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 |
|---|---|---|---|---|---|---|---|---|---|---|
37621e6a-48c1-4a83-be1a-36409315b765 | easy | alirezashirvani-jr | 2026-08-23 19:55:12.915429+00:00 | failed | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "37621e6a-48c1-4a83-be1a-36409315b765",
"created_at": "2026-08-23 19:55:12.915429+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "4807cf25-b230-4a36-ac2d-a14677980548",
"tier": "easy",
"dataset_id": "e1",
... | null
|
37651293-ac24-4b61-82d2-c0b6da2beedb | easy | chuk-uzowihe | 2026-08-12 17:11:57.287457+00:00 | failed | 0901f3164653c1d8494716e3cc741f70b48d7353c1ec04b5c9c8382d22185a09 | 69,314 | null | """One Layer Deeper submission: depth-routed MoEUT over a 2D digit lattice.
v9: the recurrent block is a single MoEUT layer (refs/moeut) — SwitchHead
attention + sigma-MoE feedforward — whose routers are conditioned on the
loop depth (Fourier features of the iteration index) so one shared bank can
express phase-depend... | {
"id": "37651293-ac24-4b61-82d2-c0b6da2beedb",
"created_at": "2026-08-12 17:11:57.287457+00:00",
"db_md5": "341554843db97c74f7fff11270e3cee6",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "ab43bb30-e5d3-4f0e-8aaa-1678f9dd739b",
"tier": "easy",
"dataset_id": "e1",
"status": "... | null
|
376723af-4f27-4694-8f7d-2346acb4514c | easy | shreyash-chonkie | 2026-08-24 23:12:33.334111+00:00 | succeeded | 5ec9435bfbf36484b842883140429e916b1a1c8140ad337663a230d72fbdd161 | 9,306 | null | """Recurrent latent operations combined by a learned sum-product pooler."""
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": "376723af-4f27-4694-8f7d-2346acb4514c",
"created_at": "2026-08-24 23:12:33.334111+00:00",
"db_md5": "de7fbbc26a3cd9d61fb103ddaef38d80",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "774c560b-a695-453b-9eeb-0d8e4ff8cf54",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 8.307411193847656,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02333333296701312
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.505904197692871,
"example_count": 100,
... |
376aa008-a7dc-459c-a1f0-a3c6bf29a0b2 | easy | shirvani-jr | 2026-08-29 04:02:14.472623+00:00 | succeeded | 4a66d63dbbbaeed5dc4c55a40cdeeea9e71e8df53320d97d7477aed27bfc777a | 49,200 | null | """Adaptive learned discrete recurrent fabric for One Layer Deeper.
A gradient-trained, operation-free recurrent computational machine. The
model learns, from the competition's endpoint supervision: how to parse the
prompt into a bank of categorical digit registers, a reusable transition
circuit applied serially, a l... | {
"id": "376aa008-a7dc-459c-a1f0-a3c6bf29a0b2",
"created_at": "2026-08-29 04:02:14.472623+00:00",
"db_md5": "aa10d8c1d47f78d3d7b774544bdbdcd6",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "5d25b2c5-9257-4f54-b7c0-982273dacf3a",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.5012683505611912,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.00916666674738129
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5170133113861084,
"example_count": 600... |
37702a41-26d3-4af2-af7c-455575587fe8 | easy | neel04 | 2026-08-11 13:03:03.586428+00:00 | succeeded | 2217c7d0b431339bef241d4131dc14b19e191ef25d502c47d9d63d493f968a60 | 15,544 | null | """Layer 5: stationary NAG with value embeddings and tied cosine decoding."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
BatchReuseContext,
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,... | {
"id": "37702a41-26d3-4af2-af7c-455575587fe8",
"created_at": "2026-08-11 13:03:03.586428+00:00",
"db_md5": "958509afd43cc08944e1cb5d0bef6e25",
"submitter": "Neel Gupta",
"github_login": "neel04",
"run_id": "d61c1e81-da1b-44ef-9513-f56a447941ce",
"tier": "easy",
"dataset_id": "e1",
"status": "succeede... | {
"score": {
"mean_loss": 2.26757550239563,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06833333149552345
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.529310464859009,
"example_count": 100,
... |
3771b997-29c3-4b22-bb90-af6e5e9a639b | easy | DDanlov | 2026-08-18 00:31:29.309361+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": "3771b997-29c3-4b22-bb90-af6e5e9a639b",
"created_at": "2026-08-18 00:31:29.309361+00:00",
"db_md5": "ef8f2b4e45393e17f33e8d6b42c30e5d",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "05a06337-a73d-466c-b108-35f53017a85a",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.319037573892076,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000012417635
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3430212169484705,
"example_count": 600... |
37827867-1479-4325-b4b5-15fee32c459f | easy | alirezashirvani-jr | 2026-08-30 21:40:26.684029+00:00 | succeeded | 0eee95f721e2b8145337475cc01ad6f9ea5e239710c8f9320448f31dd4dbcb08 | 26,505 | 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 = 64
ATTENTION_HEADS = ... | {
"id": "37827867-1479-4325-b4b5-15fee32c459f",
"created_at": "2026-08-30 21:40:26.684029+00:00",
"db_md5": "f87d171ba0f13c4dc5135ff5898b801f",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "ee14757f-7cb5-4074-99a9-b7166d73b6ac",
"tier": "easy",
"dataset_id": "e3",
... | {
"score": {
"mean_loss": 2.2532862424850464,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014374999795109034
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.254537582397461,
"example_count": 800... |
37847c0e-16ae-464a-bdf0-291e8a1ff51f | easy | sapient-sapiens | 2026-08-30 22:56:00.482511+00:00 | succeeded | f86d3aabff7a82c269180f1042669a65dcbad722092ad191843bdb03d24a764a | 19,456 | null | """Eligible E5 direct-Horner model with detached same-final suffix credit.
The ordinary eight-square branch retains full BPTT and weight 0.9. A T-aware
selector chooses a detached predecessor from X/A1--A7, and one fresh tied
square receives weight 0.1 against the same final answer labels. No
intermediate value, depth... | {
"id": "37847c0e-16ae-464a-bdf0-291e8a1ff51f",
"created_at": "2026-08-30 22:56:00.482511+00:00",
"db_md5": "735eec33e106d5cf383991cad89a9c51",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "bf3d1fe6-a1d9-41f0-8344-e6599d42f5a8",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1398997859221085,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1387362041815514,
"example_count": 60... |
3788e788-ccd6-432f-a268-5c1b01792364 | easy | karanganesan | 2026-08-06 04:12:15.232109+00:00 | succeeded | 99b9f33357552a2fff1633bf6749209f176559f7e5879bfc367751033ccee0e2 | 23,822 | 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": "3788e788-ccd6-432f-a268-5c1b01792364",
"created_at": "2026-08-06 04:12:15.232109+00:00",
"db_md5": "fdaebb2b2c80fed1181f4aab37c4de01",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "b122c07b-154d-47f8-9adc-85a1895bc998",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 7.455757141113281,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333656191826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.018708229064941,
"example_count": 100,... |
378ea852-717f-49b0-bf82-0f658542d436 | easy | gauravmishra | 2026-08-10 08:35:31.726464+00:00 | succeeded | 52f0d9ef249477fd116bdcba2b28aebc384bc94f13cd51b4743241b967d3f50d | 6,007 | null | """E53: K13-S385 with the isolated AdamW epsilon 1e-6 intervention."""
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": "378ea852-717f-49b0-bf82-0f658542d436",
"created_at": "2026-08-10 08:35:31.726464+00:00",
"db_md5": "79bd0340a33249904e5bb974dfa38cf8",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "494a2a37-491c-46dc-a4b5-32b0138e5b8e",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.3170669996448092,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000024835269
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3535596255229727,
"example_count": 60... |
379b36da-9bb3-4b29-9d96-8a5cc2ad149c | easy | lpbb | 2026-08-13 12:10:08.855313+00:00 | succeeded | d4bbda1823770effda6e9462f59bb265f43c7c5188490fa4bbad0cd261760b75 | 4,774 | null | """S1 — one tied cell looped 12*T times with prompt recall, plain final CE loss."""
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,
)
CO... | {
"id": "379b36da-9bb3-4b29-9d96-8a5cc2ad149c",
"created_at": "2026-08-13 12:10:08.855313+00:00",
"db_md5": "95fc503c5fbb0348b8b8454d2732773c",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "8465519b-ef8f-47a3-b6ea-cef72ba0c8ed",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 4.284102807094593,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000060535967
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.358159432754228,
"example_count": 800,... |
379dff7a-accd-4f5d-9e54-1966f8f422bc | easy | AFeinfield | 2026-08-22 14:49:51.736747+00:00 | succeeded | 13b18126cac1d39e27cb85c362a5e538b57a3ed7e5fecdb72d0dc4430e26df88 | 19,306 | null | """092: INSTRUMENTED BASE — two measurement fixes, no architecture change.
(1) Step-indexed LR schedule with a fixed screening step budget. The old
schedule decayed on WALL CLOCK, so throughput jitter changed the
LR-per-step trajectory: identical files on e2 gave 1470 vs 1025 steps
and train acc 0.725 vs 0... | {
"id": "379dff7a-accd-4f5d-9e54-1966f8f422bc",
"created_at": "2026-08-22 14:49:51.736747+00:00",
"db_md5": "fbfde8069d90ebb1b53792fca27f8bb0",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "1814c7f2-b269-44c2-a6e6-f48f241358eb",
"tier": "easy",
"dataset_id": "e10",
"status": "suc... | {
"score": {
"mean_loss": 1.9649362564086914,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.015575758181512356
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9601881504058838,
"example_count": 12... |
37aec27f-691a-43c6-ae6f-e69f59e6ea7b | easy | chuk-uzowihe | 2026-08-07 03:04:13.986411+00:00 | failed | 6171863c31738d98ccbc10e18d69a3ca31692db8e4621c949e4a8abbef18ee83 | 19,944 | null | """One Layer Deeper submission: FPRM over the raw token sequence.
v5: the model is a faithful FPRM (arXiv:2606.18206) applied directly to the
benchmark's token stream — no parsing, no digit lattice, no Fourier
featurization, no special slots. The point is to isolate the fixed-point
recurrence from every tokenization d... | {
"id": "37aec27f-691a-43c6-ae6f-e69f59e6ea7b",
"created_at": "2026-08-07 03:04:13.986411+00:00",
"db_md5": "bc88e1b914c4a4712a7159926e911e7e",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "11993349-5891-4454-ba4a-5b980616085f",
"tier": "easy",
"dataset_id": "e4",
"status": "... | null
|
37aff070-38f7-44a5-b47c-d963a09ff96b | easy | Dandandan | 2026-08-04 19:43:43.281039+00:00 | succeeded | 1e81e1cc4f0c09d5f84ce8772f602dba479d59dd2a26ef27f7d786b486749de8 | 17,510 | 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": "37aff070-38f7-44a5-b47c-d963a09ff96b",
"created_at": "2026-08-04 19:43:43.281039+00:00",
"db_md5": "85023805542093dbd2e9a12ce504d6da",
"submitter": "Daniël Heres",
"github_login": "Dandandan",
"run_id": "224d7c83-9c26-44fa-99da-7155eab70e32",
"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,
... |
37b41418-c42e-4558-a8aa-07b73ac7bac9 | easy | yashkant | 2026-08-15 14:17:24.519995+00:00 | succeeded | e8d88efbd590b58e76d3bacf8a7754be1d747983f0b8575f374f75b97cf0e929 | 32,226 | null | """Round1232 parsed-T allocation crossover on the y/z MinMax transition.
The complete Round1220 answer/latent MinMax candidate remains unchanged except
for the public-T-to-macro allocation. Separate answer y and latent z states,
immutable x reinjection, both MinMax updates per macro, and full BPTT remain
intact. MINMA... | {
"id": "37b41418-c42e-4558-a8aa-07b73ac7bac9",
"created_at": "2026-08-15 14:17:24.519995+00:00",
"db_md5": "85d07aed63e81eef73437429356d762a",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "04ffbd85-dba2-4bf5-8663-fd07850b7d00",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 4.742131471633911,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06499999761581421
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.357089996337891,
"example_count": 100,
... |
37b97250-ba34-4069-9343-cd03cc0f9dcb | easy | chuk-uzowihe | 2026-08-06 18:02:48.088289+00:00 | succeeded | 70fe494e3136a21e2cae54cfd85d1d9ca40855bc598c38dfb8f8358ceea25ff8 | 25,059 | null | """One Layer Deeper submission: FPRM over a compressed (N, x, T) lattice.
Structure
---------
1. The digit-token prompt is parsed (exactly, in int64, on GPU) into the three
integers N, x, T. Each becomes one latent vector via a fixed Fourier
featurization (phases of v mod P over a bank of periods P) followed by ... | {
"id": "37b97250-ba34-4069-9343-cd03cc0f9dcb",
"created_at": "2026-08-06 18:02:48.088289+00:00",
"db_md5": "1abbb11794f1bd2d181dd2a9c29a9eb2",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "24611fdd-87ca-463e-9c3c-75a3c5929c89",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 35.647592544555664,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.019999999552965164
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 37.68484878540039,
"example_count": 100... |
37bbc396-2cc2-41ce-b701-1e34ca145514 | easy | karanganesan | 2026-08-06 04:01:24.495961+00:00 | succeeded | b4c8fcdbe2b197d8dbf439627c2570634819a36ac053b71155b309520dd6c18b | 23,810 | 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": "37bbc396-2cc2-41ce-b701-1e34ca145514",
"created_at": "2026-08-06 04:01:24.495961+00:00",
"db_md5": "6a30d4ff8c7e10cb9e1b8264681cc60a",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "0b5e8768-c112-475d-a36b-74d45c1dfed0",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 11.070435047149658,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02666666731238365
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 11.859073638916016,
"example_count": 100... |
37c9e4fd-3d4e-4787-b471-cb6c6b3eebc8 | easy | jamaliki | 2026-08-06 13:53:21.600382+00:00 | succeeded | 74d866b789e41005b7edf7ba778c9655c660511c4a1e3aadf9cbdd587d424868 | 74,953 | null | """One Layer Deeper submission: clocked latent loop with FSQ restoration.
Design: an encoder reads the input once into a memory. A weight-tied core is
applied m times per macro-step over [register | scratch] slots, cross-attending
to the memory (input re-injection). At each clock edge, registers are snapped
through fi... | {
"id": "37c9e4fd-3d4e-4787-b471-cb6c6b3eebc8",
"created_at": "2026-08-06 13:53:21.600382+00:00",
"db_md5": "5192d3456d9a3ec585452efbe837f124",
"submitter": "Kiarash Jamali",
"github_login": "jamaliki",
"run_id": "b500779a-7c11-4b52-81e5-c3d994b6e87a",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 2.175204277038574,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1735498905181885,
"example_count": 100,
... |
37cae538-90b5-466b-91b7-733fef9d6c3e | easy | liam-gb | 2026-08-16 16:38:17.819924+00:00 | succeeded | ee8e3602e45d4a044761dc84e5d6b53ee9f63b01800876d8c48a95b4ec2aca98 | 15,754 | null | """rns_pm_gather_ct_widet_ptnore: widet + per-prime temperature (16-prime bank).
The wide readout is what makes long answers reachable; the trim is what pays
for it, by not running loop iterations whose update is multiplied by zero.
Base docstring:
Single change vs rns_pm_gather_ct: the readout carries one
distance-... | {
"id": "37cae538-90b5-466b-91b7-733fef9d6c3e",
"created_at": "2026-08-16 16:38:17.819924+00:00",
"db_md5": "5051d831398b6c8922c27ddf8e5e8b36",
"submitter": "liam-gb",
"github_login": "liam-gb",
"run_id": "91779f57-9b42-4e1e-887e-5cbe1dbeff9d",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded"... | {
"score": {
"mean_loss": 5.109532713890076,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.4145833421498537
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.028779983520508,
"example_count": 300,
... |
37cbdfb1-c19c-4f7c-9361-1ff603ba951a | easy | gauravmishra | 2026-08-10 19:08:53.748755+00:00 | succeeded | 00804ed05da14753413f6860f1131dfd23daa2c4aece25d8823491ffe14fd63a | 14,519 | null | """Capability reset: end-relative output queries with optional recurrent tape.
The only line changed between hosted Easy variants is ``VARIANT``. Every
variant is generic: it uses only input embeddings, positions, and the supplied
attention mask. It does not inspect token IDs, parse fields, or implement an
arithmeti... | {
"id": "37cbdfb1-c19c-4f7c-9361-1ff603ba951a",
"created_at": "2026-08-10 19:08:53.748755+00:00",
"db_md5": "de7b46321cbcf8fcc49b77012f7d9c67",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "8c5e2c64-0024-42ff-9f74-91ab58f58983",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.274579644203186,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3247427940368652,
"example_count": 100,
... |
37cd591b-4797-46f2-80c1-ee18a81d39fc | easy | chuk-uzowihe | 2026-08-09 20:53:33.195389+00:00 | succeeded | 11727f44fcbaa1771c4778943298f1d3f89c2ab1df46b7cebb9454bdbbec56a0 | 38,839 | null | """One Layer Deeper submission: depth-routed MoEUT over a 2D digit lattice.
v9: the recurrent block is a single MoEUT layer (refs/moeut) — SwitchHead
attention + sigma-MoE feedforward — whose routers are conditioned on the
loop depth (Fourier features of the iteration index) so one shared bank can
express phase-depend... | {
"id": "37cd591b-4797-46f2-80c1-ee18a81d39fc",
"created_at": "2026-08-09 20:53:33.195389+00:00",
"db_md5": "0e9edcee3d3fe5f89af969bcd7029d15",
"submitter": "Chuk Uzowihe",
"github_login": "chuk-uzowihe",
"run_id": "760b0233-7f46-45e8-8585-add276ce13b5",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 6.771840333938599,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0333333327434957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.951433181762695,
"example_count": 100,
... |
37cd87a6-ba58-4aa0-9fca-ec301a4a5957 | easy | alirezashirvani-jr | 2026-08-30 01:32:46.573092+00:00 | succeeded | 08922b04fa9afb0f2c135f923d17dc7c8a0e8e17eccd9c3a7718c9e6346d986a | 27,729 | 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": "37cd87a6-ba58-4aa0-9fca-ec301a4a5957",
"created_at": "2026-08-30 01:32:46.573092+00:00",
"db_md5": "63f8b53c4a7bd0c4ff3046b5c41c5ca8",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "2258fead-54eb-485e-a5b3-4239d2f29e0d",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 2.709320474516701,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014583333289871613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7208542823791504,
"example_count": 600... |
37cfc403-0481-415c-bca7-af2ce085bbb8 | easy | khushidahi | 2026-08-15 13:18:08.612027+00:00 | succeeded | 9e1317b6bbbd30b2d6c6fef6c31b8094d35723891a39db9cb2600333a7b3168c | 25,301 | 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": "37cfc403-0481-415c-bca7-af2ce085bbb8",
"created_at": "2026-08-15 13:18:08.612027+00:00",
"db_md5": "1ca5a56e71940c4559148c1df11dff2e",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "141cd86c-21c5-4c9c-9c7b-84fc0798b129",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1539098024368286,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011249999981373549
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.155963659286499,
"example_count": 600... |
37d9604b-99f9-4de9-b1b3-93501b0bc875 | easy | yashkant | 2026-08-13 00:00:40.858417+00:00 | failed | e241cbdfa9bd42fe96268beb30b87c9f88311fff1ac5094f7f9710a2a727e55e | 7,778 | null | """Round 783 compact bigram hash with AdamW LR 3.5e-3 and EMA 0.997.
This arm uses 63 shared pair buckets plus one boundary bucket. Its dimension
is chosen so table plus projection has exactly the same parameter count as the
collision-free and current-token arms. The table starts at zero and the
projection starts ra... | {
"id": "37d9604b-99f9-4de9-b1b3-93501b0bc875",
"created_at": "2026-08-13 00:00:40.858417+00:00",
"db_md5": "138c6376285fd7917246a05d3a373cd3",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "09fc1495-a405-42b5-b08b-47bb9a0a50bd",
"tier": "easy",
"dataset_id": "e2",
"status": "failed"... | null
|
37dae994-2ef9-489f-be41-4c16dd28134a | easy | achyutreddy24 | 2026-08-04 23:11:33.023855+00:00 | succeeded | 31702b5427a696450c0bd246cafbbd8b1ed98ab0cd05ba1c8c65b8b5644a750d | 14,919 | null | """Fuel-gated recurrent residue clock (informed initialization) — public submission.
WHAT THIS MODEL IS
==================
The task is x^(2^T) mod N from decimal digit tokens in a single parallel forward
pass. This submission computes it the way the algebra suggests: as T composed
applications of one exact squaring s... | {
"id": "37dae994-2ef9-489f-be41-4c16dd28134a",
"created_at": "2026-08-04 23:11:33.023855+00:00",
"db_md5": "d43b1e9d5c00d52f53ba68a424bdca4c",
"submitter": "Achyut Reddy",
"github_login": "achyutreddy24",
"run_id": "81085e9a-7299-4f1b-b369-f61e90bb72b3",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 0.7277971506118774,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.7277970910072327,
"example_count": 100,
"exa... |
37de3376-e0d4-437b-a2de-20d8c8449f33 | easy | KaustubhKumar05 | 2026-08-25 18:22:03.998551+00:00 | succeeded | e61cc7591ec83d725aef1ff1fdd4693c311bba37127cc76ba9b0c3df8f54b631 | 16,286 | null | """conv_dist: conv_v2s_mom05 + a circular VALUE-DISTANCE loss.
Digit cross-entropy counts wrong symbols, which is anti-correlated with
numeric correctness on the commonest arithmetic error. Measured: for true
answer 120, CE scores 119 (off by ONE) at 8.00 and 220 (off by ONE HUNDRED)
at 4.00, because the borrow flips ... | {
"id": "37de3376-e0d4-437b-a2de-20d8c8449f33",
"created_at": "2026-08-25 18:22:03.998551+00:00",
"db_md5": "fd430a49750f99034e68fb5a8169b1f4",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "3eb3c8ed-8d3b-4fc2-accf-dd6a8768f69c",
"tier": "easy",
"dataset_id": "e8",
"status": "succee... | {
"score": {
"mean_loss": 7.2537455558776855,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.1303921602666378
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.949556350708008,
"example_count": 85,
... |
37e7a93c-8602-40af-919b-b693ec0bbefc | easy | gauravmishra | 2026-08-09 22:00:33.858252+00:00 | succeeded | b0979291365095e306a4f792207dd9c45c6c4b4543405555f7cfab816652b42f | 4,249 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
D_MODEL = ... | {
"id": "37e7a93c-8602-40af-919b-b693ec0bbefc",
"created_at": "2026-08-09 22:00:33.858252+00:00",
"db_md5": "ff141e9d63f34f2c2e34713a993447aa",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "b0fe82fc-f3ac-40a1-b0a3-c0159e6c50d1",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 3.355435201315863,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.618710500777035,
"example_count": 600,... |
37ea55f0-cecb-4fe1-98e4-84e11fba9f42 | easy | andreamiele | 2026-08-06 09:12:33.481507+00:00 | succeeded | 28db2f3f862aa09c0f7525cf892c8fc386629ecaa8188916a57fe14491f088cb | 22,805 | null | """Loop model with recall, composition consistency, and deep supervision.
Motivated by what the easy-to-hard / latent-reasoning literature converges on
and which earlier attempts here lacked:
* **Recall (input injection).** Deep Thinking (Schwarzschild et al. 2021;
Bansal et al. 2022) found iterating past the trai... | {
"id": "37ea55f0-cecb-4fe1-98e4-84e11fba9f42",
"created_at": "2026-08-06 09:12:33.481507+00:00",
"db_md5": "aae991192217af5eca1017a91a6737dd",
"submitter": "Andrea Miele",
"github_login": "andreamiele",
"run_id": "133fc668-16ac-40df-9116-f9d7ae8ed010",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 1.334719720042924,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.7533333377043405
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.42353919559895103,
"example_count": 100,... |
380317c9-1986-4d50-9141-f61440f603a1 | easy | ehonig | 2026-08-20 19:05:53.801906+00:00 | succeeded | 655a110ea3902f200de6da6f509ee2bb06671705b58097ddc1a285037e394269 | 25,540 | 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": "380317c9-1986-4d50-9141-f61440f603a1",
"created_at": "2026-08-20 19:05:53.801906+00:00",
"db_md5": "f07c8dad1aafd172017084a015a40a2b",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "911463d4-6ac6-4107-ac11-2004b04874fa",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.1075154542922974,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07999999821186066
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0929758548736572,
"example_count": 100... |
38056b51-d614-4422-963a-f7e520f5a068 | easy | AFeinfield | 2026-08-28 02:20:37.878851+00:00 | succeeded | cae587efadbcb666d1204659aafaaa07ad7abfab3f36ae118af1bf26d0609bbd | 7,797 | null | """179: GROKKED RESIDUE MAP — the literature bet for fixed-N datasets.
Grokking modular arithmetic to 100% held-out accuracy is established
(Power et al. 2022; Gromov 2023 for MLPs; GrokFast for 50x acceleration).
Squaring mod a FIXED N is in that task class, so for fixed-N datasets we
learn the entire one-step map x ... | {
"id": "38056b51-d614-4422-963a-f7e520f5a068",
"created_at": "2026-08-28 02:20:37.878851+00:00",
"db_md5": "a4039c23ea39a76d70bb8d4a19a1a2c7",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "2c01f926-61b0-48a8-b7c8-554fad17360d",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.9870691895484924,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07999999821186066
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.943123459815979,
"example_count": 100,... |
380edfeb-d964-41bf-982b-96103e1fa11a | easy | sapient-sapiens | 2026-08-09 14:54:32.681007+00:00 | succeeded | 1763607be036fe83d25121757c497c2d5b7d375786d04e914da63b7224eb45c8 | 11,416 | null | """Near-500M adaptive recurrence grown from the hosted-working baseline."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
D_MODEL = 6400
NUM_HE... | {
"id": "380edfeb-d964-41bf-982b-96103e1fa11a",
"created_at": "2026-08-09 14:54:32.681007+00:00",
"db_md5": "7975d62f91a0f4a130ee0cdd08af97db",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "f5a264b9-b3b8-4732-b0b7-b11cfd0aceb4",
"tier": "easy",
"dataset_id": "e4",
"status": "su... | {
"score": {
"mean_loss": 2.364686550054137,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.356496607716783,
"example_count": 1200,
"exac... |
381441bf-0a02-41a1-a87e-cfdbf9b04243 | easy | DDanlov | 2026-08-12 21:02:34.314231+00:00 | succeeded | 0286a59512605c7ac56e5e79cc82d56962200aedefda8053c0517b29c7419562 | 25,388 | null | from __future__ import annotations
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
try:
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
except ... | {
"id": "381441bf-0a02-41a1-a87e-cfdbf9b04243",
"created_at": "2026-08-12 21:02:34.314231+00:00",
"db_md5": "181d48a2f4d63e21d76d94d1779a6fcb",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "d1d3212c-4cb4-4efa-b718-0ac8c37d709a",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.1639939546585083,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.167818784713745,
"example_count": 100,
... |
381a45b0-a689-4c7e-b170-4e9c5becab0c | easy | DDanlov | 2026-08-29 13:42:37.492097+00:00 | succeeded | 55cb2e69fc453cd549ea55dcd3aef85862518cfc98ce3b8d639aae7d3c8466a9 | 24,019 | null | """
trial_w03_l2_d1024_k8.py
W03: L=2, D=1024, H=16, d_ffn=3072, K=8, dt=0.08, mu=0.85, lr=0.003, freq=5
"""
import math
import time
import contextlib
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
try:
... | {
"id": "381a45b0-a689-4c7e-b170-4e9c5becab0c",
"created_at": "2026-08-29 13:42:37.492097+00:00",
"db_md5": "bd27dfa6348fbbe69b098a816b49d146",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "90d94374-4893-4ad4-b27f-d9b827049cf7",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.2951771114890667,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012916666703919569
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3296526657091663,
"example_count": 60... |
381fd1a6-6c3b-4c55-b9e1-58de067eb6aa | easy | jordanrubin | 2026-08-02 20:44:54.141012+00:00 | succeeded | fc83380fd71a375ea72ea5c0f6a095229d7241bed721084eff1844b20f9e70c0 | 13,994 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "381fd1a6-6c3b-4c55-b9e1-58de067eb6aa",
"created_at": "2026-08-02 20:44:54.141012+00:00",
"db_md5": "8424ff3c2b53438cdfeb088f9343f339",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "ca5dfab7-bece-4fb1-a3a3-04e827a5b092",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 3.4063422044311347,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.24020832777023315
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.4877349973140275,
"example_count": 300... |
3821ec48-d2b7-4cbc-bfbd-11e4187c264f | easy | velocizapkar | 2026-08-16 07:24:38.264723+00:00 | succeeded | 31d080bce4338d95b2551b331f3145f136937a0458c540763e3ee7542f81cd5a | 10,468 | null | """Tied standard residual Transformer with a 1-to-4 depth curriculum.
This is the matched fixed tied-block control with one training-only change:
the shared block begins at one application and advances to four as a shaped
function of evaluator-provided wall time. Evaluation always executes four
applications. Paramet... | {
"id": "3821ec48-d2b7-4cbc-bfbd-11e4187c264f",
"created_at": "2026-08-16 07:24:38.264723+00:00",
"db_md5": "3c91b549c8149953dcbf05c73709070e",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "74ec91e4-e619-4e12-b7d0-468fc965ee14",
"tier": "easy",
"dataset_id": "e6",
"statu... | {
"score": {
"mean_loss": 6.917160749435425,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02837838069535792
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.514925479888916,
"example_count": 60,
... |
38277bd8-d746-47cb-8ca3-addb6d4836db | easy | jordanrubin | 2026-08-01 21:20:10.030311+00:00 | succeeded | 2c6adbc91db77091eaa1fdb53b2b15fdc6e05306608a2440be5e6a2aa2591b8a | 18,183 | null | """Looped bidirectional selective-SSM (Mamba-style), pure PyTorch.
The statefulness bet: input-dependent SSM dynamics along the token axis,
with the block weight-tied and applied NUM_LOOPS times in depth with input
injection — recurrence in depth is where the task's T-iteration lives.
The selective scan is computed i... | {
"id": "38277bd8-d746-47cb-8ca3-addb6d4836db",
"created_at": "2026-08-01 21:20:10.030311+00:00",
"db_md5": "35fec302dbad6ab1de27b5092ede0b6a",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "1344267a-5938-4dc8-b8b1-3307729ba3c8",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 5.273348143570441,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03999999910593033
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.608862400054932,
"example_count": 100,
... |
3833f72f-7296-40ac-ad21-140ea5af7b9d | easy | oupadhyay | 2026-08-15 00:30:44.182166+00:00 | failed | 5eee2faa16c7d3cf3e46236b1bd0b7196d71fb9825564bca443c3c469f0ed8be | 3,983 | null | """Global contextual N-conditioned transducer with more optimizer updates."""
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... | {
"id": "3833f72f-7296-40ac-ad21-140ea5af7b9d",
"created_at": "2026-08-15 00:30:44.182166+00:00",
"db_md5": "62eac8600ef7e659d6318b7050a56d28",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "13491f4c-2389-4d1a-b85b-7bdb20dba6df",
"tier": "easy",
"dataset_id": "e5",
"status": "f... | null
|
38378e54-df21-4677-8cf9-a57985956ca5 | easy | sapient-sapiens | 2026-08-31 22:18:10.631721+00:00 | succeeded | 21fca420afffc8ea5f0405d180bb1ade8a684636b43df60c99863abfcd9878f5 | 16,043 | null | """One-block nested E5 model: four learned Horner calls per square, eight squares."""
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_mod... | {
"id": "38378e54-df21-4677-8cf9-a57985956ca5",
"created_at": "2026-08-31 22:18:10.631721+00:00",
"db_md5": "fc19b475557d4a536dcbfc80e24263b1",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "368b5291-7a22-4c08-b168-a21659ff77fc",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1513451319242813,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333333333333
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1514078875828218,
"example_count": 60... |
3845f008-95bd-402e-8b4b-4d3d293ebf09 | easy | DDanlov | 2026-08-31 20:19:57.790587+00:00 | succeeded | ca3bad44d796c2fc5906a8bec3ffb0adcbdda6f349daf8b19e4d760689d72560 | 32,407 | null | """
Unified Continuous Hopfield Equilibrium Propagation Deliberation Engine
=======================================================================
Architecture:
- Deliberation: Continuous Hopfield Multi-Layer Energy System (L=4, D=768, d_ffn=2304, H=16)
- Integration: In-Place Strang-Split Conformal Symplectic Velocit... | {
"id": "3845f008-95bd-402e-8b4b-4d3d293ebf09",
"created_at": "2026-08-31 20:19:57.790587+00:00",
"db_md5": "4bc1c00f59046c3ed4dec9b94bf519db",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "f1e33b87-8d4e-47d5-bc99-f3b2e452bcdb",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.7458473638875636,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011875000049670537
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7665768481020865,
"example_count": 30... |
384781e7-8fcf-48ac-9e64-83ea7fc1707b | easy | lpbb | 2026-08-21 13:05:53.027156+00:00 | succeeded | 5ff6d2e5d9b82e6af9060440b4065f1bb06f33a16b620db54b53b397818c1aac | 12,545 | null | """S0 + RoPE + abacus + looped block, with *input-adaptive* depth: instead of
a fixed `num_repeats`, each sample's T is parsed from its own token stream
(the digits after the `T` marker) and the shared block of `num_layers` layers
is applied exactly T times for that sample. One block pass = one modular
squaring step, s... | {
"id": "384781e7-8fcf-48ac-9e64-83ea7fc1707b",
"created_at": "2026-08-21 13:05:53.027156+00:00",
"db_md5": "cf9b3e5968ebefb80997467aa41c1b3a",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "b87628c2-0d02-4eba-a243-5b62305b08d8",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 2.670956392835203,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06499999826153119
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.894313097000122,
"example_count": 100,
... |
384bda3d-43c9-4879-867a-1472ee4d8161 | easy | khushidahi | 2026-08-08 02:44:07.580930+00:00 | succeeded | 1cdd583dffb7313ea6793e47b0fded6b50ed9c296f9f79e68b67cd668f6c66d2 | 15,593 | null | """R42 Fourier-number representation experiment.
All input-dependent numeric parsing stays inside differentiable GPU tensor
operations. This is a learned-model representation, not a task solver: it does
not compute modular squaring or create intermediate labels.
"""
from __future__ import annotations
import math
impor... | {
"id": "384bda3d-43c9-4879-867a-1472ee4d8161",
"created_at": "2026-08-08 02:44:07.580930+00:00",
"db_md5": "1ca4113577061d36fb195591ba7394dd",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "580f42b1-df51-46b0-8689-361bb8abb372",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.164495187970969,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004583333353511989
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.168682813644409,
"example_count": 600,... |
384c4f0d-583b-42c5-b6b6-06033961ec44 | easy | jordanrubin | 2026-08-06 19:57:50.020438+00:00 | succeeded | 09586f90ffa296e3bdbb25e6a0b059c7579344900d2f7d9eb12b0e902c4848f6 | 15,390 | null | """Bidirectional transformer, D=256 x 3 blocks, grokking-tuned AdamW.
Design notes:
- Inputs are padded to config.max_seq_len inside forward and logits sliced
back, so every training batch presents one static shape (compile/cudagraph
friendly, and uniform kernel shapes even in eager).
- Attention mask gets an iden... | {
"id": "384c4f0d-583b-42c5-b6b6-06033961ec44",
"created_at": "2026-08-06 19:57:50.020438+00:00",
"db_md5": "ef7d67d872a65aa96e81ca53991ab27d",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "ff2207a7-bf06-4492-96d4-a326c288aaa9",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 3.2098218499361426,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.18395832894990843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.876075123799111,
"example_count": 300,... |
384d3388-4e12-460f-b992-586a3d8141a5 | easy | lpbb | 2026-08-07 14:04:47.629917+00:00 | succeeded | 29788247c614406df0ad36a5b74283838128caad6352515b4869dc4b548f6c10 | 21,114 | null | """Exhaustive lever map: every knob the competition API exposes, spelled out.
Same model family as submissions/dev/submission.py but with two fixes wired
in (everything else left at its neutral default), and every lever written out
explicitly with its constraints, so this file doubles as the reference for
"what am I a... | {
"id": "384d3388-4e12-460f-b992-586a3d8141a5",
"created_at": "2026-08-07 14:04:47.629917+00:00",
"db_md5": "7995c515662faaea5e4feeb52b883a96",
"submitter": "Lpbb",
"github_login": "lpbb",
"run_id": "abed0197-5d20-4141-be7b-a61c368a4fbb",
"tier": "easy",
"dataset_id": "e3",
"status": "succeeded",
"s... | {
"score": {
"mean_loss": 4.637277874640176,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000004656612
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.131471155669742,
"example_count": 800,... |
3852e4ba-4789-4952-bebf-d3444c381b4b | easy | gauravmishra | 2026-08-08 00:08:21.062785+00:00 | succeeded | 9140fef522ec6a9afeb558e69343f534c498c0164ab7648853f9bbf90a742043 | 3,726 | null | """Four-pass Transformer with four independently parameterized blocks."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 1... | {
"id": "3852e4ba-4789-4952-bebf-d3444c381b4b",
"created_at": "2026-08-08 00:08:21.062785+00:00",
"db_md5": "067dd19f95b3a2b025d36561c727f6b0",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "8cba9c35-0fb4-46e3-a10a-d77a7e71ea2a",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 11.438741544039665,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 11.841605083824808,
"example_count": 60... |
3855e1eb-e9ac-419d-ac72-573ee02e0db1 | easy | amirmeisami | 2026-08-15 01:10:59.463180+00:00 | failed | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "3855e1eb-e9ac-419d-ac72-573ee02e0db1",
"created_at": "2026-08-15 01:10:59.463180+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Amir Meisami",
"github_login": "amirmeisami",
"run_id": "2d6d444a-1e61-4459-be91-f9e123c1925e",
"tier": "easy",
"dataset_id": "e5",
"status": "f... | null
|
3858d371-19a6-4a45-9253-662876519591 | easy | sapient-sapiens | 2026-08-30 22:39:56.497398+00:00 | succeeded | 69641ee9384602c14c2d68aa1edfdceb232428b31f02fbea3bd7eda774c84bb3 | 17,862 | null | """Eligible E5 direct-Horner model with a one-time ingress conditioner.
After prompt encoding and before any outer square, X receives one scale-0.03
SwiGLU residual. The conditioner is never part of the repeated square/Horner
operator and is never reapplied. N/X are otherwise immutable, only A is
decoded, T acts only ... | {
"id": "3858d371-19a6-4a45-9253-662876519591",
"created_at": "2026-08-30 22:39:56.497398+00:00",
"db_md5": "021ec9c6c0307342d7f2c2657d3bb867",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "297d6a82-6a3b-4e47-aea3-a986e9657922",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.2105869512448324,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2102088981679735,
"example_count": 60... |
3860faa2-c2d9-476e-a1bb-b120de5cc3b5 | easy | CodeReclaimers | 2026-08-15 05:30:44.817945+00:00 | succeeded | acaae3e694638e70e9600d2f2a9476aaee199d3cdd4ad6d8df1a49005e3dd588 | 57,228 | null | """v31: merge of the prime-agent low-rank universal-CRT design (hosted M4:
100%, Max T=64, OOD-N Max T=64) with the v30 streamed large-N readout.
From the low-rank design: per-prime residue embeddings E (sum m, DIM) with a
shared readout w (DIM, WMAX) — per-prime cost m*DIM, so the bank covers all
primes 2..4095 (~17M... | {
"id": "3860faa2-c2d9-476e-a1bb-b120de5cc3b5",
"created_at": "2026-08-15 05:30:44.817945+00:00",
"db_md5": "fc08ccee7087a70aeefaaa03c48367d2",
"submitter": "CodeReclaimers",
"github_login": "CodeReclaimers",
"run_id": "bf20c974-f469-4ff9-a5f6-95029108488a",
"tier": "easy",
"dataset_id": "e9",
"status... | {
"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": 90,
"exact_accuracy": 1.0,
"c... |
386194f5-4d9d-43a0-a053-8f6dba063ea1 | easy | jordanrubin | 2026-08-14 01:10:48.701459+00:00 | succeeded | fc98447fc19a656845f48580dd88f53c0b2d7d00c35e32a8fe2d3dfff79a73f4 | 29,581 | null | """Weight-tied MLP-cell loop over soft digit states (mlploop).
Cell = MLP over [expected digits, pairwise digit products, RNS residue
simplices (fixed differentiable mixing over Z_p), N digits]. The cell that
learns one-step modular squaring from direct pairs (52% unseen at 14k rows,
day-20 screen) inside the exact-T ... | {
"id": "386194f5-4d9d-43a0-a053-8f6dba063ea1",
"created_at": "2026-08-14 01:10:48.701459+00:00",
"db_md5": "d08ba312cd6460ffc449c5d2476626a5",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "9f9fa252-c25d-4479-8ef6-bf257f6f4c94",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.750798353104604,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.019166666160648067
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.194704412745788,
"example_count": 600,... |
3862e6f5-439f-4987-88cd-8980dd65cda4 | easy | khushidahi | 2026-08-22 01:30:07.408884+00:00 | failed | 97718428f263a95e6c61d102df8e4da30bdc6aed0cd4a525b4b51f49fb6a0c75 | 18,866 | null | """R395: minimal Universal-Transformer functional-root control for M5.
This candidate adapts the released Universal Transformer to the competition's
separate-output interface. Numeric prompt symbols are only rearranged onto
equal-length, right-aligned categorical N and X tapes. A third tape supplies
parallel learned... | {
"id": "3862e6f5-439f-4987-88cd-8980dd65cda4",
"created_at": "2026-08-22 01:30:07.408884+00:00",
"db_md5": "979d4cb136f1b64486c013ccacad420e",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "6165d21f-80ac-4c66-a9b5-e9ceb358563e",
"tier": "easy",
"dataset_id": "e5",
"status": "fail... | null
|
386a15e4-b1a7-4e6c-88ce-b17d8a300491 | easy | DDanlov | 2026-08-15 15:04:32.800786+00:00 | succeeded | a8af64ec94282ec36ca1c650edf0053ecf113264577ba3e68d997cb3edb94c99 | 19,671 | 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": "386a15e4-b1a7-4e6c-88ce-b17d8a300491",
"created_at": "2026-08-15 15:04:32.800786+00:00",
"db_md5": "2d9417d66e904d08d6f16bef24839798",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "b4892c7e-9291-4736-89dd-1c5c980eb7d7",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 6.299631243423484,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.026666666766007742
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.260930705426344,
"example_count": 100,... |
386bd022-c32b-486c-8f2d-52c8fb9370ac | easy | ayushnangia | 2026-08-11 18:13:16.428333+00:00 | succeeded | d3eac746a15be85cc616fba2d3717845ae6dbf91dd0aaa1d26175c8edd197374 | 43,330 | null | """rMXL1 (rMX1 + PARAMETER-FREE i+j LATTICE ROUTING) — organizer-blessed:
inject the convolution skeleton of multiplication into every loop step.
Per step: s_i = proj(tape_i) (72 scalars), p_k = sum_{i+j=k} s_i s_j
(parameter-free routing, k folded to slot range), tape_k += scale * phi(p_k).
The pairing structure of a ... | {
"id": "386bd022-c32b-486c-8f2d-52c8fb9370ac",
"created_at": "2026-08-11 18:13:16.428333+00:00",
"db_md5": "38039b6d9d305af6d9698107d5c9ab1c",
"submitter": "Ayush Nangia",
"github_login": "ayushnangia",
"run_id": "9d4c29b1-28f3-4b52-be81-04f703f8437e",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.315508246421814,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.320657730102539,
"example_count": 600,... |
38802924-bd0a-40da-8ade-95e1810545d2 | easy | sapient-sapiens | 2026-08-23 07:42:28.483825+00:00 | succeeded | 08da1057457d3739e91e2754695b7a5eb6727cd4c2c9a7ba0ebc49e8a5d1bc8b | 19,909 | 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": "38802924-bd0a-40da-8ade-95e1810545d2",
"created_at": "2026-08-23 07:42:28.483825+00:00",
"db_md5": "f5e2646c2afeaa677593d7b7d83396eb",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "58608edf-0884-4451-a399-432c55431573",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.1500543687425333,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0037500000000000003
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1530903175540033,
"example_count": 6... |
3885c6f8-1b44-4a45-833e-f64c60576663 | easy | briheet | 2026-08-07 02:14:20.223487+00:00 | succeeded | 65088e164e87f0a33b39a4046db181879fc74ea822ed89de572613e03a4924bd | 3,864 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 64
NUM_HEADS = 4
c... | {
"id": "3885c6f8-1b44-4a45-833e-f64c60576663",
"created_at": "2026-08-07 02:14:20.223487+00:00",
"db_md5": "8f95a49d509c033e177e3d7bdd4ce373",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "b375503b-977f-4c7a-b209-5cb2286e6630",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.076060712337494,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04833333380520344
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.181432008743286,
"example_count": 100,
... |
38862769-5cd2-4d95-9ae5-d73d01e42df5 | easy | oupadhyay | 2026-08-29 17:40:59.319720+00:00 | succeeded | e9c675c7182b58a8c5c336d13d574624f255efca79c3281e713f992445be17b5 | 3,993 | null | """Conditional optimizer screen: batch128."""
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_si... | {
"id": "38862769-5cd2-4d95-9ae5-d73d01e42df5",
"created_at": "2026-08-29 17:40:59.319720+00:00",
"db_md5": "290d0c8bb1faeccfb53e7b35639151cc",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "459a4231-e8d6-4685-9682-5b89b1883967",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 1.9878603219985962,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.026801803149282932
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.961653232574463,
"example_count": 60,... |
388727d3-0e67-4be8-8bb2-ab8679fa59a0 | easy | DDanlov | 2026-08-18 00:34:52.506952+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": "388727d3-0e67-4be8-8bb2-ab8679fa59a0",
"created_at": "2026-08-18 00:34:52.506952+00:00",
"db_md5": "ef8f2b4e45393e17f33e8d6b42c30e5d",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "7fb2d72b-8bfb-403a-85b9-9e43e420e6af",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.3062232302604366,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3363519421874677,
"example_count": 60... |
388761e7-faab-48b8-b6e8-1351d4a155b2 | easy | khushidahi | 2026-08-09 04:24:02.922491+00:00 | succeeded | 7970c3df3fb605f8ea9ac28e697060e839e5fbc115a4ecd78e978c62c3b585d8 | 14,779 | 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": "388761e7-faab-48b8-b6e8-1351d4a155b2",
"created_at": "2026-08-09 04:24:02.922491+00:00",
"db_md5": "eb9f5886c3b5505aa0b7fc9621c4c577",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "121c0836-ff0a-4f11-88d2-576ad703c22b",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.197014570236206,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666772216558
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.196056365966797,
"example_count": 600,... |
38908314-fa4c-4abd-b6ab-b0ea5bff7304 | easy | AFeinfield | 2026-08-25 01:30:42.490190+00:00 | succeeded | a7e6625237f1b04919592eae91b2358d48dd7d31e13a9013cec5d39299784503 | 10,853 | null | """113: 023 VERBATIM (no init fix) + ABACUS place-value embeddings.
The 110/111/112 ladder never fit e5: the tied-head rescale that is correct
on the 5-head d512 ensemble BREAKS fitting on this single-tied-head d256
trunk (110: loss 2.09 at 2772 steps vs historical 023: train_acc 1.0 at
3968). 113 grafts abacus onto t... | {
"id": "38908314-fa4c-4abd-b6ab-b0ea5bff7304",
"created_at": "2026-08-25 01:30:42.490190+00:00",
"db_md5": "1f41570b1e3c6afe24fc9e0bf039e608",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "d1ed858b-6679-49c0-ad8b-47e7e1c41e68",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.577472194793277,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5420125578551014,
"example_count": 600... |
38934734-9e03-4856-9c17-9fc243f3a2fd | easy | Cree0618 | 2026-08-29 11:30:26.463016+00:00 | succeeded | 6b763dedb799b544d417c561c0873b7fdbb0049eca749e0cfa44d6e9cc385c68 | 14,989 | null | """E019: chain-supervised T-aligned shared squarer.
Parent: E017 (T-aligned recurrence on the E007 untied-head model).
Hypothesis:
E017 failed its gate because the shared block was never asked to carry an
explicit "current value" through the recurrence. Supervision arrived only
after T composed applications, so nothi... | {
"id": "38934734-9e03-4856-9c17-9fc243f3a2fd",
"created_at": "2026-08-29 11:30:26.463016+00:00",
"db_md5": "94d5bdcc12c59844df3ccab75d842ec2",
"submitter": "Cree0618",
"github_login": "Cree0618",
"run_id": "f100797e-b569-4528-8528-8c826ae01cde",
"tier": "easy",
"dataset_id": "e3",
"status": "succeede... | {
"score": {
"mean_loss": 9.480329513549805,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006249999976716936
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.653474807739258,
"example_count": 800,... |
389a12b7-536d-4794-a220-9e77c8f7b1e1 | easy | 0Chris5R | 2026-08-07 11:12:53.380977+00:00 | succeeded | 135886e112857078c6d7efd68958d127602ab3a5f992b217496a4d3369a44d44 | 17,000 | 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": "389a12b7-536d-4794-a220-9e77c8f7b1e1",
"created_at": "2026-08-07 11:12:53.380977+00:00",
"db_md5": "70daa7e69afe3a042da45e093d359c09",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "2bdebe98-a321-4033-80d3-242d2d22a73b",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 7.005897317623334,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.366102231457629,
"example_count": 600,
"e... |
389bdc27-0f9d-4dfd-9c78-28ace3edaefa | easy | jamaliki | 2026-08-03 22:23:02.380840+00:00 | succeeded | 3187c326302e5e2f3a3d6d9be9093ee36f795e5633e35e782129f001346fb66a | 9,236 | null | """One Layer Deeper submission: clocked latent loop with FSQ restoration.
Design: an encoder reads the input once into a memory. A weight-tied core is
applied m times per macro-step over [register | scratch] slots, cross-attending
to the memory (input re-injection). At each clock edge, registers are snapped
through fi... | {
"id": "389bdc27-0f9d-4dfd-9c78-28ace3edaefa",
"created_at": "2026-08-03 22:23:02.380840+00:00",
"db_md5": "c8123a978b869551f215a54af5ea6cbc",
"submitter": "Kiarash Jamali",
"github_login": "jamaliki",
"run_id": "73a87409-4b96-4c54-823d-f6996dfe0817",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 3.1286404132843018,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.019999999552965164
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1518304347991943,
"example_count": 10... |
389ec18b-b875-400b-9c05-db862bd97055 | easy | alirezashirvani-jr | 2026-08-28 23:42:23.590789+00:00 | succeeded | 73dad23ac67ce34a8f5ba4257342fbd955f792c46dd910f2668b969b51df72ec | 53,560 | null | """Adaptive learned discrete recurrent fabric for One Layer Deeper.
A gradient-trained, operation-free recurrent computational machine. The
model learns, from the competition's endpoint supervision: how to parse the
prompt into a bank of categorical digit registers, a reusable transition
circuit applied serially, a l... | {
"id": "389ec18b-b875-400b-9c05-db862bd97055",
"created_at": "2026-08-28 23:42:23.590789+00:00",
"db_md5": "c68d20cdc8ff54eb88edee065db13428",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "62e7cc80-e246-41c1-b29f-c4c56eea5c96",
"tier": "easy",
"dataset_id": "e9",
... | {
"score": {
"mean_loss": 1.8524322509765625,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333333656191826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8630187511444092,
"example_count": 90... |
38b0e5d0-e7e0-4a43-b57b-5f64d757cb78 | easy | khushidahi | 2026-08-23 21:31:29.234035+00:00 | succeeded | dc30f96367eeccb5fe562f1ff3bdff32abfe024a824930c0ad6f787d71aa41f0 | 11,734 | null | """ART-DT recall candidate.
A learned recurrent sequence model with right-to-left Abacus positions and
global-sequence rotary attention. This variant changes only optimizer LR and
warmup schedule from its architectural parent. This branch ties the two
within-visit layer calls to one TransformerLayer object. One imm... | {
"id": "38b0e5d0-e7e0-4a43-b57b-5f64d757cb78",
"created_at": "2026-08-23 21:31:29.234035+00:00",
"db_md5": "25f8e724d7ec7f02d7c56a4ef3401964",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "b115560b-155c-4ff3-902c-507b12e731bc",
"tier": "easy",
"dataset_id": "e9",
"status": "succ... | {
"score": {
"mean_loss": 2.031118929386139,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.030000001192092896
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1404833793640137,
"example_count": 90,... |
38b3c2cf-5a9e-4f6d-bc48-061e0846ce79 | easy | KaustubhKumar05 | 2026-08-28 18:16:13.707432+00:00 | succeeded | 40bc192c44ce30d3edd9b62bd2716a1ca76828f8f5f626c5a3cf9c51ea9f8db4 | 19,395 | null | """w1_k8m -- K=8,388,608, spacing 2.4e-07, batch 8.
The coverage ladder on contest Easy so far: 262K 12.38%, 1M 23.71%, 2M 66.42%,
4.2M 82.71% -- still climbing even though steps fell 4,680 -> 1,572. e5's correctness
basin is 3.6e-7, so this arm sits well inside it and tests where the ceiling is.
Parameters: 33,554,43... | {
"id": "38b3c2cf-5a9e-4f6d-bc48-061e0846ce79",
"created_at": "2026-08-28 18:16:13.707432+00:00",
"db_md5": "aefbcd1d80f84c44795fd96c2da8cdfd",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "8a7dd691-40cd-4e83-9566-ad9a4f24d430",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 0.5821876433463398,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.94375
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.572303065871444,
"example_count": 600,
"... |
38be2b37-889b-4d2e-9d7d-56c090721f8a | easy | khushidahi | 2026-08-06 03:30:40.380962+00:00 | succeeded | 16bfc6f62f2ab83ef08d1d4222a060af0077d22cecee1e868439e114f438e748 | 14,364 | 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": "38be2b37-889b-4d2e-9d7d-56c090721f8a",
"created_at": "2026-08-06 03:30:40.380962+00:00",
"db_md5": "2cba24c3085f5a8a165f5b3b75afe7fd",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "3942bffb-95b4-4dd3-b9f7-c21c6d807e9c",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1585025787353516,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008333333302289248
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1627724170684814,
"example_count": 60... |
38c186f1-c99c-4e01-8c12-6f27bfe26e78 | easy | karanganesan | 2026-08-05 11:55:37.516548+00:00 | succeeded | 7d02c8c5da4773ccf5e1dfdc89c48fdcf33e7736d7cef4531e90a51a9bcef450 | 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": "38c186f1-c99c-4e01-8c12-6f27bfe26e78",
"created_at": "2026-08-05 11:55:37.516548+00:00",
"db_md5": "c3c44d28bb5fd98c8e6bdb7fc4fda49b",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "f99cc4c8-fc22-4b69-830d-d82f104d6168",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 10.449763774871826,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04333333298563957
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 13.136024475097656,
"example_count": 100... |
38c24ef5-1833-4177-9a66-7c6bc9dfece6 | easy | chad-atexpedient | 2026-08-19 21:10:38.199316+00:00 | failed | 9a1c94eec16effa240148e628f6bd890cd8bd714700b5c9d1623bbac374a792a | 22,152 | null | """X98: routed self-convolution transition.
A decimal square couples every pair of value places, and each pair contributes to
a place determined only by their relative positions. X98 adds a bounded 16-digit
learned bilinear X-X interaction grid whose features are routed by fixed place
addition into an output tape, plu... | {
"id": "38c24ef5-1833-4177-9a66-7c6bc9dfece6",
"created_at": "2026-08-19 21:10:38.199316+00:00",
"db_md5": "6b4b830040d0340a34e6ccd1d5a46315",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "f9ec8394-73a6-47c1-99c1-2e355e0cd82b",
"tier": "easy",
"dataset_id": "e5",
"st... | null
|
38c2848c-2faa-4765-908f-7fefce938908 | easy | blake-camp-surge | 2026-08-05 18:39:01.623464+00:00 | succeeded | 7e3c3a1c31d442b7ee7a38c96d16846fcb94c8fd32dc3f3271529eadc424ba33 | 10,387 | null | """Vector-state neural cellular automaton over attention edges."""
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,
)
TOKEN... | {
"id": "38c2848c-2faa-4765-908f-7fefce938908",
"created_at": "2026-08-05 18:39:01.623464+00:00",
"db_md5": "923d8b2dc95d785984bdb62e413ce7cd",
"submitter": "Blake Camp",
"github_login": "blake-camp-surge",
"run_id": "a0484fbc-db69-48e8-b865-c55c29302593",
"tier": "easy",
"dataset_id": "e3",
"status":... | {
"score": {
"mean_loss": 3.8012737689440925,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005625
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.944906297630756,
"example_count": 800,
... |
38c33554-5d66-49bd-a773-0fa29828f8fe | easy | sapient-sapiens | 2026-08-18 11:11:37.108350+00:00 | succeeded | 85729495efc1f5aae59c5f5b8f16f41efb58a5c8dbb929b814f0c89a0e3c3276 | 26,969 | null | """v44 adaptive baseline: one encoder, four-layer T-free core, bounded halt.
Locked routing: field-local one-layer encoder; recurrent core sees only N/X;
T is pooled solely for the controller; decode once on a halt-weighted state
mix. Stopping support is 8 depths, not 64. T is never parsed in forward or
the loss.
"""
... | {
"id": "38c33554-5d66-49bd-a773-0fa29828f8fe",
"created_at": "2026-08-18 11:11:37.108350+00:00",
"db_md5": "435dabebe848b3a53725188abe315d21",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "04259b96-59e0-4d0c-bc34-ae405a0d1e0b",
"tier": "easy",
"dataset_id": "e5",
"status": "su... | {
"score": {
"mean_loss": 2.145945646805782,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0033333333333333335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1491757047283273,
"example_count": 60... |
38c59b23-5b85-40ac-816b-911922778182 | easy | yashkant | 2026-08-13 00:14:22.199523+00:00 | succeeded | e5a8b81b1ed4b26dbbd05ca8efeb0659d4aed6d90c2d798713f9596f1ccce625 | 34,238 | null | """Round1282 shared continued-fraction denominator-recursion pair.
The complete Round1141 D416 MemGLU carrier remains byte-identical. Its pooled
representation seeds a positive terminal denominator. Every shared call adds
one tied continued-fraction level, ``P(x) + softplus(beta)/(state + epsilon)``.
The matched contr... | {
"id": "38c59b23-5b85-40ac-816b-911922778182",
"created_at": "2026-08-13 00:14:22.199523+00:00",
"db_md5": "79276cc2448a29e24bee51725da46792",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "aa9fc4ec-0a22-4fcb-9418-50a77eb2be19",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 4.80410361289978,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05166666768491268
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.691973686218262,
"example_count": 100,
... |
38c62871-c67e-42a8-bfbf-b6946661d5c0 | easy | shreyash-chonkie | 2026-08-31 23:04:01.467521+00:00 | succeeded | 85a7b94958371aa83a3d1b936aa2ac62f5ff8abe4cff944be09dad60cc261fec | 18,203 | null | """Parallel prompt binding with a learned spectral latent recurrence."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
ass... | {
"id": "38c62871-c67e-42a8-bfbf-b6946661d5c0",
"created_at": "2026-08-31 23:04:01.467521+00:00",
"db_md5": "8492a165c36cac955ed0ac7fc9b496da",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "9bebcdf1-50e0-4d88-affb-bdd497e38e1f",
"tier": "easy",
"dataset_id": "e1",
"status": "... | {
"score": {
"mean_loss": 1.9739575029311334,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06500000054637592
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9379386901855469,
"example_count": 100... |
38cae487-357a-4104-bffe-b7c039495ec2 | easy | jordanrubin | 2026-08-06 07:00:09.681000+00:00 | succeeded | 3cbe72033eb1c881013eb18ff96f70d0874f55a5fb001c85c4e2fda2d5091826 | 15,795 | null | """Gauge-free pair-relational continuous-boundary looped Transformer.
This is a learned transition model, not an arithmetic program. The fixed
parts expose only the public prompt grammar and six representation priors:
right-origin decimal slots, a shared ordered digit geometry, unordered pair
tokens, immutable N reca... | {
"id": "38cae487-357a-4104-bffe-b7c039495ec2",
"created_at": "2026-08-06 07:00:09.681000+00:00",
"db_md5": "1539d6a5ab39c751e21766f16114bf14",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "83077749-29a6-43ff-b4c8-30c415903167",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 1.9653446078300476,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014999999664723873
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9185973405838013,
"example_count": 10... |
38ceff13-c005-4565-9684-0f74a2265e0d | easy | jordanrubin | 2026-08-11 01:25:14.605060+00:00 | succeeded | aa0c446a01b1c50a3739ac906b125fbe05d6b1cd9cf0a23d1a564e062285e8d6 | 21,670 | 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": "38ceff13-c005-4565-9684-0f74a2265e0d",
"created_at": "2026-08-11 01:25:14.605060+00:00",
"db_md5": "344e901468d7fc4aff5006fca563aa6e",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "cf10e2e7-77f5-4b90-8ba4-d87c9103d75d",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.605735026123569,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0091666664617757
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1574028619469012,
"example_count": 600,
... |
38d174b1-95c7-4552-907a-ca2d129f8edd | easy | rougee | 2026-08-22 06:42:24.876262+00:00 | failed | e3ed96c9da5472d39dfc331b9799e4624316474ec318161156d6007f68306fd0 | 7,688 | null | """Official submittable bundle for One Layer Deeper Benchmark.
Architecture:
- Dynamic token metadata extraction for modulus N, input x, time steps T, and answer position.
- Sequential looped circuit depth recurrence for exact multi-step modular squaring (T in 1..64).
- Vectorized place-value digit decoding in pure Py... | {
"id": "38d174b1-95c7-4552-907a-ca2d129f8edd",
"created_at": "2026-08-22 06:42:24.876262+00:00",
"db_md5": "213a2447b561be7129f357cfbdb2bd14",
"submitter": "Rougee",
"github_login": "rougee",
"run_id": "ac4334ab-b9d8-400f-85b8-1f79330ee6a1",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
"... | null
|
38dd6c64-169b-4ea3-a3df-77424bf9ef50 | easy | isaac0804 | 2026-08-07 13:46:22.186384+00:00 | failed | be5e070ad35cfcc7154f1663d6d98e343cbc30016b8d405d6c27c4a3b8c1a9c3 | 13,450 | null | """Weight-tied digit-cell loop: a discrete-register arithmetic machine.
Full theory: docs/digit_cell_spec.md.
The per-iteration block is intended to be a tiny, N-agnostic digit-serial ALU
(multiply-add with carry, shift-compare-subtract) whose state is a block of
discrete digit registers (residue A, modulus N, produc... | {
"id": "38dd6c64-169b-4ea3-a3df-77424bf9ef50",
"created_at": "2026-08-07 13:46:22.186384+00:00",
"db_md5": "cd0acc269ce26cc3637bcd71f00a22ed",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "96d0e9b0-ef6f-42cd-b527-be8661fa0a42",
"tier": "easy",
"dataset_id": "e1",
"status": "faile... | null
|
38e27890-9a34-4e65-88ed-7719ce4b23eb | easy | yashkant | 2026-08-23 02:53:43.025306+00:00 | failed | 5670f8bd7becbbd14604ca6eb67c54895cf1d7ba6873397f0983a79dc7ed1801 | 7,088 | null | """One Layer Deeper — CRT residue-register machine (vectorized).
State = per-modulus registers packed as P[B,K,M], init onehot(x mod m_k). Shared
transition P <- S P (one einsum over all moduli), S_k learned column-stochastic,
applied T times. Since (y^2 mod m) depends only on (y mod m), each register
exactly tracks y... | {
"id": "38e27890-9a34-4e65-88ed-7719ce4b23eb",
"created_at": "2026-08-23 02:53:43.025306+00:00",
"db_md5": "78a53c165c178715a98a4db0b65bf396",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "0eeb0388-38b5-4a18-8549-64b834a617b3",
"tier": "easy",
"dataset_id": "e1",
"status": "failed"... | null
|
38e2e59e-ad46-4813-8b0a-d31b97684db1 | easy | DDanlov | 2026-08-28 21:45:32.776437+00:00 | succeeded | 2c8d38cdda8944bd3252b7adf7234b5d44d3e0bc66bbfd9487b5ced3f789a94b | 14,939 | null | """
Trial 15: Model B: D=256, H=4, M=16, u=2 (T=32)
"""
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, ... | {
"id": "38e2e59e-ad46-4813-8b0a-d31b97684db1",
"created_at": "2026-08-28 21:45:32.776437+00:00",
"db_md5": "b4fd46c81030be735aa9967e72056d67",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "c196acfd-257b-4f2d-8454-015360641264",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.83321353302519,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0008333333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8332135109356167,
"example_count": 600... |
38ec8326-343f-415c-a592-308d8f626eeb | easy | 0Chris5R | 2026-08-01 06:28:40.675363+00:00 | succeeded | 25d05da3e0ad0b7a927fef68963a953bc127365a0e4be0001bdc02294b33ac86 | 20,718 | null | """Shared-digit Neural GPU with a compute-matched operator curriculum.
The model aligns decimal digits by significance, forms a learned low-rank
second-order feature map of the current state, and evolves a two-dimensional
workspace with the original convolutional-GRU equations. Each arithmetic
transition applies the ... | {
"id": "38ec8326-343f-415c-a592-308d8f626eeb",
"created_at": "2026-08-01 06:28:40.675363+00:00",
"db_md5": "0a4ed1ff27fb5e39509d04a9ae7b58d8",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "d30ebc95-7596-45ca-b5e9-431e39e3c5d7",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.683605568008273,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.017916666666666668
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.669698292899025,
"example_count": 600,... |
38f228d3-0a1e-4eca-ac38-c9e5f3cc1a06 | easy | DDanlov | 2026-08-13 20:34:14.099520+00:00 | succeeded | f64ff3cd0684f4fa9ae5db95971ae798b7a58b467b4249cf736f4a6ed56c0ee1 | 27,054 | null | from __future__ import annotations
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
try:
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
except ... | {
"id": "38f228d3-0a1e-4eca-ac38-c9e5f3cc1a06",
"created_at": "2026-08-13 20:34:14.099520+00:00",
"db_md5": "8c940d6a1a5959c43d1a96afef12c71c",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "d3f6ccc1-7376-42cc-a986-926a4a10ff31",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.353529485778223,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666666666667
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.351234259644175,
"example_count": 300,... |
38f38c27-8787-48c0-aa52-876f80b4163c | easy | oupadhyay | 2026-08-22 17:10:53.535092+00:00 | succeeded | 4596fc187c0d54b7d54ada1c79ec07ba53006df424ede715b0348d9054a8680b | 4,196 | null | """Two-microtick learned role--filler tensor-product transition."""
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle,Submission,assert_model_state
PAD,N,X,T,ANS,DIGIT=0,2,3,4,5,7
P,FILLER,ROLE,C,MAX_T,TICKS=4,32,16,4,64,2
STATE=14186
class Config:
def __init__(s,s... | {
"id": "38f38c27-8787-48c0-aa52-876f80b4163c",
"created_at": "2026-08-22 17:10:53.535092+00:00",
"db_md5": "5cadd8b8b35c3c92c5518828ed7e1785",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "f3ac5aa2-396d-4bbd-a305-5421749b4a7c",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.150765581651773,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0054166666790843015
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1510008985151625,
"example_count": 60... |
390be00f-c0eb-48dd-a3d6-ca69f5e8942f | easy | AFeinfield | 2026-08-18 23:21:35.153550+00:00 | succeeded | 9e08b14d3bc25f054832e2fd53fba2bbf69c529596340bc176309f34cf34049f | 17,388 | null | """080: 077 with contractive residual alpha init 1.0 (was 0.5).
Each iteration could previously only move the state halfway; over a
16-step chain that heavily damps signal AND gradient. Tests whether the
contraction is buying stability or just slowing the fit.
Original 077 notes follow.
077: ACTIVE-ROW trajectory loop... | {
"id": "390be00f-c0eb-48dd-a3d6-ca69f5e8942f",
"created_at": "2026-08-18 23:21:35.153550+00:00",
"db_md5": "52c4b63dba0025234cc44e49c75327e4",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "3ee84563-2214-4044-80a1-68ff5f6361d5",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 2.964237928390503,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006041666609235108
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4289462566375732,
"example_count": 300... |
390ce303-9d0b-4c91-b8d3-1e1f4064ca44 | easy | falloficaruss | 2026-08-06 06:06:09.816460+00:00 | succeeded | 9e84d421817180114b307e9e20266750ccdcb7697c402d821c90de2ce589f72f | 8,972 | null | """Deep encoder v11 — Medium m1 (fixed N) attempt.
Lessons
-------
m3/v7: 31k steps, train 69% exact / test 0.25% — memorized varying-N table.
e3/v9 T-latent: healthy gap, slow learning (loss 1.84, test 0.8%).
e3/v10 multi-digit latent: same, ood≈test≈0.75%.
m1 is different: N=10403 is constant. The map is only (x,T)... | {
"id": "390ce303-9d0b-4c91-b8d3-1e1f4064ca44",
"created_at": "2026-08-06 06:06:09.816460+00:00",
"db_md5": "3394368b267a6dbba432fb2639e38679",
"submitter": "Abhishek Shinde",
"github_login": "falloficaruss",
"run_id": "9b358632-41a7-4fea-9544-136265962bcf",
"tier": "easy",
"dataset_id": "e2",
"status... | {
"score": {
"mean_loss": 3.884492516517639,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004375000135041773
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.881488800048828,
"example_count": 300,... |
390dd97e-7794-44c0-8a5b-c298de9f091e | easy | sapient-sapiens | 2026-08-16 19:19:39.392367+00:00 | succeeded | 74bd52a04f6b3cbc180fa4427c00f4deafffec8f5c137ceda66ba5db574f3b91 | 19,561 | null | """E6 width-2048 optimizer swap: AdamW+SAM, no Muon.
Frozen v73 (WIDTH=2048, d_h=4, L=6, batch 128) except core matrices use
AdamW instead of Muon. SAM (rho=0.02, two backward passes) is unchanged.
T is diagnostic-only.
"""
from __future__ import annotations
import math
import time
import torch
import torch.nn.func... | {
"id": "390dd97e-7794-44c0-8a5b-c298de9f091e",
"created_at": "2026-08-16 19:19:39.392367+00:00",
"db_md5": "613b6dc1d3a459a21853610495684534",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "c4abb0e8-88e3-4a09-9857-54ec17ed8989",
"tier": "easy",
"dataset_id": "e6",
"status": "su... | {
"score": {
"mean_loss": 1.9233465480477843,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08536036433400335
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.7495356798171997,
"example_count": 60,... |
390e8647-8807-4625-a786-633df4899363 | easy | oupadhyay | 2026-08-14 16:16:01.418939+00:00 | succeeded | 30aef0bfb17b73f13a7b7e0aea5a027bf05be7759b7ea8884bb46151c97e75e3 | 3,984 | null | """Low-capacity recurrent transducer with global contextual N conditioning."""
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_siz... | {
"id": "390e8647-8807-4625-a786-633df4899363",
"created_at": "2026-08-14 16:16:01.418939+00:00",
"db_md5": "0e8f62666fd6920e0662ac77c1545174",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "0563c19a-03fa-4ae2-962e-1eefb3c564fa",
"tier": "easy",
"dataset_id": "e3",
"status": "s... | {
"score": {
"mean_loss": 2.2692788616042727,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000013969839
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2395920240934224,
"example_count": 80... |
3917a5c1-584d-4f25-a68e-e6ebeda2c5ff | easy | yashkant | 2026-08-09 03:02:40.685478+00:00 | succeeded | 075b8b560e2bc7aa6189a86079343a506a8a99429e6db01db1e1aee5ebf84746 | 22,841 | null | """Round 211 common TRM-style x/y/z shared-transition template."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
from torch import... | {
"id": "3917a5c1-584d-4f25-a68e-e6ebeda2c5ff",
"created_at": "2026-08-09 03:02:40.685478+00:00",
"db_md5": "6e76b66c89d2b63b0d67174df1ee1970",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "04aa2a62-4978-47cd-a1d5-9c6118708d38",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 2.1179708425618164,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01687500009313226
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1160716661446024,
"example_count": 800... |
3920e505-babb-48cf-9a47-8c310b9aa1ed | easy | isaac0804 | 2026-08-05 08:33:02.284386+00:00 | succeeded | 0ab40eb7cfa42b9700f7e4a4f170eba07ae49a93a935f124b68d54c84d706635 | 9,365 | null | """Weight-tied looped Transformer (v6 core: damped gate, 2 sublayers/iteration)
plus explicit place-value embeddings for N's and x's digit tokens.
Motivation (see docs/findings.md, N-conditioning section): every architecture
tried so far treats N's (and x's) digits as an opaque span for generic
attention to notice imp... | {
"id": "3920e505-babb-48cf-9a47-8c310b9aa1ed",
"created_at": "2026-08-05 08:33:02.284386+00:00",
"db_md5": "6c332bc632593f83d44af5125cddee51",
"submitter": "Isaac Yong",
"github_login": "isaac0804",
"run_id": "6a789b02-5cf2-4b93-9635-14114a8a8cc7",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.993898630142212,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.056666663847863674
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0064151287078857,
"example_count": 100... |
392150bf-1fe5-4591-b083-19a6687656b8 | easy | gauravmishra | 2026-08-09 19:09:53.529749+00:00 | succeeded | ba2afdc01064dec98d35b7c2d308bb0e9f3b003326153aa05a53541dc3828d36 | 4,268 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
D_MODEL = ... | {
"id": "392150bf-1fe5-4591-b083-19a6687656b8",
"created_at": "2026-08-09 19:09:53.529749+00:00",
"db_md5": "f1b8443587827c8185c0470874cb2698",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "c35f6d70-0e62-445e-bac0-0bb89f9e63e5",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.254117936518532,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2842779822413695,
"example_count": 600... |
3922b29b-1a31-4405-9ec8-a421c75869a8 | easy | DDanlov | 2026-08-21 02:01:52.613029+00:00 | succeeded | 91112618bdd4f4934ac04635020693646dbe83b4b0eccc5196648acd6981cb59 | 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": "3922b29b-1a31-4405-9ec8-a421c75869a8",
"created_at": "2026-08-21 02:01:52.613029+00:00",
"db_md5": "6502c33bd7751df57f597f23503893f1",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "621115cc-9932-4665-a775-9efa71481c35",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.7365588280456787,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002916666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.7338732722896095,
"example_count": 60... |
3925c020-9996-457c-bd2b-f618ae170e35 | easy | khushidahi | 2026-08-17 13:33:40.236073+00:00 | succeeded | 6328b7e499c3f2df6f3023d9b61809d506cbecc4451d18b954802702677af329 | 30,134 | null | """R192: warm-started task-aligned TRM answer/workspace transition.
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 su... | {
"id": "3925c020-9996-457c-bd2b-f618ae170e35",
"created_at": "2026-08-17 13:33:40.236073+00:00",
"db_md5": "e5f1d5833178ea62c216b22dc3fe22e8",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "f58b1292-ad11-43b1-afd7-23aba98d2978",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.5477185692817703,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014583333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4847476268563033,
"example_count": 60... |
392e28ce-114f-4b15-b5d9-403e1ba9e8a4 | easy | shirvani-jr | 2026-08-29 00:49:56.694846+00:00 | succeeded | ec63e3be3be21d162cdd713f8b6a6f8eeae78e8e6bb4f8b885675655198a81f6 | 49,678 | null | """Adaptive learned discrete recurrent fabric for One Layer Deeper.
A gradient-trained, operation-free recurrent computational machine. The
model learns, from the competition's endpoint supervision: how to parse the
prompt into a bank of categorical digit registers, a reusable transition
circuit applied serially, a l... | {
"id": "392e28ce-114f-4b15-b5d9-403e1ba9e8a4",
"created_at": "2026-08-29 00:49:56.694846+00:00",
"db_md5": "44ce50c9cda91dc66aadee2488c2b2f5",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "a9534452-ba4a-47dc-9145-d66330de0118",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.8950763940811157,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.36666666343808174
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.587101936340332,
"example_count": 100,... |
392eaa57-3790-4bb0-a5a9-f2d2eb762a6e | easy | epicosy | 2026-08-20 12:13:16.357784+00:00 | succeeded | 271f11a54caa1b23d6ec3ed7999d93aa33bd02b537c0a34abea07c3fd104c349 | 4,460 | null | """Shared-refinement Transformer baseline for the Easy tier."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEAD... | {
"id": "392eaa57-3790-4bb0-a5a9-f2d2eb762a6e",
"created_at": "2026-08-20 12:13:16.357784+00:00",
"db_md5": "81bdf8de54eee0924b64d79557021ff0",
"submitter": "Eduard Pinconschi",
"github_login": "epicosy",
"run_id": "5bf2ec9f-dc9b-4a9e-b93e-7556dccd8016",
"tier": "easy",
"dataset_id": "e6",
"status": "... | {
"score": {
"mean_loss": 2.129182517528534,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.021846847608685493
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3304998874664307,
"example_count": 60,... |
3930407a-647b-49a4-af89-e8857a57fca1 | easy | kushasareen | 2026-08-10 22:19:51.918407+00:00 | succeeded | 2cf825ea522bc483b60db3516c1529da720bb843f164d15c083e63f7f809a389 | 5,777 | 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": "3930407a-647b-49a4-af89-e8857a57fca1",
"created_at": "2026-08-10 22:19:51.918407+00:00",
"db_md5": "4275feeff758e8157fadbaff6bf088c8",
"submitter": "Kusha Sareen",
"github_login": "kushasareen",
"run_id": "dc8e882d-5454-4c13-b113-5ac87518a506",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.4837318996684408,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000049670536
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.399765347151478,
"example_count": 600... |
393f6545-14e3-4a2e-b996-559e3afad2c8 | easy | Bananafly | 2026-08-12 09:56:41.023068+00:00 | succeeded | 529c23011f7700627f7967cd7524d76f79934234c4d6c03db9e55fd724793e94 | 9,768 | null | """Fast T-aligned decimal-residue MLP for Easy T<=3 tasks."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = ... | {
"id": "393f6545-14e3-4a2e-b996-559e3afad2c8",
"created_at": "2026-08-12 09:56:41.023068+00:00",
"db_md5": "6523708511fdc1862f32b001999a9a46",
"submitter": "Andre Kreidemann",
"github_login": "Bananafly",
"run_id": "81b497d1-f7aa-4517-8f2f-5ac8d8562d64",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.031863808631897,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06666666455566883
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8569660186767578,
"example_count": 100,... |
393f8737-d19f-43f7-953c-af72ef27fa22 | easy | pjjajal | 2026-08-07 03:37:50.143225+00:00 | succeeded | ac79edb51f86d810a7ec47673aaa63386c69fd13840b5505e01e342a58f2b212 | 12,472 | null | """Looped Transformer with an outer-product residual state.
Each token carries several residual streams. A dynamic read contracts those
streams to one Transformer-width vector; a dynamic write forms an outer product
between per-stream write weights and the Transformer output, then adds that
matrix to the identity-car... | {
"id": "393f8737-d19f-43f7-953c-af72ef27fa22",
"created_at": "2026-08-07 03:37:50.143225+00:00",
"db_md5": "db984d164af63cc4ef78923d77fda3ef",
"submitter": "pjjajal",
"github_login": "pjjajal",
"run_id": "c4db8cd2-d274-499e-a3e2-d7085b4e7be7",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 1.9750056862831116,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.07333333045244217
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9784643650054932,
"example_count": 100... |
394d852b-5e9b-44f9-8f09-cd717c2cacc1 | easy | oupadhyay | 2026-08-07 12:47:46.285498+00:00 | succeeded | 4d884b144cfce849c718ac4475afae469a7f366404828172b455e8d8d8f08501 | 8,556 | null | """Factorized pair reducer with compute-efficient additive N conditioning."""
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle, Submission, assert_model_state
PAD, N, X, T, ANS, DIGIT = 0, 2, 3, 4, 5, 7
D, PLACES, MICROPHASES, MAX_STEPS = 64, 4, 1, 64
STATE_ELEME... | {
"id": "394d852b-5e9b-44f9-8f09-cd717c2cacc1",
"created_at": "2026-08-07 12:47:46.285498+00:00",
"db_md5": "0981b2c259848d76aee0739a91c8569d",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "243c1a92-ee6d-46a4-9188-2e102b189237",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.2103483593737385,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0075000000496705365
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.216489208118798,
"example_count": 60... |
395049ae-9edb-41af-a8d4-a319035361f9 | easy | oupadhyay | 2026-08-07 13:23:08.515129+00:00 | succeeded | bbdcb3cefb6e0ff2199333924b0ad85c0ee8b5b71c41e6f9dd4361103fe929da | 8,512 | null | """Factorized pair reducer with compute-efficient additive N conditioning."""
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle, Submission, assert_model_state
PAD, N, X, T, ANS, DIGIT = 0, 2, 3, 4, 5, 7
D, PLACES, MICROPHASES, MAX_STEPS = 64, 4, 1, 64
STATE_ELEME... | {
"id": "395049ae-9edb-41af-a8d4-a319035361f9",
"created_at": "2026-08-07 13:23:08.515129+00:00",
"db_md5": "5c422b7d962fbd41ff9976e48647d33b",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "b9c8fe07-1c49-40de-a797-fc101abbe3b2",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.1546903282124923,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1565283392576893,
"example_count": 60... |
3955252c-afe2-4d8d-81b4-d88530441e27 | easy | alirezashirvani-jr | 2026-09-01 00:15:36.422828+00:00 | succeeded | 4e1627e58581890aade082c87345df5801df094a6ea1ceeb45d8543da364193e | 16,829 | 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": "3955252c-afe2-4d8d-81b4-d88530441e27",
"created_at": "2026-09-01 00:15:36.422828+00:00",
"db_md5": "241d7f2990929d86b32dfafd9255f4c0",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "7b330869-91a1-4cab-8581-5dba86f0935c",
"tier": "easy",
"dataset_id": "e5",
... | {
"score": {
"mean_loss": 2.384216941926662,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.010000000484287739
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3886051177978516,
"example_count": 600... |
3956b053-f9e0-4000-9dfe-b23c564a1767 | easy | erdavis0 | 2026-08-22 07:40:00.423000+00:00 | succeeded | 4052302344c2ee5544752d8dd2b682cc62f86ed61a7d43e055b398661db547e4 | 4,774 | null | """Masked butterfly-style global mixing over sequence positions."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
WIDTH = 64
HIDDEN... | {
"id": "3956b053-f9e0-4000-9dfe-b23c564a1767",
"created_at": "2026-08-22 07:40:00.423000+00:00",
"db_md5": "7b255a5697e39c756d9c195ef0c4c794",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "f4a70865-f8af-4b30-a928-0d09972aa21e",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 8.153066396713257,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0283333333209157
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.91909646987915,
"example_count": 100,
... |
39597565-75d8-4a88-adbf-557824e28b6e | easy | adambrzosko | 2026-08-30 14:47:16.863365+00:00 | succeeded | 8df848dc7ed778cc7565eb6b555106090c05ca4790ec1b9f8402a063938ac6b4 | 12,028 | null | """One Layer Deeper — digit-slot recurrent machine ("dm").
The prompt is [N] digits(N) [X] digits(x) [T] digits(T) and the label is the
decimal expansion of x^(2^T) mod N, tail-aligned: the last prompt position
carries the units digit, the one before it the tens digit, and so on.
Architecture. The model parses the ... | {
"id": "39597565-75d8-4a88-adbf-557824e28b6e",
"created_at": "2026-08-30 14:47:16.863365+00:00",
"db_md5": "35eb7d319679f1481c817075fbaed148",
"submitter": "Adam Brzosko",
"github_login": "adambrzosko",
"run_id": "3c0ca58f-322e-46d8-9995-7c9df427f980",
"tier": "easy",
"dataset_id": "e6",
"status": "s... | {
"score": {
"mean_loss": 2.402464509010315,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0468468489125371
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3274991512298584,
"example_count": 60,
... |
395b2385-09de-4a45-bb3a-2cf3692ca64e | easy | repst | 2026-08-09 21:40:56.980677+00:00 | succeeded | 36dd679dcb477f7aa19beb9d8cd039a7f44fea0c10036f14f82fddcba0cbe31f | 23,177 | null | """Strict forced-T operator-learning diagnostic for the public prompt contract."""
from __future__ import annotations
from typing import NamedTuple
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,... | {
"id": "395b2385-09de-4a45-bb3a-2cf3692ca64e",
"created_at": "2026-08-09 21:40:56.980677+00:00",
"db_md5": "1db907953e7b39fe8d31af5b24eee25b",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "e1fb2ced-66ce-4190-99ef-9e5a42118bac",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 7.931864023208618,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.08500000089406967
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.786234855651855,
"example_count": 100,
... |
395efec1-7c41-4616-9697-52c12314347f | easy | alirezashirvani-jr | 2026-08-30 21:41:51.782872+00:00 | succeeded | 3fa6488bbbb697ef739fb15cafcce44b37ecb85477b348af20ab3ea038d5545c | 24,483 | null | from __future__ import annotations
import math
import time
from contextlib import nullcontext
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, TokenLossBatch, assert_model_state
DIGIT_CARDINALITY = 10
STEP_FIELD_W... | {
"id": "395efec1-7c41-4616-9697-52c12314347f",
"created_at": "2026-08-30 21:41:51.782872+00:00",
"db_md5": "41fd7a05980039664511de3256ab2ebb",
"submitter": "alirezashirvani-jr",
"github_login": "alirezashirvani-jr",
"run_id": "450f29a7-6b1b-4292-abb3-80136b4c6c70",
"tier": "easy",
"dataset_id": "e1",
... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 100,
"exact_accuracy": 1.0,
"... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.