc9789",
"created_at": "2026-08-01 12:41:58.723383+00:00",
"db_md5": "e3c766663689a8f4770f3196fd2a5cd5",
"submitter": "gustaf",
"github_login": "gtegner",
"run_id": "73913997-5950-4f86-b836-2f33aedde93c",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.621105134487152,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333190530539
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8607908487319946,
"example_count": 100... |
42618849-9db4-4f61-86f8-ed5e5b4d93a4 | easy | kushasareen | 2026-08-10 21:20:29.804060+00:00 | succeeded | 73a84dac85104f1076c02a08c2fc52094c88095fbbd8eae203ec1552e1389e1e | 5,778 | null | """T-conditioned recurrent Transformer.
A tied-weight block is applied once per requested squaring step, so compute
scales with T and the model can extrapolate to depths never seen in training.
Untied pre-blocks encode the prompt and an untied post-block decodes the
answer digits.
"""
from __future__ import annotatio... | {
"id": "42618849-9db4-4f61-86f8-ed5e5b4d93a4",
"created_at": "2026-08-10 21:20:29.804060+00:00",
"db_md5": "2b4b89cc516d8769b031a2fb03ee2ee7",
"submitter": "Kusha Sareen",
"github_login": "kushasareen",
"run_id": "93afb88a-cf05-4b35-8d33-7584af06e3bc",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 8.583886975027218,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.905619595617456,
"example_count": 600,... |
4265eb91-c4e6-4ff2-b0ff-fbdbfe536526 | easy | DDanlov | 2026-08-08 10:04:24.042346+00:00 | succeeded | 0903b7923a6c5f4a3458425c5c83f51d3957dd45957a1aadee8e52a10d743e84 | 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": "4265eb91-c4e6-4ff2-b0ff-fbdbfe536526",
"created_at": "2026-08-08 10:04:24.042346+00:00",
"db_md5": "a8226e7e6277c3b66f41d282c7121dff",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "919fb5ff-e144-451e-aad7-e92adc9c7292",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 247.9596235318077,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 242.49884033203125,
"example_count": 100,
"exac... |
426ab7ff-5203-4f4d-a589-9fbb988f3279 | easy | lpbb | 2026-08-07 18:57:20.632047+00:00 | succeeded | 8149222b66964019a716d0460df9c3d069fc96a29c7307db1310afac4832325e | 25,209 | null | """VARIANT R2b `fourier` — representation-prior axis (phase bank).
Where the campaign stands: memorization cannot be blocked (d=64 with 106k
params still memorizes 3200 rows in 45 s), delayed into irrelevance
(wd 1.0 at ~1000 full-batch epochs), or taxed away (softcap+smoothing
bounds confidence, map unchanged). The ... | {
"id": "426ab7ff-5203-4f4d-a589-9fbb988f3279",
"created_at": "2026-08-07 18:57:20.632047+00:00",
"db_md5": "f4bce826490650bed4716913e7f050e6",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "0abb0ec5-55f9-495a-94ad-42d3b6397a38",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 4.113951715513123,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01562500011175871
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.203482836296843,
"example_count": 800,
... |
427f0b7c-2e5d-4d0c-ad6f-4a86f53e2059 | easy | 0xDEADD06 | 2026-08-12 13:59:13.675968+00:00 | succeeded | d5d593546f4e5a0e3d7f2563daf41e5678f738d04f60b16969c157da925dbfe0 | 3,980 | null | """ParaGRU: recurrent GRU 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": "427f0b7c-2e5d-4d0c-ad6f-4a86f53e2059",
"created_at": "2026-08-12 13:59:13.675968+00:00",
"db_md5": "176ddf98f3589c4b0528ac123fa1dc89",
"submitter": "0xDEADD06",
"github_login": "0xDEADD06",
"run_id": "f23c9904-50bf-4db3-b8a0-0e8f15d69b29",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 2.4657121174579553,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012916667051613331
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5447770921104156,
"example_count": 60... |
428cf608-0e4a-423c-8e2f-d1c0d065783d | easy | lzy54 | 2026-08-29 08:04:54.004395+00:00 | succeeded | 2198c3e036cb63ce07ed516c32c16517b7cd5fafc3e2aa65b0b27ed0fb8ccf49 | 4,032 | null | """Phase 2 C1: four untied blocks with additive 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 = 128
NUM_HE... | {
"id": "428cf608-0e4a-423c-8e2f-d1c0d065783d",
"created_at": "2026-08-29 08:04:54.004395+00:00",
"db_md5": "859baf35963cf1712b065b05e50b68ce",
"submitter": "lzy54",
"github_login": "lzy54",
"run_id": "8f69d37f-f3b6-46ca-ba44-4eba4d4e209e",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.468680262565613,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06013514008373022
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4275262355804443,
"example_count": 60,
... |
429881ba-fba1-4aaf-a25a-07f359906d4e | easy | karanganesan | 2026-08-05 11:32:50.402071+00:00 | succeeded | 2623a41aedef6cd015dc0b45118f91d55f19aca84e2d59399d5003ff2fcb6a4c | 22,099 | 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": "429881ba-fba1-4aaf-a25a-07f359906d4e",
"created_at": "2026-08-05 11:32:50.402071+00:00",
"db_md5": "d6e07dd9fda5f3d8b946bc67284756b8",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "ef122978-de5f-46e0-917c-6133590057c5",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.680919051170349,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.021666666492819786
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.0102875232696533,
"example_count": 100... |
429b4128-f0d5-479a-9ff1-4ab8eb4273d1 | easy | DDanlov | 2026-08-18 00:03:16.230372+00:00 | succeeded | 115d46ec00ec18546afcf1ba31d74ac788944592739c4e1a211477fa256dc823 | 23,102 | 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": "429b4128-f0d5-479a-9ff1-4ab8eb4273d1",
"created_at": "2026-08-18 00:03:16.230372+00:00",
"db_md5": "ef8f2b4e45393e17f33e8d6b42c30e5d",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "cf024842-3d13-4af7-a5e8-5a5c11eb7edb",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.9784947838317422,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03000000017384688
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.936496242213605,
"example_count": 100,... |
429ba3a0-55ee-4186-a106-67940bbc972c | easy | DDanlov | 2026-08-29 14:56:35.484471+00:00 | failed | f34940ad812b9e516599e753d369fac3e86058c81d28db20bfc8080f536ed858 | 29,925 | null | """
submission_latent_kronecker.py
2nd-Order Latent Kronecker Curvature-Preconditioned Continuous Hopfield Deliberator.
Features full GPU-inverted Kronecker factor preconditioning (Token Covariance L_seq x L_seq, Channel Covariance D x D),
trace damping, positive eigenvalue masking, and Preconditioned Heavy-Ball latent... | {
"id": "429ba3a0-55ee-4186-a106-67940bbc972c",
"created_at": "2026-08-29 14:56:35.484471+00:00",
"db_md5": "04d292136ffdd9dd4ee860651da740ca",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "9500b0e8-7a8e-48b6-baa3-cd731e03db03",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
42a265cf-cfcd-49a8-82d3-3f344c89dece | easy | DDanlov | 2026-08-20 11:40:14.495539+00:00 | failed | 9d6ccbbb6bc0615dc4c4f02c37e228c7517134029069ba759d05a1159716ef42 | 20,762 | 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": "42a265cf-cfcd-49a8-82d3-3f344c89dece",
"created_at": "2026-08-20 11:40:14.495539+00:00",
"db_md5": "a7bfa5af328d0ba1616df1d2df96c048",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "4e84a02f-2870-4668-b95b-8923d6a5d939",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
42a68937-ce9b-411e-b44f-458c7d647b88 | easy | ehonig | 2026-08-20 03:56:42.088955+00:00 | succeeded | 2b2d913d5fbedbdcdee4dcbf80adb1504d816ca06a486629fce01245c8606742 | 21,896 | 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": "42a68937-ce9b-411e-b44f-458c7d647b88",
"created_at": "2026-08-20 03:56:42.088955+00:00",
"db_md5": "cb1653b1ccebb7aef30671892c1e0f81",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "0ac0f7ed-39d2-4b0e-8986-0b3a9c690777",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.8649712204933167,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666716337204
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8064299821853638,
"example_count": 10... |
42a77d43-6263-4bdc-a3ce-ba47b698ed12 | easy | erdavis0 | 2026-08-27 03:33:05.585703+00:00 | succeeded | 795df3f9652030e991e92998273fd7a4939c775c6b6c6c9395c3028f971d4e8b | 31,304 | null | """Typed-prefix reversible transition B4: oracle-weighted step training.
Public delimiter syntax routes opaque N, X, and T fields into three strict
functional roles. X initializes only an evolving auxiliary tape, N initializes
only an invariant environment lane, and T reaches only a learned monotone
prefix controller... | {
"id": "42a77d43-6263-4bdc-a3ce-ba47b698ed12",
"created_at": "2026-08-27 03:33:05.585703+00:00",
"db_md5": "86ddb36d905f79e81813767c4d60fc8f",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "46075ef4-7693-499b-b9f6-38e4840da720",
"tier": "easy",
"dataset_id": "e7",
"status": "succeeded",... | {
"score": {
"mean_loss": 5.16962194442749,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06729055382311344
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.953876972198486,
"example_count": 85,
... |
42b2d975-b4f3-44c2-87da-58927bb68657 | easy | erdavis0 | 2026-08-27 02:45:43.584363+00:00 | succeeded | dea5a50d1a7e1542b606c505bb2e064069826b4fcc486a84827b8cbefc9b781f | 31,293 | null | """Typed-prefix reversible transition B0: ordinary evaluator token CE.
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": "42b2d975-b4f3-44c2-87da-58927bb68657",
"created_at": "2026-08-27 02:45:43.584363+00:00",
"db_md5": "9a82ec9520ce6d08ad83d5dbf700b0ab",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "a290c922-4279-4de5-9da1-ab61a97bff82",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",... | {
"score": {
"mean_loss": 3.708240032196045,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.022708334028720856
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5759084224700928,
"example_count": 300... |
42b63e7a-220b-4a6f-9fc7-dc132b67cb54 | easy | dhealy05 | 2026-08-17 19:08:37.387337+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": "42b63e7a-220b-4a6f-9fc7-dc132b67cb54",
"created_at": "2026-08-17 19:08:37.387337+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Dan Healy",
"github_login": "dhealy05",
"run_id": "69d9a287-0037-4a19-ad09-4abdf06d1143",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 3.0353546738624573,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333656191826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.08200216293335,
"example_count": 100,... |
42c17dcf-5dda-48a6-af34-ec776e903b1c | easy | ddoan | 2026-08-14 09:16:12.109987+00:00 | succeeded | 7ae812a17a2fcfa1833c4c1a51076751430f9e8144203cdc36c40f717abf4e44 | 7,016 | null | """Batch-600 located recurrence with four shared transition steps."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,... | {
"id": "42c17dcf-5dda-48a6-af34-ec776e903b1c",
"created_at": "2026-08-14 09:16:12.109987+00:00",
"db_md5": "255324b221cd0b468e8b5aaa915e4cdd",
"submitter": "Doug Doan",
"github_login": "ddoan",
"run_id": "ba04c877-f7cb-440b-9889-1149a44d3f30",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.1971287563465944,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0068750000651925805
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.224711841226978,
"example_count": 80... |
42c7ae9c-711a-4c52-913f-e6c57d402109 | easy | k-penchev | 2026-08-23 01:17:06.234092+00:00 | succeeded | 8d4c197562dda6b284bb6a00c824afb3517afb3e681e78a71a4d0267774d7c25 | 8,427 | null | """Schedule-Free gated recurrence with outer momentum beta 0.95."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
ass... | {
"id": "42c7ae9c-711a-4c52-913f-e6c57d402109",
"created_at": "2026-08-23 01:17:06.234092+00:00",
"db_md5": "ada6b1b04f6d6fde6e95ffea7d7deff7",
"submitter": "Kaloyan Penchev",
"github_login": "k-penchev",
"run_id": "93a65e66-71d1-4903-9448-12ac974d50c1",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 2.338625133037567,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.697514533996582,
"example_count": 100,
... |
42cb0ca2-bc52-4926-b1bf-0ad450543c3e | easy | LauraGomezjurado | 2026-08-30 21:37:35.440679+00:00 | succeeded | 3a4c84f399938bc23b62858a45f8ee73b6cd377aaf13be77d425b34b7ebedf70 | 9,045 | 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": "42cb0ca2-bc52-4926-b1bf-0ad450543c3e",
"created_at": "2026-08-30 21:37:35.440679+00:00",
"db_md5": "a272a8bd2d2a38736cc4b3cff7b0e401",
"submitter": "LauraGomezjurado",
"github_login": "LauraGomezjurado",
"run_id": "e77b3f5b-28d9-4bfc-8f54-68ea9cb7f095",
"tier": "easy",
"dataset_id": "e7",
"st... | {
"score": {
"mean_loss": 2.8323974533686562,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.053832442527919105
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1820156574249268,
"example_count": 85... |
42cc41b8-cf20-4cae-8f14-bc8c0a46c1d6 | easy | jordanrubin | 2026-08-14 01:34:57.999091+00:00 | succeeded | 69ece06bd94846b5a380dc94d4c00c6798b85f8c1e9f0ba91ba31bcdb9376a77 | 29,580 | 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": "42cc41b8-cf20-4cae-8f14-bc8c0a46c1d6",
"created_at": "2026-08-14 01:34:57.999091+00:00",
"db_md5": "0cdede812da1d9957f86c9ea89ed1f0f",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "87fbc6e8-79d3-4fd7-8b01-dc1a9d08a087",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.5058358892574484,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08583333141480884
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.128060714412698,
"example_count": 600,... |
42cfa20d-60c8-494f-8822-03cb2dd06bc3 | easy | shirvani-jr | 2026-08-27 00:09:22.182877+00:00 | succeeded | fe3fda400ac8eb48e0af5aab2e4632776465c4a7ed9c08c75e20f2ae775f7b44 | 26,962 | 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 represent
the prompt, what to store in a bank of categorical registers, a reusable
transition circuit applied se... | {
"id": "42cfa20d-60c8-494f-8822-03cb2dd06bc3",
"created_at": "2026-08-27 00:09:22.182877+00:00",
"db_md5": "eab28881ee62fa5b14c92e3cb55af10d",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "165a5bf6-d452-4f10-9dc2-247e634dc25b",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.9745791554450989,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333334140479565
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9898244142532349,
"example_count": 1... |
42dd316a-ecf9-4090-93c5-ad25a1ef7af4 | easy | 0xDEADD06 | 2026-08-12 13:53:56.232492+00:00 | succeeded | 9fff0c8f734f8f5ad0dbc75dbcf6e5a919ac1856db9ceed45bfaf1eda12e4ce4 | 3,980 | null | """ParaGRU: recurrent GRU 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": "42dd316a-ecf9-4090-93c5-ad25a1ef7af4",
"created_at": "2026-08-12 13:53:56.232492+00:00",
"db_md5": "28a81ab800a87b8ca6adbcac37ac4f91",
"submitter": "0xDEADD06",
"github_login": "0xDEADD06",
"run_id": "2b99889d-89cc-4fdd-ac1c-ce38b2d18ad6",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 3.1330254609944777,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010416666977107525
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.293783705464393,
"example_count": 600... |
42e0b684-1158-49cb-a6de-9787b45f69e6 | easy | oupadhyay | 2026-08-28 02:29:01.067754+00:00 | succeeded | c7c9f1c1e86c6fe49731adf84381680ac4db9c34183201cc57f020236fe3523e | 10,054 | null | """A fixed-capacity relational digit transducer with continuous latent recurrence."""
import torch
from torch import nn
from benchmark import OptimizerBundle, Submission, assert_model_state
PAD, N, X, T, ANS, DIGIT = 0, 2, 3, 4, 5, 7
D, HEADS, MAX_PLACES, SCRATCH, MAX_STEPS = 64, 4, 16, 4, 64
STATE_ELEMENTS = 160_202... | {
"id": "42e0b684-1158-49cb-a6de-9787b45f69e6",
"created_at": "2026-08-28 02:29:01.067754+00:00",
"db_md5": "830240dbaae62da5c4a5ca9ede32ae48",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "76955de9-445c-4f56-925b-845fac983278",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 5.268832206726074,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013513513840734959
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.4847187995910645,
"example_count": 60,... |
42e37713-8506-435c-b9b3-5ca8b09272c3 | easy | erdavis0 | 2026-08-26 03:12:37.899896+00:00 | succeeded | f323ea27a95b25c17ef240c64fd37d3915d7f3e1d86c503b6dc262e4235c4217 | 30,352 | null | """Latent-role continuous REV C1: selector-oracle distillation.
Public delimiter syntax only routes opaque token categories into distinct,
right-aligned N, X, and T tapes. Learned local cells compile separate
interaction, carry, and modulus-conditioned workspaces. N is isolated from
interaction and carry and first e... | {
"id": "42e37713-8506-435c-b9b3-5ca8b09272c3",
"created_at": "2026-08-26 03:12:37.899896+00:00",
"db_md5": "1658fcec88828e061d8640031c475500",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "78e44446-a6c4-454f-a6ad-7bc2a24658f1",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",... | {
"score": {
"mean_loss": 3.44337936652846,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010833333420256774
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.409446935482624,
"example_count": 600,
... |
42e68a4c-2a6e-4d52-8f55-7358cc536f73 | easy | Dandandan | 2026-08-04 23:06:12.101779+00:00 | succeeded | 4cb9654c6146f8e988c53f41af9d38983820356a2441e2329090f03440641fc8 | 28,109 | null | """Multi-start learned periodic recurrence for One Layer Deeper.
Every coefficient that determines a prediction is randomly initialized and
updated end to end. Randomly initialized program selectors compose stable
arithmetic interactions and are unrolled to the prompt's training depth;
self-calibration and an integer-... | {
"id": "42e68a4c-2a6e-4d52-8f55-7358cc536f73",
"created_at": "2026-08-04 23:06:12.101779+00:00",
"db_md5": "fc64a48db4faee986a9210b901de1e71",
"submitter": "Daniël Heres",
"github_login": "Dandandan",
"run_id": "3c687606-c256-47b7-9741-65c3899a7936",
"tier": "easy",
"dataset_id": "e4",
"status": "suc... | {
"score": {
"mean_loss": 1.1314449047815334e-05,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.1313680157761708e-05,
"example_count": 1200,
... |
42e84227-1d6b-4bb4-b361-1039cf32fff8 | easy | DDanlov | 2026-08-18 10:37:02.822027+00:00 | succeeded | 9e2e886a7f72218c6f6788250c79c4edcba3908cfb09a02021dbdad2941008ba | 23,584 | 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": "42e84227-1d6b-4bb4-b361-1039cf32fff8",
"created_at": "2026-08-18 10:37:02.822027+00:00",
"db_md5": "d6174d2a9211082098f18de894eca987",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "f6866bd4-cfcf-4a6b-bacd-0452dce4cba8",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.040294085456275,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666703919571
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0022640628601187,
"example_count": 100... |
42ed6d5f-a9c4-4c5f-a251-355a67b97a6c | easy | arthurfeeney | 2026-08-07 02:19:43.220348+00:00 | succeeded | d9fb2684b825a5bbd2dbfc094b68e1650fce99bfdff860d4a656227eb9cade36 | 7,219 | null | r"""
Notes
1. goal is basically learn y = G(x_0, T, N).
G is always a recurrence `x_t+1 = f(x_t, t) mod N`.
Shouldn't be possible in general since f isn't determinable from a dataset...
I.e., multiple f can generate the same training dataset.
2. can't really use any info on structure of f...
"""
from __future... | {
"id": "42ed6d5f-a9c4-4c5f-a251-355a67b97a6c",
"created_at": "2026-08-07 02:19:43.220348+00:00",
"db_md5": "f4f8c4ce43bd562a48da598319a5e229",
"submitter": "Arthur",
"github_login": "arthurfeeney",
"run_id": "dbc8ae86-3f10-48d7-a7a7-9c3aaeddf78e",
"tier": "easy",
"dataset_id": "e1",
"status": "succee... | {
"score": {
"mean_loss": 2.8398653268814087,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0283333333209157
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.0075883865356445,
"example_count": 100,... |
42f72f76-d365-4a08-90e1-f006dc4df4a3 | easy | DDanlov | 2026-08-28 18:34:21.187716+00:00 | succeeded | e9fcd42a47651f3a7af880c65b37a4c41b33aa3e1653795b1a7c760cb8c5efd8 | 15,206 | null | """
Model A Variant: Baseline Arch1 (K=1, continuous z0 + zt injection, DKS, RohanShampoo)
Config: D=384, H=6, d_ff=1536, T_rec=64, 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
t... | {
"id": "42f72f76-d365-4a08-90e1-f006dc4df4a3",
"created_at": "2026-08-28 18:34:21.187716+00:00",
"db_md5": "119cb4578ba3f60d8c9f0905c1498772",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "8933874b-99d4-4137-bbc8-b5ece08aa9ad",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.290271280863079,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00875
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.305264394245875,
"example_count": 600,
"e... |
42f9a8b7-e8b1-482f-a3dd-b0d87fdc81f1 | easy | viridale | 2026-08-13 21:53:37.099297+00:00 | succeeded | 40e00468e7689bdf9ab456c542d380dc829b72af8306dd44104253533dcd1339 | 25,430 | null | """Explicit register products with stochastic bit reduction.
hypothesis: explicit generic learned register-pair product streams remove latent
operand search, while stochastic two-sided bit-column control marginalizes carry
and reduction paths instead of committing to one brittle affine trajectory.
axis: arch/l... | {
"id": "42f9a8b7-e8b1-482f-a3dd-b0d87fdc81f1",
"created_at": "2026-08-13 21:53:37.099297+00:00",
"db_md5": "ca01ea97966dd0387932dd8a76fc17e8",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "73a8795f-dc93-40bf-b33a-d27a3b90b09b",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 18.010025024414062,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05833333171904087
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 18.5866641998291,
"example_count": 100,
... |
42fa7d32-f7b2-4716-8fa4-fba4771f800a | easy | DDanlov | 2026-08-24 11:22:01.406408+00:00 | succeeded | dde261269ab63a58cd464bcf45f5c69fb17ac0049867e704380b0d97c2d98585 | 43,289 | null | """
Unified Hybrid Masked Deliberation Model (44.0M Parameters) - Pure Skipless, No Cross-Attention
Architecture:
- Scale: dim=1536, num_heads=24, d_ff=6144 (Exactly 43,978,752 parameters)
- Hybrid Prefix-Bidirectional + Causal Target Attention Mask
- Token Layout: [Prompt (L_p), 4 Pause Registers, Target Tokens (L_tgt... | {
"id": "42fa7d32-f7b2-4716-8fa4-fba4771f800a",
"created_at": "2026-08-24 11:22:01.406408+00:00",
"db_md5": "32895fc48922153a0675d5dff8cc0bce",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "cde2a810-b8cb-4469-8d99-4cf4a3a8cb08",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.311038200573245,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0075
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.34158647541508,
"example_count": 600,
"exa... |
43001680-5590-4285-93b1-70be821e879b | easy | shreyash-chonkie | 2026-08-24 07:03:33.509244+00:00 | succeeded | c3695b8bfc9cf1913b213b4c8e46e1f19e81aa64d832d4615d0df0679c365f3a | 8,233 | null | """Four-stage recurrent register Transformer for One Layer Deeper."""
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 = 1024
... | {
"id": "43001680-5590-4285-93b1-70be821e879b",
"created_at": "2026-08-24 07:03:33.509244+00:00",
"db_md5": "089c7bc3558fdaef1af6b37e19105e43",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "d2537728-7dab-4dff-bc31-575cf1c2ae0b",
"tier": "easy",
"dataset_id": "e6",
"status": "... | {
"score": {
"mean_loss": 2.845352292060852,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.043468470685184
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.733504056930542,
"example_count": 60,
... |
430bb8b7-06d3-42a9-8e19-70781f53077a | easy | rishivg | 2026-08-04 05:37:25.060331+00:00 | succeeded | b35bd60a4eefe9eaa88af1035fd71eaf7f05dd96f6480bc1310654bdba9b93eb | 131,327 | null | """Looped transformer with place-value coordinates, Muon, and a clock-paced LR.
Design notes, in rough order of expected value:
1. Weight-tied recurrence with input injection. The task is function
composition of depth ``T``, so one block applied ``LOOPS`` times is a much
better prior than a single feed-forw... | {
"id": "430bb8b7-06d3-42a9-8e19-70781f53077a",
"created_at": "2026-08-04 05:37:25.060331+00:00",
"db_md5": "5cdc8956d07eb3fe563d5d24dfbe06ee",
"submitter": "Rishi Gottumukkala",
"github_login": "rishivg",
"run_id": "1d398399-b866-4589-af2b-1be4cad77ff5",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 14.713037967681885,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014166667126119137
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 14.342918395996094,
"example_count": 60... |
430ec584-881b-42d0-9b5d-74140b753980 | easy | khushidahi | 2026-08-20 22:46:36.818384+00:00 | succeeded | 7df75f30377f104924804f6db69799b0a9305505758e71fb46a472d95de9cc0a | 30,622 | null | """R321: balanced fixed-loop answer/workspace TRM for Hard.
This is a learned architecture, not an arithmetic program. Decimal fields are
rearranged into categorical tapes, but no tensor is converted to a native
number and no multiplication, reduction, quotient, carry, or task transition
is specified. All such seman... | {
"id": "430ec584-881b-42d0-9b5d-74140b753980",
"created_at": "2026-08-20 22:46:36.818384+00:00",
"db_md5": "9e3b4e3c8844590981dcf0bc8f20c489",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "b74e28eb-b0d6-4cfe-936a-c2e136a40398",
"tier": "easy",
"dataset_id": "e4",
"status": "succ... | {
"score": {
"mean_loss": 2.168360922441482,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0017708333333333335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.167864799143131,
"example_count": 120... |
430fd342-98e4-4c96-996a-5e9de8cbdf7a | easy | DDanlov | 2026-08-23 21:53:32.600188+00:00 | succeeded | 1f782ab9e82ba2eeab0e803321b02cd9d3ce331447bd2535a6a4f0b419ffd5cf | 31,834 | null | """
Unified Hybrid Masked Deliberation Model (71.9M Parameters) - Tier 1 (Pure Skipless, No Cross-Attention)
Architecture:
- Scale: dim=2048, num_heads=32, d_ff=8192 (Exactly 71,889,920 parameters)
- Hybrid Prefix-Bidirectional + Causal Target Attention Mask
- Token Layout: [Prompt (L_p), 4 Pause Registers, Target Toke... | {
"id": "430fd342-98e4-4c96-996a-5e9de8cbdf7a",
"created_at": "2026-08-23 21:53:32.600188+00:00",
"db_md5": "bb622dec1a621956e22f8121aea6d68c",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "1f4d63f7-eb6b-48c0-a287-0382d42ef857",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.6317884145980375,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.640164305810971,
"example_count": 60... |
43148a2b-58df-4eef-aa2e-8ce2807b5052 | easy | viridale | 2026-08-15 03:07:18.216904+00:00 | succeeded | 222cf0b0e2f8fb03230f960acb9d089fd133f6e67b6cc4e86dae0c764670ae36 | 42,027 | null | """Clip-normalized one-pass Cayley expert with translation-shared fallback.
hypothesis: a proof-triggered dataset-global boundary expert and a learned state-general scan cover complementary identifiable and non-atlas tasks without halving optimizer decisions.
axis: arch/loss/opt
target: P127 controls plus unchanged of... | {
"id": "43148a2b-58df-4eef-aa2e-8ce2807b5052",
"created_at": "2026-08-15 03:07:18.216904+00:00",
"db_md5": "17871de6e39baceb85db53b5aa3d8aba",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "e17afd9f-f5d0-4a0a-9283-e5f9f8f95277",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 3.3629742860794067,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.12333333492279053
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.7223994731903076,
"example_count": 100... |
431a2426-9244-49f2-8e53-85bcd2f7cf12 | easy | DDanlov | 2026-08-21 02:05:25.406429+00:00 | succeeded | 427bbc47b55d69da20c8db0a30565c9a72efaefca6a9c2225c200f079f0ed2c6 | 25,118 | 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": "431a2426-9244-49f2-8e53-85bcd2f7cf12",
"created_at": "2026-08-21 02:05:25.406429+00:00",
"db_md5": "162b1fb8a03a2b79d504dda4e81ad2e8",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "ae9f0a9d-7b2b-4844-8d71-750b9a050075",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.7841803895540056,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7797018194679723,
"example_count": 60... |
431d0b8b-a645-4720-bda8-f34aa0738e9b | easy | yashkant | 2026-08-31 19:05:57.387391+00:00 | succeeded | cc5d2ad30f57bf801c8a4e51f9a1eb5869bc13ebd80bd8ae8947a98d6f154ade | 106,951 | 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": "431d0b8b-a645-4720-bda8-f34aa0738e9b",
"created_at": "2026-08-31 19:05:57.387391+00:00",
"db_md5": "b993eb5178e0e97ca19e74f6102bd017",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "3873d74f-04b0-48b1-8530-1b6babc0b54d",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 4.619848728179932,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.7009429931640625,
"example_count": 100,
"exac... |
431e705c-8cec-4de4-91e0-ebb46aea4d88 | easy | himalalps | 2026-08-07 16:01:43.698603+00:00 | succeeded | f0bfadf8f9e7b4073fa80fe4f61cdb035d13e566e9af0658a9a4fcf6559bda9c | 69,773 | null | """Algorithmically constrained latent recurrence for One Layer Deeper.
One masked cross-attention pass creates two independent representations:
``state_0`` from X only and an immutable context from N only. T is excluded
from both representations and is used solely to choose how many times the
learned recurrent MLP tr... | {
"id": "431e705c-8cec-4de4-91e0-ebb46aea4d88",
"created_at": "2026-08-07 16:01:43.698603+00:00",
"db_md5": "e2fecf4d821c25a9a976aad8be80055a",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "1a804d31-c013-464e-9b5e-4b3ddcee4ce4",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 3.1077595865270036,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666707023978
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.261776924133301,
"example_count": 600... |
43234ed9-e60b-4249-b1db-fd4170b20b42 | easy | sapient-sapiens | 2026-08-20 15:20:42.831821+00:00 | succeeded | 46c5da5656337e1c91212746ead09b87499b3e301692a2610e41b146ec2bdcfd | 34,760 | null | """Categorical-modulus-conditioned recurrent operator.
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 separate deco... | {
"id": "43234ed9-e60b-4249-b1db-fd4170b20b42",
"created_at": "2026-08-20 15:20:42.831821+00:00",
"db_md5": "26900ab744e48e147b5acce950baa4ea",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "8843f125-a8ad-4104-b6a9-dc94a5f86619",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.263749876161442,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333358168601
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2659249337799348,
"example_count": 600... |
4325834a-65cf-4204-895a-cf2ba2a711f7 | easy | DDanlov | 2026-08-21 21:51:23.907887+00:00 | failed | 56f6d8a2dc9f27fee535e49c32c6c3308b0b30b02d5518c279f5001ca8439b36 | 27,620 | null | """
Unified Pure Skipless Recurrent Deliberation Model (14.02M Parameters)
========================================================================
- Dimension: 660, Heads: 22 (head_dim = 30)
- FFN: 4-stage SwiGLU (d_ff = 2640)
- Positional Encoding: Rotary Position Embedding (RoPE) in Self and Cross Attention
- Recurr... | {
"id": "4325834a-65cf-4204-895a-cf2ba2a711f7",
"created_at": "2026-08-21 21:51:23.907887+00:00",
"db_md5": "7110d53433ab62246b63f3499f23fc60",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "9ad5a55b-317a-4a77-9d96-5de1b95dc0b2",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
43305765-0e3e-4df9-862b-6cbd30dd4ab2 | easy | viridale | 2026-08-28 11:16:07.483617+00:00 | succeeded | ee90717fce2713e9b6f0165bd4719089c43365d58b37860bde9b46ca88c70637 | 32,318 | null | """Shallow-evidence absolute-coordinate learned coefficient slab.
hypothesis: v974 selects the dominant same-N group but preserves dataloader order,
causing its complete 63.7M-program evidence pass to execute up to the deepest of
four arbitrary rows. For learning a shared one-step cell and certifying the requi... | {
"id": "43305765-0e3e-4df9-862b-6cbd30dd4ab2",
"created_at": "2026-08-28 11:16:07.483617+00:00",
"db_md5": "f40d1412f888f4107e7749447a37159c",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "6536cf19-0466-4b11-a3d8-177315f160f1",
"tier": "easy",
"dataset_id": "e10",
"status": "succ... | {
"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": 125,
"exact_accuracy": 1.0,
"... |
433346e6-4825-4c92-b282-02b6ebf75171 | easy | karanganesan | 2026-08-06 05:47:31.185215+00:00 | succeeded | 44afcfafc4d049a26f4f791e2374aa9270ab891f5af4e3be4b591330a0028758 | 25,616 | 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": "433346e6-4825-4c92-b282-02b6ebf75171",
"created_at": "2026-08-06 05:47:31.185215+00:00",
"db_md5": "905cc4dfb44c8d0ec559173c60e66bbb",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "eaf940c0-854d-4231-bc46-eee34f064c0c",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.3262561559677124,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014999999664723873
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.222905158996582,
"example_count": 100... |
43359583-d58f-4ae1-adc9-0c50d43c2b8b | easy | khushidahi | 2026-08-21 17:18:28.622132+00:00 | succeeded | 6fdcc86aa244153456b66e0c1e9a2d99152468752210505c7c4879cad275f54c | 29,571 | null | """R343: update-matched eight-visit inner-recurrence arm of R340.
This submission tests the strongest rules-safe generic baseline suggested by
the public competition discussion. It does not implement squaring, modular
reduction, or any other task solver. Numeric prompt fields are only rearranged
onto equal-length, r... | {
"id": "43359583-d58f-4ae1-adc9-0c50d43c2b8b",
"created_at": "2026-08-21 17:18:28.622132+00:00",
"db_md5": "e2d6ff2b09c7fb9de2c94d3a3d4a6065",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "ddaab710-2598-482d-a5ca-1fdace10b631",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 0.7511908113956451,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.8148649036884308
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.45561712980270386,
"example_count": 60,... |
4339dc11-a540-481d-b6d6-b2d063fbd0a2 | easy | khushidahi | 2026-08-06 04:03:02.409895+00:00 | succeeded | 420c0204ffa0517e1c19259874dc4f803dc16ebb2c19057d70183ce7a19e444b | 14,402 | 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": "4339dc11-a540-481d-b6d6-b2d063fbd0a2",
"created_at": "2026-08-06 04:03:02.409895+00:00",
"db_md5": "6303218fcfec4b7afd1327fe0c940cb3",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "5b0e42d2-0a5d-40c6-8bc5-1411758fde50",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1875118017196655,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011250000447034836
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1861536502838135,
"example_count": 60... |
4341034a-d89d-4da1-a0db-a89453033c47 | easy | sapient-sapiens | 2026-09-01 01:04:03.607935+00:00 | succeeded | c50e29b7579dc4acd102df5d006ddcdd55cc4477995766e7a1764fb5f309bc87 | 24,835 | null | """E5 transfer of the current D6 shared-character residual nested square."""
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,... | {
"id": "4341034a-d89d-4da1-a0db-a89453033c47",
"created_at": "2026-09-01 01:04:03.607935+00:00",
"db_md5": "a7424d760ed9f3906c758cc2375226fa",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "b2adf3ae-5ffb-4716-a005-a231fb729697",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.345920327680324,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.349975780253988,
"example_count": 600... |
43484d42-20c7-4745-aad3-0743f1ff791d | easy | ehonig | 2026-08-14 01:07:39.406353+00:00 | succeeded | 651ce58492e147c9c0c3e56c77e0291eb53224091e02c58eea3974270d487af8 | 49,676 | null | """DigitLoop — a place-value recurrent model for repeated modular squaring.
The task is ``y = x^(2^T) mod N`` with the answer read back tail aligned, one
decimal place value per token position from the right. A stack of attention and
MLP layers has no primitive for the two operations this needs — a partial-product
co... | {
"id": "43484d42-20c7-4745-aad3-0743f1ff791d",
"created_at": "2026-08-14 01:07:39.406353+00:00",
"db_md5": "e6cb7469f33c47cc7f8f1d06da83b86f",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "711a939e-d31c-474b-b7de-3a4d1f2b4cdc",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 4.317456245422363,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.2916666567325592
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.504920482635498,
"example_count": 100,
... |
434e57a5-0f90-4b82-86d5-64ac40d70113 | easy | gauravmishra | 2026-08-10 03:25:03.405045+00:00 | succeeded | 350d876cd225f4280d9ab80241d4801218d0cb1f520cd062dea33a0a18df0719 | 7,374 | null | """E09: K4 hidden matrices with Muon and auxiliaries with AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)... | {
"id": "434e57a5-0f90-4b82-86d5-64ac40d70113",
"created_at": "2026-08-10 03:25:03.405045+00:00",
"db_md5": "9d545b570d03c80ff7dfcd6d98cf71c0",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "ec676384-1072-4b44-af48-83f3b7e286a5",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 8.73051135247945,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333370586236
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.608678612473835,
"example_count": 600,
... |
4351cdd3-128a-42fb-a7bd-4ec7785070e4 | easy | liam-gb | 2026-08-12 12:00:57.961553+00:00 | succeeded | 63560835218f6cc8d2426c2dfe473ed6ed0b400ea97d7ccbdbf29d9eabbb74fc | 13,923 | null | """rns_pm_gather_ct_wide: the champion with a dataset-sized digit head.
Single change vs rns_pm_gather_ct: the readout carries one
distance-from-units slot per sequence position instead of three, so
answers longer than three digits are reachable. Behaviour on shorter
answers is unchanged — the extra slots are simply n... | {
"id": "4351cdd3-128a-42fb-a7bd-4ec7785070e4",
"created_at": "2026-08-12 12:00:57.961553+00:00",
"db_md5": "6160238a93c003b99ce5e4e8e2555bad",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "19a43c3e-9453-4b99-911e-e3e60cadf350",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 4.6104196310043335,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4349999949336052
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.136412620544434,
"example_count": 100,
... |
435dbc6c-906d-4766-99d2-2475deaeb129 | easy | oupadhyay | 2026-08-29 18:47:04.422761+00:00 | succeeded | 2c22960c47c04744fcec5acae44b161d3cf914642ac0d54c71de767c75befc57 | 3,994 | null | """Conditional optimizer phase 3: warm8_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... | {
"id": "435dbc6c-906d-4766-99d2-2475deaeb129",
"created_at": "2026-08-29 18:47:04.422761+00:00",
"db_md5": "4aa3204f4eaa733a8e4235b72dd88b75",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "3f837a27-4031-4acd-95f4-a5e1e0a64f03",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 1.966839849948883,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.048648651689291
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9320961236953735,
"example_count": 60,
... |
435ef0ee-fd3b-4b6a-9327-0299367839f1 | easy | isaac0804 | 2026-08-04 17:10:26.235948+00:00 | succeeded | 69f197af1df5b482a6f4c9837a04847d08a22434bf8643b4166d4c173edaf3a0 | 7,552 | 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": "435ef0ee-fd3b-4b6a-9327-0299367839f1",
"created_at": "2026-08-04 17:10:26.235948+00:00",
"db_md5": "5f5ef9a721a4c2b15ee9eb425ece78dc",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "470747b5-d8b0-4ce4-8007-6f4cf48a8cba",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.9722319841384888,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05499999783933163
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9978245496749878,
"example_count": 100... |
435f3af7-a569-489c-913c-52dadcfde264 | easy | jayyala | 2026-08-16 22:40:04.497414+00:00 | succeeded | 67093a88f1d7e42eee17a538c42e3d548ccde180b37a92e5bab1d2e0c5e8fb48 | 13,859 | null | """A scalable recurrent machine with canonical categorical byte states.
Each integer is represented by a short little-endian tape of 256-way slots.
One tied bidirectional mixing cell learns the unknown state transition from the
current tape and the prompt context, then its categorical output is fed back
for exactly T ... | {
"id": "435f3af7-a569-489c-913c-52dadcfde264",
"created_at": "2026-08-16 22:40:04.497414+00:00",
"db_md5": "67c0ab3c7d809072290bb9f45e40f517",
"submitter": "jayyala",
"github_login": "jayyala",
"run_id": "5563661b-a071-41c1-8f9b-64845f13cab4",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 5.334052681922913,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.8466666638851166
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.089552164077759,
"example_count": 100,
... |
4362db26-3bf6-43a2-a531-49e2482e61fc | easy | shirvani-jr | 2026-08-31 03:05:34.285617+00:00 | succeeded | edb22e9fd325e2252fdd8363a839d4097cc78e4b5421b2753528de1fb8941e43 | 29,538 | 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": "4362db26-3bf6-43a2-a531-49e2482e61fc",
"created_at": "2026-08-31 03:05:34.285617+00:00",
"db_md5": "a955506c7a7239ce8b432472fb6d43ad",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "e54dfb68-47c6-4e7a-9ae2-6529da90f22b",
"tier": "easy",
"dataset_id": "e7",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.1367915272712708,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.5857843458652496
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.223343014717102,
"example_count": 85,
... |
4363724c-5756-4fa5-83cd-c82bef33f081 | easy | repst | 2026-08-24 21:57:02.903192+00:00 | succeeded | 42a3699050e0c6cebae70793dedd3199ac5cb9dca2c2dab12a01eeaa057515df | 31,002 | null | # R639: adaptive-width compute-scaled persistent-register machine.
#
# R639 WIDTH-GENERALIZATION DELTA:
# The learned finite-state cell is width invariant. Evaluation now allocates its
# tape from the largest modulus in the current batch instead of silently fixing
# every computation at 18 bits. No weights, program t... | {
"id": "4363724c-5756-4fa5-83cd-c82bef33f081",
"created_at": "2026-08-24 21:57:02.903192+00:00",
"db_md5": "99ed2477a1909b5c69e0226194de22ce",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "c292790d-cac3-4d99-b0e5-a2b2b9e25de7",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 0.06506542112991026,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.06534162840718245,
"example_count": 600,
"e... |
436ac649-c156-487e-8b19-84ed1d8ac458 | easy | gauravmishra | 2026-08-08 06:13:09.211680+00:00 | succeeded | f909f86b3b62d4a59bfa3d0a2d1d42ec6eb92b0601dd73c431cdd4d9431e8bda | 5,473 | null | """Four-block Transformer with four mutable per-forward memory tokens."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 1... | {
"id": "436ac649-c156-487e-8b19-84ed1d8ac458",
"created_at": "2026-08-08 06:13:09.211680+00:00",
"db_md5": "ceae26ba231943df422f344bbdf6c0fa",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "5df04625-70c2-4894-9595-0549446f2459",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 9.578243249596497,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 10.30409972870831,
"example_count": 600,
"e... |
4370148f-705b-4921-bf8c-7dc6f009ab64 | easy | sapient-sapiens | 2026-08-15 21:08:46.408041+00:00 | succeeded | 469ae48173121f8a328b5c1f44d0bff0e2f098d243dcd227975cee12399b6829 | 17,772 | null | """T-only adaptive chunk routing over a learned shared operator.
The controller sees only the categorical tokens inside the T prompt field and
is trained solely by final-answer loss. It never parses T numerically and T is
not used as an intermediate supervision target. Inputs use learned additive
token, semantic-gro... | {
"id": "4370148f-705b-4921-bf8c-7dc6f009ab64",
"created_at": "2026-08-15 21:08:46.408041+00:00",
"db_md5": "38cda1d694d318221e86f82dd23bbc39",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "61a5e39d-f0d2-46ea-aeb1-74ce66ddcf15",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 2.1342633210512765,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03499999870856603
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.024549722671509,
"example_count": 100,... |
4372eb46-b15a-4720-9184-3dcaedb8fdf2 | easy | sapient-sapiens | 2026-08-23 10:03:43.687897+00:00 | succeeded | a23bdfb6394e19f95069f73f9e0fc4a79c3ccc6b6c158df3cf70454f48eaf64c | 22,055 | null | """Approximately 10x-scale adaptive child of the D6 pair/quotient architecture.
The architecture uses the required learned additive token/group/left-relative
representation. Marker-derived field lengths additionally route categorical
digit embeddings into right-aligned lanes; this alternative place alignment is
an ex... | {
"id": "4372eb46-b15a-4720-9184-3dcaedb8fdf2",
"created_at": "2026-08-23 10:03:43.687897+00:00",
"db_md5": "8ead2b36fb2068f23eba85cbe19e2f1c",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "4bcf41be-0629-421e-8b42-a418154351bb",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1487144448434394,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.152522169020144,
"example_count": 600... |
43758181-2156-4b4a-a718-a873a6cb487f | easy | repst | 2026-08-28 21:58:30.671471+00:00 | succeeded | 6d7bd4777d58cb9e66d552a7e2644b3e39eb56f497dacf2d0c64e1af426af0e1 | 55,483 | null | """
R794: endpoint-learned declarative recurrent programs with compiled execution.
Learned jointly from ordinary endpoint answer digits:
* execution modulus offset;
* recurrence-clock offset;
* setup, recurrent transition, and readout expressions; and
* decimal digit-token lexicon.
The task vocabulary contains only ... | {
"id": "43758181-2156-4b4a-a718-a873a6cb487f",
"created_at": "2026-08-28 21:58:30.671471+00:00",
"db_md5": "7fcd8d4477c7ae397b14a98a0f526676",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "660956c7-04e9-4c03-bb6d-db50f904ca44",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 0.41005186298095675,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.4099893271923065,
"example_count": 100,
"ex... |
4378c3bd-6f52-41a1-affd-38aed6cf0e1f | easy | alirezashirvani-jr | 2026-09-01 00:52:42.581733+00:00 | succeeded | 165528669f04dbd5c7f71d3a663f274dc291e32a119b96831234efb80b98a5c8 | 33,617 | null | from __future__ import annotations
import math
import time
from contextlib import nullcontext
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
MLP_EX... | {
"id": "4378c3bd-6f52-41a1-affd-38aed6cf0e1f",
"created_at": "2026-09-01 00:52:42.581733+00:00",
"db_md5": "61965aa1bc38d536f47dfceb82dc6d36",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "f55450eb-ff14-4684-958c-71a62e2e85f2",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 2.848644410100606,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.030833333258827526
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8860645294189453,
"example_count": 600... |
4384a348-5082-4a4a-bf4a-a536cf57350d | easy | sapient-sapiens | 2026-08-04 15:59:57.976453+00:00 | succeeded | f823bd965a4cc756ba381dcd25ad7b5460f8eaf5cb5f2709817928a3d6adb980 | 3,659 | 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": "4384a348-5082-4a4a-bf4a-a536cf57350d",
"created_at": "2026-08-04 15:59:57.976453+00:00",
"db_md5": "4284cd6becbf0eaeddc953eb51695222",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "ab2c27c1-1017-451f-a9a9-f0c2735ceb29",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 4.530428424737697,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0066666667660077415
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.956821441650391,
"example_count": 100... |
438731a2-4c59-43c4-9425-7ac5d3d2b17a | easy | DDanlov | 2026-08-22 15:40:48.500035+00:00 | succeeded | e40ecde14da3663b14f8ad842ae6efca4965363fbab3c8eb095b3b49e194df2d | 30,670 | null | """
Unified Hybrid Masked Deliberation Model (85.3M Parameters) - Easy Tier (e5) WITH Self-Conditioning
- Scale: dim=1536, num_heads=24, d_ff=6144 (85,315,584 parameters)
- Hybrid Prefix-Bidirectional + Causal Target Attention Mask
- Self-Conditioning: TRUE (Continuous Softmax Probability Feedback from DiffusionGemma)
... | {
"id": "438731a2-4c59-43c4-9425-7ac5d3d2b17a",
"created_at": "2026-08-22 15:40:48.500035+00:00",
"db_md5": "41d1427e0a845721b476997ef6965080",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "84518755-5636-41e1-9477-c9ea5e8afb99",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 34.61864303556234,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 34.402680007866145,
"example_count": 600,
"exac... |
4389c948-e8d2-4632-9707-f2f92417b047 | easy | khushidahi | 2026-08-06 05:25:29.440529+00:00 | succeeded | 62af8441a10cde9937d824c8e9ebef5896eb8203f7f8cd1078d6113aa535fc59 | 14,691 | 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": "4389c948-e8d2-4632-9707-f2f92417b047",
"created_at": "2026-08-06 05:25:29.440529+00:00",
"db_md5": "b712590409c7bee4fc742ac581a6f2b5",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "b27e797d-c278-474f-8c27-2218e379c74a",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.230131983757019,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333656191826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2300710678100586,
"example_count": 300... |
438f7843-1db4-41a3-a0b6-6571a9507768 | easy | erdavis0 | 2026-08-26 04:35:17.804343+00:00 | succeeded | 6fe6fdf46bcdc9c76ffef64fbdc6b302f2262f8879ac4d7daea4e62dccf8b390 | 30,354 | null | """Latent-role continuous REV C5: hybrid token CE and sequence mixture.
Public delimiter syntax only routes opaque token categories into distinct,
right-aligned N, X, and T tapes. Learned local cells compile separate
interaction, carry, and modulus-conditioned workspaces. N is isolated from
interaction and carry and... | {
"id": "438f7843-1db4-41a3-a0b6-6571a9507768",
"created_at": "2026-08-26 04:35:17.804343+00:00",
"db_md5": "14ff631f83062f36483f24b1bd94c4c9",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "9dfad443-5824-485d-bdf7-acc77aeea71f",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 4.220247507095337,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0566666666418314
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.360044956207275,
"example_count": 100,
... |
43951bcf-443d-47a4-8759-a95d1a26be14 | easy | 0Chris5R | 2026-08-07 10:06:17.333234+00:00 | succeeded | 2d7c46f039c584fb60a24e4f59a1044967875136e4fa693d5f964f094bef186f | 14,519 | null | """Learned convolutional-GRU over an aligned field-by-digit workspace.
The public field markers and decimal significance define only the layout of
the neural workspace. Every transition and output is learned; there is no
arithmetic routine, recurrence assumption, lookup table, or generated target.
"""
from __future__... | {
"id": "43951bcf-443d-47a4-8759-a95d1a26be14",
"created_at": "2026-08-07 10:06:17.333234+00:00",
"db_md5": "5c8a419fdd1c13a47c10fcf08d6c9c30",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "bc689de7-7f9b-4a5c-b96f-7cb2fca8d5ec",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.382078916505049,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010833333358168603
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.1629262569239325,
"example_count": 600... |
43a9c40d-0c7d-4864-8bf3-896fb764d9e3 | easy | alirezashirvani-jr | 2026-09-01 00:03:36.020612+00:00 | succeeded | a533a947fbf8209e3cc9eadd7af3947f1b398f6895c1cf1eed1403cde5c9eaa5 | 21,686 | null | from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
HP = {
"maximum_outer_steps": 64,
"batch_size... | {
"id": "43a9c40d-0c7d-4864-8bf3-896fb764d9e3",
"created_at": "2026-09-01 00:03:36.020612+00:00",
"db_md5": "2ff358139a85301adde20a7d30e1169f",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "12f05c21-478c-43c1-825c-bde8a476ad71",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 2.5749144768106467,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333333333333335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.572237990094942,
"example_count": 60... |
43ac366e-fbe6-4219-ab2c-d3174fbdcfa6 | easy | sapient-sapiens | 2026-08-30 21:37:32.167251+00:00 | succeeded | 8cd245609f1b9b1c208a073ca5ee6769f91c73b37c69542241b51e6f7ce34ff7 | 17,403 | null | """Eligible E5 direct-Horner model with separate S-compute and A-commit blocks.
The two blocks have identical geometry but independent weights. At every one
of four Horner positions, the S block writes latent scratch once and the A
block consumes that scratch to write answer state once. N and X are immutable
within ... | {
"id": "43ac366e-fbe6-4219-ab2c-d3174fbdcfa6",
"created_at": "2026-08-30 21:37:32.167251+00:00",
"db_md5": "1996801c41e7c56f35e33a261bf06010",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "2f19416d-0c93-4eec-976d-94dc3512b3b8",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.151315522434616,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1525971386999294,
"example_count": 600... |
43c00e30-df64-4573-b296-a5768ea996e9 | easy | blake-camp-surge | 2026-08-04 16:27:09.570337+00:00 | succeeded | 4966bd38a7111e37a686785764acb5467adade180a09cddc5126cb3fc086028f | 15,789 | null | """Vector DANs communicating through recurrent linear-attention memory."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
a... | {
"id": "43c00e30-df64-4573-b296-a5768ea996e9",
"created_at": "2026-08-04 16:27:09.570337+00:00",
"db_md5": "658ffbec0dc204ff7bb962b0175192dd",
"submitter": "Blake Camp",
"github_login": "blake-camp-surge",
"run_id": "e33a954e-2b03-488a-a037-eff141490fe3",
"tier": "easy",
"dataset_id": "e3",
"status":... | {
"score": {
"mean_loss": 2.1635737602799763,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013125000013969837
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1598412360671593,
"example_count": 80... |
43c09b85-d96e-45e3-8b01-00befeff9456 | easy | alirezashirvani-jr | 2026-08-29 13:22:16.583243+00:00 | succeeded | f52aae5f2fa7fbcfa35a2a305f484ddaa7b0587265925de340678caf1f904317 | 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": "43c09b85-d96e-45e3-8b01-00befeff9456",
"created_at": "2026-08-29 13:22:16.583243+00:00",
"db_md5": "06297fe2a13beca444f27a1131bfe93e",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "09f9d4b4-39a6-4410-b15a-fde459e5173b",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 2.93021917124992,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04541666728754838
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.9453957080841064,
"example_count": 600,
... |
43c29d8f-5c97-454b-b737-a0046587f4b8 | easy | epicosy | 2026-08-22 23:11:31.722370+00:00 | succeeded | f2b4fc3990badd482c871fb9e72e934e1c20fe1983061bdeba9a84fdd6146baa | 5,431 | null | """Latent-state Universal Transformer baseline for the Easy tier.
The prompt stream is encoded once and kept fixed. A single shared block
recurrently updates learned latent/query tokens through cross-attention to the
prompt. The final logits remain aligned with the original input sequence, as
required by the public su... | {
"id": "43c29d8f-5c97-454b-b737-a0046587f4b8",
"created_at": "2026-08-22 23:11:31.722370+00:00",
"db_md5": "8c76ca64351aa089c857955699c72afe",
"submitter": "Eduard Pinconschi",
"github_login": "epicosy",
"run_id": "2e9e422e-9232-4016-9836-eb3c7fd559c6",
"tier": "easy",
"dataset_id": "e6",
"status": "... | {
"score": {
"mean_loss": 2.269412040710449,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02004504599608481
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.221578598022461,
"example_count": 60,
... |
43c64f24-fbb0-4b36-8208-21e5a0e8df8e | easy | khushidahi | 2026-08-17 15:45:24.203778+00:00 | succeeded | 3d4a278246081b4164c1e9f9b6058722b795d9950a1ffdd0305c0dd87b9d234d | 26,596 | null | """R203C: R202 plus dense final-label supervision over inner iterates.
This is the third rung of a matched adaptive-recurrence ladder. Architecture,
width, optimizer, input reinjection, and recurrence caps are identical to R202.
The only intervention is the loss: every active inner iterate is decoded and
compared wit... | {
"id": "43c64f24-fbb0-4b36-8208-21e5a0e8df8e",
"created_at": "2026-08-17 15:45:24.203778+00:00",
"db_md5": "2899677c14c36fdbffd69638db2b0984",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "95570b22-676b-4aca-a8af-5bda927d3c25",
"tier": "easy",
"dataset_id": "e6",
"status": "succ... | {
"score": {
"mean_loss": 1.8934924602508545,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.44099099934101105
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8319072723388672,
"example_count": 60,... |
43cc58c7-5dd3-40dc-b15c-7b00c915bc8c | easy | gauravmishra | 2026-08-07 19:07:50.241666+00:00 | succeeded | 088db7a0455abbaaa8c55a2bcc99dc42047c8116798e66a3236997a236e0df75 | 3,682 | null | """Four-pass Transformer that reuses one parameter-tied block."""
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_H... | {
"id": "43cc58c7-5dd3-40dc-b15c-7b00c915bc8c",
"created_at": "2026-08-07 19:07:50.241666+00:00",
"db_md5": "c733c62f23624484c93c5a6a1b0bb7f1",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "34ed18e3-a463-4e2b-bd2e-2c57a2a4b12a",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.2176026105880737,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3862414360046387,
"example_count": 10... |
43d9ac9b-6d8f-4ef0-a943-367610e12575 | easy | sapient-sapiens | 2026-08-31 01:41:08.567470+00:00 | succeeded | 6d09e34c3d6d754f15ff94c62b010645e5752995e4baaf06648c7fda0ca52f8f | 15,761 | null | """Eligible E5 live-X-digit 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. It
receives no step index or step tag; the loop iterates over the four live X
lanes directly. The ... | {
"id": "43d9ac9b-6d8f-4ef0-a943-367610e12575",
"created_at": "2026-08-31 01:41:08.567470+00:00",
"db_md5": "ba60ff1a14856292286a691a349c2263",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "2f1eaf40-3a06-4ae1-b1e7-1191286eaf05",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1780487595333193,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333370586236
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.179561628889075,
"example_count": 600... |
43dd043e-7f03-4349-8923-5b130106c176 | easy | briheet | 2026-08-07 03:22:18.594031+00:00 | succeeded | ece0f4d3d83497d71be5188fa37cab273cb15392b5ca45e5879c67515fc19289 | 4,097 | 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 = 48
NUM_HEADS = 4
NU... | {
"id": "43dd043e-7f03-4349-8923-5b130106c176",
"created_at": "2026-08-07 03:22:18.594031+00:00",
"db_md5": "d3a230c823239c1c72463a3b29337154",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "975b7cc5-b1cb-4672-8a28-0c459b2b500e",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.5390721559524536,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03666666615754366
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.9083194732666016,
"example_count": 100... |
43e71ba2-14df-40e3-9039-181c55a05c84 | easy | khushidahi | 2026-08-05 23:06:21.938751+00:00 | succeeded | eba2605abb8237f6fc3d0aa809bb7402c38bec7e7dc8075e0d64fdce5009fad4 | 15,047 | 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": "43e71ba2-14df-40e3-9039-181c55a05c84",
"created_at": "2026-08-05 23:06:21.938751+00:00",
"db_md5": "e6627e28269b6733e2e729aa1caf0465",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "bb9061f4-8269-4c33-8fa0-e95c5d915af0",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.2338117361068726,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009583333972841501
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2345433235168457,
"example_count": 30... |
43e85a11-5b2c-424e-ae3e-f015d4ae8e2e | easy | alirezashirvani-jr | 2026-08-30 20:30:24.691296+00:00 | succeeded | b9175ea749088ce73f1acb54d926640b8af718ce3781cc20009e94afbafac36e | 23,652 | 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_WIDTH = 128
ATTENTION_HEADS =... | {
"id": "43e85a11-5b2c-424e-ae3e-f015d4ae8e2e",
"created_at": "2026-08-30 20:30:24.691296+00:00",
"db_md5": "90e44baa72c95f059cca19cc200869d7",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "3a5ff16d-e64b-4ce5-a2ba-ff8af81ab235",
"tier": "easy",
"dataset_id": "e4",
... | {
"score": {
"mean_loss": 2.674950666400605,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004270833345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6631363850656764,
"example_count": 120... |
43f1ec1f-3584-4b13-b141-91e87ab313b5 | easy | viridale | 2026-08-06 10:03:41.236562+00:00 | succeeded | 4c0ac8fed7079e4957791f91672761e02e337a4cb24af8fb003de325c844bc1a | 7,667 | null | """Hard-binary straight-through tied categorical interface.
hypothesis: the correct reusable categorical circuit may already arise during
memorization but require weight sharpening; a unit-norm binary token/head table
can enforce that sharpening while preserving the full continuous recurrent core.
axis: arch
t... | {
"id": "43f1ec1f-3584-4b13-b141-91e87ab313b5",
"created_at": "2026-08-06 10:03:41.236562+00:00",
"db_md5": "49f2759b652ab62456821e31568b56e2",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "9d33d3ff-2def-4c03-92b5-3b8e4bcf427a",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 2.2266043424606323,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2532668113708496,
"example_count": 100,... |
440347c7-1e5d-4fb8-8c90-c690efaa81b6 | easy | DESU-CLUB | 2026-08-30 19:16:18.230555+00:00 | succeeded | d03d033bb1a4e0245290f3e6ba2adb31ffb6b1df1390e2f9ccbedcd5b9fb2315 | 15,215 | null | """Discrete-state loop over a place-value tape, trained with orthogonalize-only Muon.
The recurrent state carried between squarings is the *digit string*, not a
residual stream. Each outer iteration reads (N, x) off an LSB-first place-value
tape, emits digit logits for one squaring, hard-argmaxes them (straight-throu... | {
"id": "440347c7-1e5d-4fb8-8c90-c690efaa81b6",
"created_at": "2026-08-30 19:16:18.230555+00:00",
"db_md5": "f51bb980e87aa9b9f12c6b1605ac103d",
"submitter": "Low Keng Hoong, Warren",
"github_login": "DESU-CLUB",
"run_id": "bdb92bb9-d226-40b3-8663-381958cb3d61",
"tier": "easy",
"dataset_id": "e6",
"sta... | {
"score": {
"mean_loss": 1.0646589573932022,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.6436937024464479
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.7399755120277405,
"example_count": 60,
... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.