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 |
|---|---|---|---|---|---|---|---|---|---|---|
15a744db-fe47-421c-824f-a299d913143a | easy | jordanrubin | 2026-08-02 18:08:17.828570+00:00 | succeeded | d40f14a7b67e0c2e4e77f8e65b9c97e34f0d14f363da65a9ba371a1f0c8c8143 | 13,992 | 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": "15a744db-fe47-421c-824f-a299d913143a",
"created_at": "2026-08-02 18:08:17.828570+00:00",
"db_md5": "51382647a8e95a70e3475a660194a03a",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "10200997-f6c6-4458-abc6-6eff1ff85b37",
"tier": "easy",
"dataset_id": "e1",
"status": "s... | {
"score": {
"mean_loss": 4.149927522922097,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03833333247651657
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.217482209205627,
"example_count": 100,
... |
15a955aa-a35b-4465-b7c8-57b97850b816 | easy | Yalyenea | 2026-08-11 19:24:13.447367+00:00 | succeeded | 2f2e0971646a6d921b39fa3ee4f15ec8f361e50f59ef1738c3ea1f0f73d7dda9 | 12,139 | null | """R4 full-bandwidth castnorm candidate with fixed source recall after feedback."""
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... | {
"id": "15a955aa-a35b-4465-b7c8-57b97850b816",
"created_at": "2026-08-11 19:24:13.447367+00:00",
"db_md5": "e4b0fd656fc62f35681d78f316e70ce9",
"submitter": "yfff",
"github_login": "Yalyenea",
"run_id": "849188e1-5127-455a-b0b2-81bee9c1b602",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 6.51559673692729,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013333334075286984
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.421787261962891,
"example_count": 600,
... |
15a9d7e6-0953-44e5-aa8d-a1208644a624 | easy | blake-camp-surge | 2026-08-05 19:12:00.508576+00:00 | succeeded | 7aeccf4ed7ce8e58dff503129f29834dc0178b8b8dcd6054f225ef82c052f3a9 | 10,517 | 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": "15a9d7e6-0953-44e5-aa8d-a1208644a624",
"created_at": "2026-08-05 19:12:00.508576+00:00",
"db_md5": "e35950247776e2834eb81f7fe89a78b1",
"submitter": "Blake Camp",
"github_login": "blake-camp-surge",
"run_id": "f85164d4-0a7f-4ffe-baee-064d8a5db119",
"tier": "easy",
"dataset_id": "e3",
"status":... | {
"score": {
"mean_loss": 2.123493756068178,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005625000000000001
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.121503811600733,
"example_count": 800,... |
15acb005-58bb-4d12-a35d-32e6cde63d1b | easy | DDanlov | 2026-08-28 21:53:01.667223+00:00 | succeeded | b43d8e5ca9191158bb89ee37908673803ade8f1dc0e1eb80bd0acc9681a52329 | 14,935 | null | """
Trial 21: Model B: D=512, H=8, M=8, u=2 (T=16)
"""
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, T... | {
"id": "15acb005-58bb-4d12-a35d-32e6cde63d1b",
"created_at": "2026-08-28 21:53:01.667223+00:00",
"db_md5": "83677b590c6dfc3802915ab5c24729e4",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "26fee6db-262e-4bfd-b995-72b7b7de5909",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.83321353302519,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0025
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8332135109356167,
"example_count": 600,
"ex... |
15b53e20-ff50-4064-91b1-3c24890eae4b | easy | sapient-sapiens | 2026-08-18 10:23:12.110721+00:00 | succeeded | 4f57e9b4fb11ccc98eade9fa49befc0c2a4757ef36e79691646ce253c90f41c9 | 24,080 | null | """v43: v40 lineage with the halting controller removed and a fixed unroll.
Diagnosis this arm responds to. Across every stored local run the learned
controller emitted a single constant depth for the whole dataset (1032 of 1037
depth-vs-T cells had exactly zero per-example variance), and forced-depth
sweeps showed lo... | {
"id": "15b53e20-ff50-4064-91b1-3c24890eae4b",
"created_at": "2026-08-18 10:23:12.110721+00:00",
"db_md5": "5158bfac4bfd0f659f580f6e6eb0fa62",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "1cfcee54-ea60-4da0-855a-45448a3d5285",
"tier": "easy",
"dataset_id": "e9",
"status": "su... | {
"score": {
"mean_loss": 1.9348184076004125,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.034444444750746095
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.9291605949401855,
"example_count": 90... |
15ba7144-b314-45c8-a13b-384f2d34dc32 | easy | DDanlov | 2026-08-26 20:39:37.290642+00:00 | succeeded | 6b2a41d8a86ba26f5b2915c748dea4c27166d088f380d0036207a0318894d3ba | 34,345 | null | import math
import time
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from onelayerdeeper.api import Submission, OptimizerBundle, ModelSpec, OptimizerSpec
except ... | {
"id": "15ba7144-b314-45c8-a13b-384f2d34dc32",
"created_at": "2026-08-26 20:39:37.290642+00:00",
"db_md5": "6698f98a9dbbb1997edeb6751f3e2d22",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "5e938e60-4b43-4efe-99e6-5e2a08393c54",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.3196196853547155,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.31783143714939,
"example_count": 600,
... |
15bb923c-be8b-416a-bd85-eea5d36c800d | easy | velocizapkar | 2026-08-06 21:47:14.181160+00:00 | succeeded | 2bf193da97239338231b279f71fffe736d0afacbe988e9b36a6ff595b5d2ee0e | 52,965 | null | """Fast packed residual-free Attention-Residual transformer with Muon.
The model is a generic learned sequence model. It contains no parser,
arithmetic primitive, task constant, generated data, label feedback, or answer
construction. Every macro-state is produced by a serial pack of attention and
MLP transformations... | {
"id": "15bb923c-be8b-416a-bd85-eea5d36c800d",
"created_at": "2026-08-06 21:47:14.181160+00:00",
"db_md5": "3258b6dad0913ef455225090617907d0",
"submitter": "Aakanksh Zarapkar",
"github_login": "velocizapkar",
"run_id": "5a484806-5f96-4c97-8eff-37d4660c79f8",
"tier": "easy",
"dataset_id": "e3",
"statu... | {
"score": {
"mean_loss": 3.224522113800049,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01687499973922968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.227999687194824,
"example_count": 800,
... |
15bdc610-aa95-410d-8e0c-f9d0dc434b1c | easy | Avyukth | 2026-08-11 20:23:45.581131+00:00 | succeeded | 258a605997c9529871d8aa5233a220dafa9457234e24aeed54d7d6f93b98fc8b | 15,011 | null | """T-adaptive recurrent Transformer: loop count scales with T read from the prompt.
The benchmark family composes a transition function T times. This model bakes
that structure in: it decodes the integer T from the prompt's digit tokens and
applies one shared block BASE + K*T times. Phase embeddings are cyclic (period... | {
"id": "15bdc610-aa95-410d-8e0c-f9d0dc434b1c",
"created_at": "2026-08-11 20:23:45.581131+00:00",
"db_md5": "964119335c789f1cd6bd06a2b769692a",
"submitter": "Subhrajit Makur",
"github_login": "Avyukth",
"run_id": "1ede3d71-7e72-47c2-a8cc-05b95ee6fd6a",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 2.212733507156372,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.044999998062849045
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.066493034362793,
"example_count": 100,... |
15c553d9-8da0-4e71-8274-a282fc777e52 | easy | 0Chris5R | 2026-08-01 06:34:30.275624+00:00 | succeeded | 24ae55811e6c5cab92709f271deb2bf90d503d398c9c81aa9ea33f81bc2894cf | 20,776 | 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": "15c553d9-8da0-4e71-8274-a282fc777e52",
"created_at": "2026-08-01 06:34:30.275624+00:00",
"db_md5": "98b5de46788ad3ac9d0f6a9a161ce482",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "7205385b-3d60-4cc3-a3e0-7afe09d3c0f1",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.4986197788220017,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009166666679084301
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.341090015231761,
"example_count": 600... |
15d16012-90fa-4745-a8cc-75371a611584 | easy | briheet | 2026-08-07 03:17:56.235069+00:00 | succeeded | 8361aae532f50ee9f49623f5e9e7f2c5d3512b888852ab97711d83a76bd80bbc | 4,097 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 32
NUM_HEADS = 4
NU... | {
"id": "15d16012-90fa-4745-a8cc-75371a611584",
"created_at": "2026-08-07 03:17:56.235069+00:00",
"db_md5": "eef568cd5d53d93ec3cb21c75af50c8a",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "85275966-612c-4954-817d-e2e4d4aaaeee",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.3580538034439087,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05333333043381572
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.798952579498291,
"example_count": 100,... |
15d1af8e-01ad-462f-a60c-1f0e16f1213e | easy | 0Chris5R | 2026-08-07 10:44:40.922398+00:00 | succeeded | 1e99f7146df160ce450b7ea3a381dfa41d96ec7374ec35dbeba607c4368500a3 | 20,840 | 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": "15d1af8e-01ad-462f-a60c-1f0e16f1213e",
"created_at": "2026-08-07 10:44:40.922398+00:00",
"db_md5": "f60abe8cb876d37f63c1f571e0e5c3c2",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "2de35ce4-c08e-40a5-a9b1-e5f94e1b2f3f",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.7461985429539775,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03750000002483527
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6822458799644435,
"example_count": 600... |
15d5d970-df9d-4717-a6cc-730e77e512e8 | easy | briheet | 2026-08-05 11:11:01.884848+00:00 | succeeded | c7282923f8e8d33715355454420c96e039182c24fe8c0d50dcd61ddd151dfe3b | 4,176 | 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": "15d5d970-df9d-4717-a6cc-730e77e512e8",
"created_at": "2026-08-05 11:11:01.884848+00:00",
"db_md5": "6aad821116984cd487aec62bc03665c6",
"submitter": "Aizenn",
"github_login": "briheet",
"run_id": "5ac9812d-d160-4763-9594-c041170e64f6",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",... | {
"score": {
"mean_loss": 2.4606672525405884,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0283333333209157
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.537832736968994,
"example_count": 100,
... |
15dfcfbf-88fe-49c8-9053-b2eb734cce00 | easy | sapient-sapiens | 2026-08-18 19:37:16.506163+00:00 | succeeded | 409e8baf080a4dccc21e0e196f290c11712557a961cf2d5a4032dd38516d098c | 31,707 | null | """Depth-quantized recurrence with a mixture-over-depth objective.
The recurrent state is a bank of right-aligned digit slots. Every operator
application is followed by a soft quantization back onto the token simplex, and
the digit logits produced by that quantization are the answer logits at that
depth. There is no s... | {
"id": "15dfcfbf-88fe-49c8-9053-b2eb734cce00",
"created_at": "2026-08-18 19:37:16.506163+00:00",
"db_md5": "4e9d07298f3be899224809f5a201e005",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "a3ab7540-4162-4292-b11d-19100ddbc70b",
"tier": "easy",
"dataset_id": "e9",
"status": "su... | {
"score": {
"mean_loss": 5.634714603424072,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04888889007270336
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.653122425079346,
"example_count": 90,
... |
15e1c0ab-938f-47c9-83a8-9cebd84b3a09 | easy | viridale | 2026-08-05 21:38:30.232742+00:00 | succeeded | 248d09e659399a89a42f935911249ad91b721c6eb694fb50d14e5be3b6cba5e8 | 9,273 | null | """Clean v177 recurrence with loss-adaptive power-law batch reuse.
hypothesis: uniform endpoint sampling creates a pathological compositional landscape;
repeating easier-than-current batches with multiplicities 1/2/4/8 may create the
asymmetric stepping-stone frequency distribution predicted to accelerate impl... | {
"id": "15e1c0ab-938f-47c9-83a8-9cebd84b3a09",
"created_at": "2026-08-05 21:38:30.232742+00:00",
"db_md5": "b82096d8098b203a4d1e562c93c9966c",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "a4a66aa0-84a5-4435-b01e-d910059834e9",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 2.181787133216858,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.194035768508911,
"example_count": 100,
... |
15e57aad-e3e8-42c3-adc9-a740b358abd9 | easy | shan23chen | 2026-08-09 02:45:15.022410+00:00 | succeeded | f4f5fe15140f16834d8d178fc085cc98481a55f43a3fd4849ce5628e2764e52c | 39,361 | null | """Dual-root boundary-reset recurrent register Transformer.
One learned transition is applied once per requested outer step. The
transition receives an immutable decimal N register and the current learned
boundary state, but never receives T or a separate copy of the original X.
Within each outer application, scratch... | {
"id": "15e57aad-e3e8-42c3-adc9-a740b358abd9",
"created_at": "2026-08-09 02:45:15.022410+00:00",
"db_md5": "dcabd7f95bf9105a936035b7eede2450",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "b3f0c15a-0291-4b74-97d2-f725e92bd218",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 2.2508745193481445,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.013749999925494194
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2455711364746094,
"example_count": 80... |
15eae087-eeba-48fc-b1c9-a563d7716de2 | easy | khushidahi | 2026-08-05 22:29:49.692409+00:00 | succeeded | 37edf81b755d423fbc19a954b3b484f979aebc585c27d2da32d3632011b6c14c | 25,699 | null | """R16 hard-transfer hybrid experiments for One Layer Deeper.
Motivated by complementary Hard behavior:
- the generic sequence recurrent model retained seen-N first-rung progress;
- the structured decimal-tape model retained OOD-N first-rung progress.
The dual variants co-train both learned representations, supervise... | {
"id": "15eae087-eeba-48fc-b1c9-a563d7716de2",
"created_at": "2026-08-05 22:29:49.692409+00:00",
"db_md5": "2b39ee603e92d37d22690f6bdca13acb",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "0641a4ef-df5a-4ae0-97eb-3656cf4f5fa2",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.175693482462619,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.015416667262713114
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.177978038787842,
"example_count": 600,... |
15edbb07-e362-4aad-b53f-40e340a6e52c | easy | shirvani-jr | 2026-08-21 21:29:43.728478+00:00 | succeeded | 880be9db70e690b8edff7ad8c57033a6a2b8bdc4d6fc982ed340ceae431082d8 | 8,666 | null | """Looped digit-state squaring machine.
One recurrent application of a shared block corresponds to one modular
squaring. The recurrent state is the decimal digit string of the current
value, re-quantised every step with a straight-through estimator, so the
learned map composes exactly and depth extrapolates in T.
"""... | {
"id": "15edbb07-e362-4aad-b53f-40e340a6e52c",
"created_at": "2026-08-21 21:29:43.728478+00:00",
"db_md5": "de6bbc1d13f150654971bc3d162380f2",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "abd38919-4cb2-4559-9a16-542673d25aa6",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.3227699398994446,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03500000108033419
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.875057578086853,
"example_count": 100,... |
15eff7c4-c18c-4c04-8c31-d2549da76de3 | easy | c-nnr | 2026-08-04 12:32:29.963758+00:00 | succeeded | f82225f95b301c628947c9aff23eb0351b5edd3063e52d183c888cd339d6c045 | 5,683 | null | """One Layer Deeper E2 round-two experiment: Per-example stochastic depth without torch.item GPU synchronisation."""
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_s... | {
"id": "15eff7c4-c18c-4c04-8c31-d2549da76de3",
"created_at": "2026-08-04 12:32:29.963758+00:00",
"db_md5": "8808d42b0b284e6ab0982720b4347652",
"submitter": "c-nnr",
"github_login": "c-nnr",
"run_id": "6cd596d3-d97b-4ac9-aa1f-12b5dc49482a",
"tier": "easy",
"dataset_id": "e2",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.2596105337142944,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008125000167638063
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2660982608795166,
"example_count": 30... |
15f79149-67a6-46d0-a9db-d2d33656f92b | easy | oupadhyay | 2026-08-15 22:16:42.429782+00:00 | succeeded | f3b61bba268d41957c0e1e6f3152d38fbb08811b8e8c3eb8d94f77ba74f49b43 | 3,997 | null | """Conditional optimizer phase 2: lr10_warm64."""
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.voca... | {
"id": "15f79149-67a6-46d0-a9db-d2d33656f92b",
"created_at": "2026-08-15 22:16:42.429782+00:00",
"db_md5": "71842ae21d8d21b2e861c8e83eff9f69",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "8e409c3e-fc12-403b-97e5-0ff3c7fb4c79",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.383123722936949,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2723221468818564,
"example_count": 600,
"ex... |
15fd435c-14ed-436e-adcd-173c6197f51e | easy | himalalps | 2026-08-06 12:09:23.781046+00:00 | succeeded | 4847956c2c897ac00cb00d889770660c8cece919dea285f021741e19100cc696 | 50,245 | null | """Algorithmically constrained latent recurrence for One Layer Deeper.
One masked cross-attention pass creates two independent representations:
``state_0`` from X only and an immutable context from N only. T is excluded
from both representations and is used solely to choose how many times the
learned recurrent MLP tr... | {
"id": "15fd435c-14ed-436e-adcd-173c6197f51e",
"created_at": "2026-08-06 12:09:23.781046+00:00",
"db_md5": "b64f784d9aa381b3543e2c19791c8c13",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "d955a311-30ea-4878-971d-85b2bfbcd6b8",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 8.830445173960012,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007916666883975267
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.942784309387207,
"example_count": 600,... |
15fdca9c-7cb6-44f8-8f5c-db01e7525a07 | easy | Avyukth | 2026-08-11 19:55:14.729639+00:00 | succeeded | e394b6e5092f966caae16c89a246b403b377b916a64bc4f6a173a6034c8d5496 | 13,386 | null | """T-adaptive recurrent Transformer: loop count scales with T read from the prompt.
The benchmark family composes a transition function T times. This model bakes
that structure in: it decodes the integer T from the prompt's digit tokens and
applies one shared block BASE + K*T times. Phase embeddings are cyclic (period... | {
"id": "15fdca9c-7cb6-44f8-8f5c-db01e7525a07",
"created_at": "2026-08-11 19:55:14.729639+00:00",
"db_md5": "b6f632a87d9f79acdfd85c179b260f9f",
"submitter": "Subhrajit Makur",
"github_login": "Avyukth",
"run_id": "0299d7c3-2afc-49cd-8456-ff7d2c6cd1b8",
"tier": "easy",
"dataset_id": "e1",
"status": "su... | {
"score": {
"mean_loss": 9.810303688049316,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.02666666731238365
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 10.556775093078613,
"example_count": 100,... |
1602e63a-73e8-4c4b-b978-261091abcfd1 | easy | pculaf | 2026-08-25 07:34:53.779541+00:00 | succeeded | b93ca9e7007334f2e3677be544b6bf5c4c04856f55e152154e9bebcbede4bf07 | 7,754 | null | """Small single-pass post-RMSNorm Transformer for One Layer Deeper."""
from __future__ import annotations
from collections.abc import Iterable
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
... | {
"id": "1602e63a-73e8-4c4b-b978-261091abcfd1",
"created_at": "2026-08-25 07:34:53.779541+00:00",
"db_md5": "e7b2e63d0abf73d0bceead3c172ed55a",
"submitter": "Pavle Culafic",
"github_login": "pculaf",
"run_id": "28e6fb6e-d3ec-4cc2-805e-70ca58dc5dbc",
"tier": "easy",
"dataset_id": "e6",
"status": "succe... | {
"score": {
"mean_loss": 1.909726619720459,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05855856090784073
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8726625442504883,
"example_count": 60,
... |
16032926-6b40-4b5f-ae82-45801b12400b | easy | KaustubhKumar05 | 2026-08-28 18:19:34.767431+00:00 | succeeded | d4954b5727635a3dc379e5cbde8e30c9f388702617dc3c9c2cde0d172983a611 | 19,564 | null | """w1_k4m -- w1_value at a larger population, batch traded down to pay for it.
K=4,194,304 -- spacing 4.8e-7, essentially AT the correctness basin 3.6e-7 at
N=2047, i.e. the point where coverage alone should be near-exact with no refinement.
Coverage has been the decisive lever so far: 262k -> 1M took contest Easy fro... | {
"id": "16032926-6b40-4b5f-ae82-45801b12400b",
"created_at": "2026-08-28 18:19:34.767431+00:00",
"db_md5": "40f947fa79607895cbfc08ad84c2373e",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "c574a61b-b41d-4904-8969-e9a45c2ff7da",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 0.36170946081620897,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.83125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.35177238993847854,
"example_count": 600,
... |
1606cd00-d6d4-47ec-bccd-51c1d4e5b69d | easy | khushidahi | 2026-08-02 00:54:33.213429+00:00 | succeeded | 41bbf3d39167753777b0a80d8bd1eec3860498d3395cb23e7cd622994f6d3342 | 7,576 | null | """Prelude -> recalled shared core -> coda model for One Layer Deeper.
R2 architecture reset:
- use a real one-block prelude and one-block coda;
- apply one shared recurrent Transformer core four times;
- inject immutable prelude context at every recurrent update;
- use sandwich RMS normalization for stable repeated a... | {
"id": "1606cd00-d6d4-47ec-bccd-51c1d4e5b69d",
"created_at": "2026-08-02 00:54:33.213429+00:00",
"db_md5": "b07bdaf2046ac321dcb5bcb3ca77d31a",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "4ed18501-e1d8-42aa-a2a6-43620e4b3bc0",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.1709060668945312,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000325962901
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1782610416412354,
"example_count": 60... |
1607299e-8c77-4f08-aca2-2940f7d343ae | easy | shan23chen | 2026-08-08 02:02:25.607349+00:00 | succeeded | be8818e200b6d652d00b9f969748e98043ced9f9fcb0cc7421ea45a5c25cae2b | 20,813 | null | """Generic recurrent Transformer with normalized numeric and CRF objectives.
The model is deliberately task-generic: it learns a transition over prompt and
workspace tokens instead of embedding the public modular-squaring algorithm.
A learned linear-chain CRF models dependencies among adjacent output digits;
its endpo... | {
"id": "1607299e-8c77-4f08-aca2-2940f7d343ae",
"created_at": "2026-08-08 02:02:25.607349+00:00",
"db_md5": "4f2695290075ab91887fc664814a90c9",
"submitter": "Shan Chen",
"github_login": "shan23chen",
"run_id": "ffa5fac8-bc5e-4244-8b18-c7dbee1c4d42",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 2.93568754196167,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011250000447034836
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.898953676223755,
"example_count": 600,
... |
160990e0-3d11-42dc-8f14-2e0651f1ad65 | easy | himalalps | 2026-08-06 16:24:51.908905+00:00 | succeeded | 5d8d406eaef223e35b79243276a3741df8d026178cd06d0fc3675cbe0fd75e7c | 64,424 | null | """Algorithmically constrained latent recurrence for One Layer Deeper.
One masked cross-attention pass creates two independent representations:
``state_0`` from X only and an immutable context from N only. T is excluded
from both representations and is used solely to choose how many times the
learned recurrent MLP tr... | {
"id": "160990e0-3d11-42dc-8f14-2e0651f1ad65",
"created_at": "2026-08-06 16:24:51.908905+00:00",
"db_md5": "fb21427279a83e45a9073691b41cf670",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "cd25a6d7-41fa-4166-a4e6-f14b797ae5e3",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 3.9711836025387317,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008750000093132256
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.5695443153381348,
"example_count": 60... |
160d37fb-486d-4aba-9a0e-ee0a0812c83b | easy | neel04 | 2026-08-11 11:38:13.222055+00:00 | succeeded | 25ad08b1795939f8dfd8de3ab14a91e8a9da6940e3285eb086b664e1b30863c4 | 7,549 | null | """Segment-size-one online training with normalized attractor dynamics."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
BatchReuseContext,
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBa... | {
"id": "160d37fb-486d-4aba-9a0e-ee0a0812c83b",
"created_at": "2026-08-11 11:38:13.222055+00:00",
"db_md5": "d2addca3732d3f00cec2ce733f4166e4",
"submitter": "Neel Gupta",
"github_login": "neel04",
"run_id": "4a5a5728-9068-4cd8-b4d0-5bace89cc2fa",
"tier": "easy",
"dataset_id": "e1",
"status": "succeede... | {
"score": {
"mean_loss": 2.1832600831985474,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0833333320915699
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2008047103881836,
"example_count": 100,... |
160fb77b-451b-4e83-9bba-a8154f4eab9f | easy | yashkant | 2026-08-19 02:20:29.517400+00:00 | succeeded | 9902c3faaf0097b2b49155295b205bb154ed7375c7c14c298993e6824ea93011 | 43,303 | null | """Round2213 learns a channel-wise reset decision from fixed context.
The confirmed parent resets both width-8 Dykstra correction memories on
every parsed-T cycle. One shared zero-start 9-to-8 affine gate sees the
immutable pooled input/answer anchor and explicit parsed T each cycle.
Its clipped channel outputs inter... | {
"id": "160fb77b-451b-4e83-9bba-a8154f4eab9f",
"created_at": "2026-08-19 02:20:29.517400+00:00",
"db_md5": "dd303641f17e6c1a3bdc87edc607bb61",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "61bc6c13-803c-42de-887b-3e19471e3385",
"tier": "easy",
"dataset_id": "e5",
"status": "succeed... | {
"score": {
"mean_loss": 2.1453483811755465,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666679084302
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1496733464467686,
"example_count": 60... |
16126764-024e-4d64-81e8-c2310aa9aa14 | easy | ehonig | 2026-08-19 21:48:00.320661+00:00 | succeeded | 9237b002b1a87c176f5572697d2243ebcecf46ef196dd8925a7d1c70c634fc8d | 20,518 | 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": "16126764-024e-4d64-81e8-c2310aa9aa14",
"created_at": "2026-08-19 21:48:00.320661+00:00",
"db_md5": "6cd65ceba855c792c1ac62e2d32589c9",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "48533070-cbf3-419b-902c-3522a4c0891e",
"tier": "easy",
"dataset_id": "e3",
"status": "succ... | {
"score": {
"mean_loss": 2.177429676055908,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009374999441206455
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.182441234588623,
"example_count": 800,... |
1615dd17-908e-49ee-9d20-8399fcee7aa0 | easy | viridale | 2026-08-28 12:40:21.285374+00:00 | succeeded | 25b1883751cd4406d6b3ab706b6f5b66931f5cdfd1f3b0707619174f58c4f0ee | 33,469 | null | """Globally learned quadratic-program evidence slab.
hypothesis: E5's few 10-bit modulus identities repeatedly trigger the same-N branch,
allowing the N-conditioned controller to learn modular aliases: latent evidence falls
below .001 while full-batch hard accuracy stays exactly one bit-setting out of two.
... | {
"id": "1615dd17-908e-49ee-9d20-8399fcee7aa0",
"created_at": "2026-08-28 12:40:21.285374+00:00",
"db_md5": "a81ce9296f55a044e1ab64b02b2dc26e",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "c2db08d6-735a-4dd3-9ee1-74b0731e909f",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 600,
"exact_accuracy": 1.0,
"... |
16164433-92a0-4686-b848-6d93e61fe3a9 | easy | ehonig | 2026-08-14 01:17:15.031588+00:00 | succeeded | 6262315d4c69ed5af0e318d30f72428d3db4e9acc8d402facb3207dcdc5080ee | 52,757 | null | """DigitLoop — a place-value recurrent model for repeated modular squaring.
The task is ``y = x^(2^T) mod N`` with the answer read back tail aligned, one
decimal place value per token position from the right. A stack of attention and
MLP layers has no primitive for the two operations this needs — a partial-product
co... | {
"id": "16164433-92a0-4686-b848-6d93e61fe3a9",
"created_at": "2026-08-14 01:17:15.031588+00:00",
"db_md5": "c2731c9be308dbd2b26ef7ee917bad94",
"submitter": "Edouardo Honig",
"github_login": "ehonig",
"run_id": "3fc7723a-8660-45fe-ae7e-9187a2a90c21",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 5.296449780464172,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.18000000342726707
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.651521682739258,
"example_count": 100,
... |
1616b48b-43b8-4543-a9e3-b1c01cee7fc9 | easy | akkolli | 2026-08-04 03:46:04.870454+00:00 | succeeded | a3237b12d4e3bef7be51a9fa79eeb0deb103fa95fde5e09f0490450263200905 | 17,821 | null | """Dual-format CE-trained bank of exact modular recurrence programs.
The model supports separate-output prompts and causal ANS/result/EOS records.
Answer-token values never enter parsing or recurrence state; causal routing uses
only public marker positions. The random selector is trained by token CE only.
"""
from _... | {
"id": "1616b48b-43b8-4543-a9e3-b1c01cee7fc9",
"created_at": "2026-08-04 03:46:04.870454+00:00",
"db_md5": "48c15d97f9be79619a33c6cc2aafca7a",
"submitter": "Akshay Kolli",
"github_login": "akkolli",
"run_id": "95d34878-d307-4a16-983f-3b5217122b42",
"tier": "easy",
"dataset_id": "e4",
"status": "succe... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 1200,
"exact_accuracy": 1.0,
... |
16172aa9-d077-40a5-9685-d50e242c66ad | easy | Cree0618 | 2026-08-29 06:58:44.939292+00:00 | succeeded | 987cdba8bb1a09b42e34b438388317783602083f87f1ed0b3359aadf9c1abb8d | 4,914 | null | """E008: E002 tied recurrence trained with StableMax cross-entropy.
Single-variable ablation: architecture and AdamW match E002; only the training
loss changes. StableMax uses a logarithmic transform before normalization,
reducing the incentive for already-correct logits to grow without bound.
Evaluation is unchanged ... | {
"id": "16172aa9-d077-40a5-9685-d50e242c66ad",
"created_at": "2026-08-29 06:58:44.939292+00:00",
"db_md5": "d36ec710ec76b2ea616265b723485858",
"submitter": "Cree0618",
"github_login": "Cree0618",
"run_id": "6489b176-27ff-4cc6-8f01-2e02d798fbc5",
"tier": "easy",
"dataset_id": "e5",
"status": "succeede... | {
"score": {
"mean_loss": 16.084232709480453,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333345750967
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 16.323879874875193,
"example_count": 60... |
16184a13-7c7e-4164-ab7e-2276cd9c878b | easy | MarcoF1 | 2026-08-13 19:17:08.902332+00:00 | succeeded | a1c6353084c68921aae2605cb5306e932cff534d64e0f0a811804e81685a74f1 | 14,313 | null | """Residue-clock model for fixed-N generalization.
Representation: parsed integer values of x and N feed (a) exact residue-phase
features over a bank of small integer periods, (b) learned double-hashed value
embeddings (per-residue identity), and T feeds cyclic phase features
(T mod 2..48) plus saturating channels — p... | {
"id": "16184a13-7c7e-4164-ab7e-2276cd9c878b",
"created_at": "2026-08-13 19:17:08.902332+00:00",
"db_md5": "b33cc189bc4163a2b8a14cdd231154f9",
"submitter": "Marco Fleming",
"github_login": "MarcoF1",
"run_id": "0d308645-6f24-46e4-bfb9-59c42a6aa7f3",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 2.4799392945748613,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.04833333214124044
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.601139545440674,
"example_count": 100,... |
1625291e-91f9-47fa-ab86-69683b80c804 | easy | 0Chris5R | 2026-08-08 10:31:45.495719+00:00 | succeeded | 2d7e8fb7138289ce99936b2897754bc3b706765cb5d3bee382350bd6388073ca | 21,401 | 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": "1625291e-91f9-47fa-ab86-69683b80c804",
"created_at": "2026-08-08 10:31:45.495719+00:00",
"db_md5": "726edf70ca897e154299de1db3fc9486",
"submitter": "Chris ",
"github_login": "0Chris5R",
"run_id": "1bb749a9-695b-472b-bcf5-d5feafe1ae0e",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.290900592415994,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.5031847280237174,
"example_count": 600... |
162a67d6-9e3c-494e-af9c-c73445ab50b6 | easy | DDanlov | 2026-08-19 21:19:13.222732+00:00 | failed | 12d31634a3344f4ff59645ee6c23e2990330c9a579bad96f9d8bdcc09b4a92c4 | 17,730 | null | """
Dynamic Submission for One Layer Deeper API Benchmark
Config: dim=1024, num_heads=32, d_ff_mult=4, fixed_trec=2000
"""
from __future__ import annotations
import math
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
... | {
"id": "162a67d6-9e3c-494e-af9c-c73445ab50b6",
"created_at": "2026-08-19 21:19:13.222732+00:00",
"db_md5": "f1e532e5b3d751c4bc1ed8b0d6f5493b",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "bbd38371-90d0-43d3-865a-8d7fb687d112",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
162d654d-d91e-4617-b75e-d36e01857970 | easy | newjordan | 2026-08-07 03:07:55.005010+00:00 | succeeded | e468e9e8a14153b37f757d337aaf7d6e0bce23eb2a522a41667109808bbcf42e | 46,384 | null | """Helix-Crawler — a recursive-weight-shared "crawler" ALU for modular squaring.
The task is r_{i+1} = r_i^2 mod n, chained T times: a *recurrence* over a single
operator. This build embodies the crawler / Frugendorff lineage (spark:~/sota_crawler)
mapped onto that structure. n never enters as a table row (unseen modu... | {
"id": "162d654d-d91e-4617-b75e-d36e01857970",
"created_at": "2026-08-07 03:07:55.005010+00:00",
"db_md5": "dad842ba8e773e326f0d4a0ec602d22a",
"submitter": "Frosty40",
"github_login": "newjordan",
"run_id": "42de9d96-9d1d-4124-beb5-e89d35a05ac9",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 7.094239234924316,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.23333333432674408
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.786130905151367,
"example_count": 100,
... |
1648eb38-a19c-4e24-ac2b-8ef4b19a9cdc | easy | yashkant | 2026-08-09 03:01:24.680943+00:00 | succeeded | d5fde88fe21de21e443530ce520067a660084202845f6e63cc3e645dcbef5720 | 22,836 | null | """Round 211 common TRM-style x/y/z shared-transition template."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
from torch import... | {
"id": "1648eb38-a19c-4e24-ac2b-8ef4b19a9cdc",
"created_at": "2026-08-09 03:01:24.680943+00:00",
"db_md5": "c941879f160765469a08cd6840b8b992",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "3861797b-54c0-4ffb-8c47-bd0e3b8d8e11",
"tier": "easy",
"dataset_id": "e4",
"status": "succeed... | {
"score": {
"mean_loss": 2.06421317443805,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007395833345750967
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0667295174391995,
"example_count": 1200... |
164bb785-dea5-43be-b2c3-2a84a45015a8 | easy | khushidahi | 2026-08-20 22:35:28.006137+00:00 | succeeded | 2c5f1aa1227cd86dc6408ec98bc457762a24b0b7feed235d60ad691a7ad88086 | 32,376 | null | """R320: Hard-aligned balanced causal functional-root candidate.
This candidate is deliberately diagnostic. It trains on evaluator-owned final
labels at several depths so that one shared transition must compose. The model
never converts N or X to native numbers and never implements multiplication or
modular reductio... | {
"id": "164bb785-dea5-43be-b2c3-2a84a45015a8",
"created_at": "2026-08-20 22:35:28.006137+00:00",
"db_md5": "2ff6ef9baa42b1407fccf7a1fdc163d0",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "6ca05c16-45ec-4ec7-b401-3f1e040832c2",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.3686421264437207,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0054166666790843015
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3601732296793987,
"example_count": 6... |
164f06c6-16b0-4bdc-a3f6-ecc122443d0b | easy | DDanlov | 2026-08-10 08:32:27.299726+00:00 | succeeded | 8457831b04b08c80ba8bafd1958d196c020429023c58ce93fc00d8baf6aebd21 | 16,674 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
def init_cutting_the_skip(module):
"""
SVD-based initialization to ensure initial Jaco... | {
"id": "164f06c6-16b0-4bdc-a3f6-ecc122443d0b",
"created_at": "2026-08-10 08:32:27.299726+00:00",
"db_md5": "c879f1c875452a32ba34e8d07ffc366b",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "342606de-770e-421e-933e-50b116a2c0c1",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 4.253035924083871,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.026666666219631832
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 4.249160289764404,
"example_count": 100,... |
1654225e-7af9-4b0f-b1eb-9efbf9fac96f | easy | nikolageorgiev2000 | 2026-08-24 09:46:28.855211+00:00 | succeeded | c2757380590c01e55acf0f70b11d5e994683d5b9e252410056362d05e4801f97 | 12,381 | null | """Masked semantic residue simplex with a shared variable-modulus transition."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,... | {
"id": "1654225e-7af9-4b0f-b1eb-9efbf9fac96f",
"created_at": "2026-08-24 09:46:28.855211+00:00",
"db_md5": "72b06d5dcc848ca49584d2edd9f1312d",
"submitter": "Nikola Georgiev",
"github_login": "nikolageorgiev2000",
"run_id": "af111c58-0423-4ef8-bf6d-b6b025a66ba8",
"tier": "easy",
"dataset_id": "e7",
"s... | {
"score": {
"mean_loss": 0.27839144319295883,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9713903963565826
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.13685698807239532,
"example_count": 85... |
16564a95-e672-4b03-b3b7-cbf96f88ae9d | easy | shirvani-jr | 2026-08-31 01:44:03.867306+00:00 | succeeded | f36b045c50800cbdceef864d677daf420a793727133eaedab87b1505981fd664 | 21,222 | null | from __future__ import annotations
import math
import time
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_state,
)
PERIOD_MINIMUM = 3
PERIOD_MAXIMUM = 100
EX... | {
"id": "16564a95-e672-4b03-b3b7-cbf96f88ae9d",
"created_at": "2026-08-31 01:44:03.867306+00:00",
"db_md5": "52d288f37eb9ac99c594e70a5f2357e8",
"submitter": "Ali",
"github_login": "shirvani-jr",
"run_id": "20dac411-4ef0-4040-b0f3-f70839cccf0d",
"tier": "easy",
"dataset_id": "e9",
"status": "succeeded"... | {
"score": {
"mean_loss": 0.7099649533629417,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9155555665493011
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.35706627368927,
"example_count": 90,
... |
16574bf5-f8e1-432e-b245-e30875cdf15d | easy | Aaquib111 | 2026-08-05 04:24:49.525715+00:00 | succeeded | d3323ae22ad049feebdeed092e42732373afe9a457619a759b1f47500b551da9 | 8,463 | null | """Input-recalled recurrent Transformer with randomized train-time depth."""
from __future__ import annotations
import random
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
... | {
"id": "16574bf5-f8e1-432e-b245-e30875cdf15d",
"created_at": "2026-08-05 04:24:49.525715+00:00",
"db_md5": "6ddab633e32411d56bda637691cff625",
"submitter": "Aaquib Syed",
"github_login": "Aaquib111",
"run_id": "eb30943f-8f92-4ae8-9598-82df72014b68",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 6.696669017438917,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006250000024835269
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.670586036459747,
"example_count": 600,... |
1658851b-59b1-43c1-b045-6e7d6c7a1acc | easy | khushidahi | 2026-08-03 03:00:34.648964+00:00 | succeeded | 3c376f50a7c6e072ecc55a877ebadbda986e6f08342124dd5d486167815a822b | 15,252 | 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": "1658851b-59b1-43c1-b045-6e7d6c7a1acc",
"created_at": "2026-08-03 03:00:34.648964+00:00",
"db_md5": "9b27ed17d540fed6c2c0ab41ced83653",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "5d75005a-cde0-4152-a238-82a0960a9e7f",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.200993061065674,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01083333371207118
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1985843181610107,
"example_count": 600,... |
166915f0-0ac7-4e6f-955f-6eafd729bb11 | easy | int3 | 2026-08-18 12:37:47.729461+00:00 | succeeded | 4f80168e3c089d7e12cb79b1e0f5c9323080a7f2389a4b9224826a226532cf86 | 12,185 | null | """Automata bank with bilinear pair-table readout and deep state supervision.
Architecture (all predictive computation in learned weights):
- The prompt's N / X / T decimal fields are parsed to integers (discrete index
math, the same class as T-conditioned loop counts / adaptive computation).
- State: one distributi... | {
"id": "166915f0-0ac7-4e6f-955f-6eafd729bb11",
"created_at": "2026-08-18 12:37:47.729461+00:00",
"db_md5": "5dcf7664ce49ce9468c223c1ee009ae3",
"submitter": "Jez Ng",
"github_login": "int3",
"run_id": "60fd6724-8f6c-4eea-a4b2-fd69a2ed2b5b",
"tier": "easy",
"dataset_id": "e7",
"status": "succeeded",
... | {
"score": {
"mean_loss": 0.43362991511821747,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.3795228600502014,
"example_count": 85,
"exa... |
1671b153-280b-4e45-8f62-32bf691a0cd9 | easy | KaustubhKumar05 | 2026-08-26 10:13:30.286351+00:00 | succeeded | ff3c24b119d8b41247f7a23b06e5f884d54976a083f23d920d60970370615026 | 13,328 | null | """sq_reduce — candidate A: learn to SQUARE, then conditionally reduce.
Motivated by FINDINGS 24. Trivial-predictor analysis showed the 0.78% leaderboard
cluster (ranks 4-10) is reproduced exactly by "compute x*x and never reduce mod N":
it wins precisely on {x : x^2 < N}, scoring 6/1280 at T=1 and 0 at every deeper
r... | {
"id": "1671b153-280b-4e45-8f62-32bf691a0cd9",
"created_at": "2026-08-26 10:13:30.286351+00:00",
"db_md5": "4fdbfdd9cea5e06314b1dfd2426ec716",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "d0732cb5-69a1-4a21-8b86-48c3edf45f21",
"tier": "easy",
"dataset_id": "e6",
"status": "succee... | {
"score": {
"mean_loss": 2.0073383152484894,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.5927928239107132
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.9266133904457092,
"example_count": 60,
... |
16845167-1c99-499e-94d6-5306a0f136e0 | easy | repst | 2026-08-22 23:28:36.237494+00:00 | succeeded | 2c7861bb2c63cf2d98e214e0ceba7501c6a7945257f530b6fb91cbf177cc268f | 27,550 | null | # R529: streaming-MDL grammar-frontier recurrent machine.
#
# The R525 factor graph and exact recurrent interpreter are retained, including
# their small dense associative memory for direct credit. Expression syntax is
# now streamed in minimum-operation order and stopped at a semantic compute
# budget, rather than ma... | {
"id": "16845167-1c99-499e-94d6-5306a0f136e0",
"created_at": "2026-08-22 23:28:36.237494+00:00",
"db_md5": "c128bfd6cace8a3bb540d49c5a1f15ad",
"submitter": "Ryan Epstein",
"github_login": "repst",
"run_id": "701a4eaf-9791-46e1-8a19-19a01131181f",
"tier": "easy",
"dataset_id": "e1",
"status": "succeed... | {
"score": {
"mean_loss": 0.0004790582024725154,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.00046960674808360636,
"example_count": 100,
... |
168600c2-1adb-41e7-85d6-127d518df9d1 | easy | DDanlov | 2026-08-26 19:32:53.395198+00:00 | succeeded | 314fef195ebd267bb29718b2eeb6c161f86cf74036b09e8b333da34d7ad46e8f | 31,773 | null | import math
import time
from dataclasses import dataclass
from typing import Optional, Tuple, Dict, Any, List, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from onelayerdeeper.api import Submission, OptimizerBundle, ModelSpec, OptimizerSpec
except ... | {
"id": "168600c2-1adb-41e7-85d6-127d518df9d1",
"created_at": "2026-08-26 19:32:53.395198+00:00",
"db_md5": "6fe0a3cfd44d943b3440c80a881db085",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "4742b578-d4d0-4b3f-8a83-b7c6d670e87a",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 3.684468692125024,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.002083333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.6840024411411028,
"example_count": 600... |
1687c7ed-4f84-484a-b33d-302bb363a67d | easy | AFeinfield | 2026-08-28 02:30:44.470985+00:00 | succeeded | f7bac9d1bd9eb9489f2bf0ebc6ba4f47b3d04d59185913117c0f4905d1c96d71 | 10,286 | null | """181: 180 + RESIDUE-SPACE SUPERVISION of F (the literature setting).
Local probe: F_map_acc stayed 0.000 while loss fell — the wide digit
head shortcuts around F. Fix: re-encode provided label digits into a
residue index inside the loss and add CE on F's own-T logits — direct
map supervision, exactly how published g... | {
"id": "1687c7ed-4f84-484a-b33d-302bb363a67d",
"created_at": "2026-08-28 02:30:44.470985+00:00",
"db_md5": "982b958b819ec3e4de109a367242d892",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "a1e5fe25-a5be-4d59-9d08-f7a26a629a21",
"tier": "easy",
"dataset_id": "e1",
"status": "succ... | {
"score": {
"mean_loss": 1.2400842308998108,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.9033333361148834
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.1178820133209229,
"example_count": 100,... |
169079bb-8bc6-47c3-acad-5b418ca6613d | easy | viridale | 2026-08-02 09:56:01.332181+00:00 | succeeded | 47454763eedce78565fc0b8c053a439542d52ccf7ca3f38cc6141e3132aa977a | 30,747 | null | """Synchronization-free exact-T=1 active transition learning.
hypothesis: v088 preserves every semantic gate and official 64/64 result, but
its optimizer calls .item() on the table-scale gradient every step to skip
mathematically zero Adam updates. That GPU synchronization lowers m1 from
v087's 30,515 to 2... | {
"id": "169079bb-8bc6-47c3-acad-5b418ca6613d",
"created_at": "2026-08-02 09:56:01.332181+00:00",
"db_md5": "44c8d1dec39d40d6d5a92bdb1f7f0256",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "3dd69371-1e69-4edc-ac88-5aa765038525",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 0.003014546214626436,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.00301454615599774,
"example_count": 800,
"... |
1697296e-f4dd-46a0-94bc-72b03a1845a5 | easy | KaustubhKumar05 | 2026-08-29 05:09:47.482234+00:00 | succeeded | 465e1261fbbafa405f20e04324eac584b24d298b5cb9e19bc524f212d16ec1d5 | 15,252 | null | """op2_t1w -- stock cell_op2 + T=1 loss upweighting (the scored rung).
Forked from cell_op2. ORIGINAL DOC BELOW.
cell_op2 -- ONE shared squaring operator with internal depth, applied T times.
WHY THIS SHAPE (verified on the real data, not assumed):
answer(T=t) == x squared t times mod N 20000 / 20000 e... | {
"id": "1697296e-f4dd-46a0-94bc-72b03a1845a5",
"created_at": "2026-08-29 05:09:47.482234+00:00",
"db_md5": "3f149399b3699343f2f5e891ad556b00",
"submitter": "koz",
"github_login": "KaustubhKumar05",
"run_id": "80a6ae5f-86db-4ade-a302-c4c4b2f94214",
"tier": "easy",
"dataset_id": "e5",
"status": "succee... | {
"score": {
"mean_loss": 2.632482326376908,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333358168601
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.6117962462485105,
"example_count": 600... |
169d6ac6-6404-4e95-b308-4a3a4cbcda13 | easy | sapient-sapiens | 2026-08-15 08:17:54.480329+00:00 | succeeded | cfa2d5ad3daf9c1fbb1288b322672a7bc1a400c5c9d65e165862e656026b9d9e | 11,188 | null | """Narrow/deep looped Transformer with optional latent workspace tokens.
Research candidate for sampled-modulus fixed-T=2. The encoder builds immutable
prompt memory. A stack of learned operator blocks updates a mutable latent
stream, and the whole stack is reused for each task recurrence. Optional blank
workspace ... | {
"id": "169d6ac6-6404-4e95-b308-4a3a4cbcda13",
"created_at": "2026-08-15 08:17:54.480329+00:00",
"db_md5": "73e276376bb5710a1ed15990e994132d",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "94e80b6d-1611-4db8-886c-5cc433bf1796",
"tier": "easy",
"dataset_id": "e4",
"status": "su... | {
"score": {
"mean_loss": 2.0716750353695006,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.003125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.082343257061569,
"example_count": 1200,
... |
16b3d275-0db9-4833-aa1b-1135fd4ac694 | easy | oupadhyay | 2026-08-14 15:36:16.286524+00:00 | succeeded | 43a50cca3e3adeb51c05d67d3acde8e513fac50ea5d8f6acebb730bd93cb3dd3 | 7,313 | null | """Dual-stream N-conditioned recurrent transducer for Easy E3."""
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
PAD, N_MARK, X_MARK, T_MARK, DIGIT_BASE = 0, 2... | {
"id": "16b3d275-0db9-4833-aa1b-1135fd4ac694",
"created_at": "2026-08-14 15:36:16.286524+00:00",
"db_md5": "ae2941c1407fe06ea4d897237b7ac712",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "2d1e2831-4bee-4bba-86f8-344281844113",
"tier": "easy",
"dataset_id": "e3",
"status": "s... | {
"score": {
"mean_loss": 3.2030380816497224,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.008125000018626451
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.1668072332591635,
"example_count": 80... |
16b57898-2373-48d2-8855-a26a6abddd1f | easy | karanganesan | 2026-08-06 03:36:52.381186+00:00 | succeeded | d6fd4b484846bf39c009f2ba2eed0ffbc0f70adce010c638822c0d4175ff88db | 23,793 | 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": "16b57898-2373-48d2-8855-a26a6abddd1f",
"created_at": "2026-08-06 03:36:52.381186+00:00",
"db_md5": "89bf4129e7fe8a311135fa95ff278b20",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "a0b5b537-575e-4fbb-881b-26e7b3565567",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 9.490567684173584,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03166666720062494
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.540064811706543,
"example_count": 100,
... |
16c74e01-7492-4257-8821-025d7d5f2924 | easy | DDanlov | 2026-08-16 21:21:32.026812+00:00 | succeeded | f0e401937a11ec95eea9869879b93ced5ba3d7468b913ba8432527a2348d9ae3 | 16,633 | null | """
One Layer Deeper Competition Submission
================================================================================
Architecture: Universal Harmonic Recurrent Transformer (Pure No-Norm No-Skip DKS)
- Scale-Free Continuous Harmonic Coordinates spanning multiple octaves (omega_j in [1.0, 1024.0])
- Dynamic Rotar... | {
"id": "16c74e01-7492-4257-8821-025d7d5f2924",
"created_at": "2026-08-16 21:21:32.026812+00:00",
"db_md5": "34b5246f5da9e978fa4c45e9dac68cbb",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "44b713f3-20ce-4096-9ad1-7798241ab9bb",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.292065222869641,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0141666666790843
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.295728618120399,
"example_count": 600,
... |
16c7ecba-eba2-43fe-acbb-c960a8b31b5e | easy | himalalps | 2026-08-06 14:42:56.123719+00:00 | succeeded | 76a9371266e7d1517bd5591a44f55b6b864deb3eb7bc87b868ecc8f519f04d7a | 59,499 | null | """Algorithmically constrained latent recurrence for One Layer Deeper.
One masked cross-attention pass creates two independent representations:
``state_0`` from X only and an immutable context from N only. T is excluded
from both representations and is used solely to choose how many times the
learned recurrent MLP tr... | {
"id": "16c7ecba-eba2-43fe-acbb-c960a8b31b5e",
"created_at": "2026-08-06 14:42:56.123719+00:00",
"db_md5": "9f396ac99dad2905128cbaea40161a07",
"submitter": "Haoyu Tang",
"github_login": "himalalps",
"run_id": "ad644fa6-f0f3-4af5-bb92-5e4746d7e0e9",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 6.3916066582943944,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.009583333563059567
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.034897327423096,
"example_count": 600... |
16d1cdf6-39ff-46d7-9964-eb5530c9a75a | easy | yashkant | 2026-08-14 15:39:44.506584+00:00 | succeeded | 619287a9c20cfcbcbc9cf70b507a01807ca0fdc6f7b15b85b34cffb853713e20 | 32,372 | null | """Round1780 contrastive centering of learned Dykstra set codes.
The proven Round1766 W8/D2 carrier is retained. Each projection computes the
shared mean of the two learned set codes and its named code's contrast from
that mean. The candidate feeds the contrastive code into the shared projection
network, making A/B ... | {
"id": "16d1cdf6-39ff-46d7-9964-eb5530c9a75a",
"created_at": "2026-08-14 15:39:44.506584+00:00",
"db_md5": "3a728bdeb40a44afd74a4739f742a879",
"submitter": "Yash Kant",
"github_login": "yashkant",
"run_id": "14273990-1299-4924-be77-73bd053399ec",
"tier": "easy",
"dataset_id": "e3",
"status": "succeed... | {
"score": {
"mean_loss": 2.1216499117566636,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01875000004656613
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1206259776626157,
"example_count": 800... |
16dd02da-472c-4bd0-9b4f-4bd670ccb276 | easy | oupadhyay | 2026-08-15 22:01:53.317994+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": "16dd02da-472c-4bd0-9b4f-4bd670ccb276",
"created_at": "2026-08-15 22:01:53.317994+00:00",
"db_md5": "290d0c8bb1faeccfb53e7b35639151cc",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "12afa3c7-4ae3-4218-9394-5fc27c12c38a",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.171539370837582,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005833333345750967
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.146068888394822,
"example_count": 600,... |
16f6a0c9-0e93-4a41-9b97-487d845e2ebd | easy | gauravmishra | 2026-08-10 03:53:25.721527+00:00 | succeeded | bd841cffbcf21dd9a30a3a830fc32b64e07a664efa8cfe0c79c189a849b4b186 | 12,546 | null | """Stage-1 information-gain K4 representation candidate."""
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,
)
POSITI... | {
"id": "16f6a0c9-0e93-4a41-9b97-487d845e2ebd",
"created_at": "2026-08-10 03:53:25.721527+00:00",
"db_md5": "b740b100bece9070971479bf51546e53",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "f5ef09bf-b807-48ae-a815-3881b4cc8a4a",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 7.228999455152813,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01000000002483527
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.158620214248452,
"example_count": 600,
... |
16f99cbe-2f4f-40f4-ae77-0cb753d9aa05 | easy | erdavis0 | 2026-08-25 01:50:19.183544+00:00 | succeeded | 6a159a154131e895c8eb2487b66c205def59d80d58fa77e0e333c0b7323c43bf | 24,075 | null | """Matched categorical-boundary panel: C0 continuous SEQ-CE control.
Public delimiter syntax is used only to route opaque token identities onto
separate right-aligned tapes. One T-blind learned macro maps a categorical X
tape and a digitwise N tape to a new categorical tape through freshly reset
interaction, carry, a... | {
"id": "16f99cbe-2f4f-40f4-ae77-0cb753d9aa05",
"created_at": "2026-08-25 01:50:19.183544+00:00",
"db_md5": "a33c7a23d21e22630f1e654c27cfd4ab",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "96695070-b54f-4098-810c-e4380d7179b1",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 1.8552555441856384,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06036036275327206
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8331050872802734,
"example_count": 60,... |
1703bb68-bd41-4275-beed-92ac28a49257 | easy | jordanrubin | 2026-08-02 20:35:34.550163+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": "1703bb68-bd41-4275-beed-92ac28a49257",
"created_at": "2026-08-02 20:35:34.550163+00:00",
"db_md5": "8424ff3c2b53438cdfeb088f9343f339",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "ec8e2762-00c6-458c-98f7-7e0f09d88842",
"tier": "easy",
"dataset_id": "e2",
"status": "s... | {
"score": {
"mean_loss": 3.087649010690744,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.2870833274597923
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 5.179686799974253,
"example_count": 300,
... |
170833f5-e2fb-4e51-aaa5-074b69c1bb1d | easy | DDanlov | 2026-08-29 12:47:30.179793+00:00 | succeeded | 29d0ac7a8bce5003b73d479bfd61a532e9488edf2bf3f21b24a5a8d0a4011c36 | 23,746 | null | """
trial_03_wide_latent.py
Trial 3: High-Capacity Wide Latent Engine (D=384, H=12, d_ffn=1536) with Symplectic Verlet (K=10).
"""
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.fu... | {
"id": "170833f5-e2fb-4e51-aaa5-074b69c1bb1d",
"created_at": "2026-08-29 12:47:30.179793+00:00",
"db_md5": "81f19fc44ba5f2cd5d2f8dfff7060cd8",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "99f80b03-52d5-4cae-b845-1e7b65b20f5c",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.2379315825574952,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2476904909439686,
"example_count": 600,
"... |
170a7803-d5d6-4aaa-b7d9-6e53da425d08 | easy | DDanlov | 2026-08-19 19:46:30.232742+00:00 | succeeded | 5eebf92b7febad536744ad010a1ad7da8aa37504f1796a558ffda84402a02035 | 17,727 | null | """
Dynamic Submission for One Layer Deeper API Benchmark
Config: dim=1696, num_heads=32, d_ff_mult=4, fixed_trec=112
"""
from __future__ import annotations
import math
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
... | {
"id": "170a7803-d5d6-4aaa-b7d9-6e53da425d08",
"created_at": "2026-08-19 19:46:30.232742+00:00",
"db_md5": "727d345cb72cec9b277c32edd81835fc",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "c08a90ad-a0c8-446b-8a79-ef18a54b7d00",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.378657912578051,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012083333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.44707888618713,
"example_count": 600,
... |
17148ba6-c63b-42a3-a9a3-63a7cdac1150 | easy | karanganesan | 2026-08-07 07:17:53.399813+00:00 | succeeded | 428fb1d9e4bbf76f20b89e2d72968079c2d73ef30df225a854dbfba0a342435c | 25,600 | null | """Parametric looped-transformer family (P1).
One weight-tied transformer block applied k times in latent space. Config
flags cover four P1 families with one file:
looped recall=0 gated=0 tfilm=0 plain weight-tied loop
looped-recall recall=1 re-inject the input embedding each
... | {
"id": "17148ba6-c63b-42a3-a9a3-63a7cdac1150",
"created_at": "2026-08-07 07:17:53.399813+00:00",
"db_md5": "d7f8c91c19835ec15284f898056cb16f",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "f1c78f3b-b045-4e9e-9d5f-b18b11679aa1",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 6.074803352355957,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004999999888241291
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 6.519436359405518,
"example_count": 600,... |
1717c616-52cf-4bc3-92b2-ccc18fbea869 | easy | DDanlov | 2026-08-09 00:24:40.580829+00:00 | succeeded | ce054471391d89652f4d62ce463901380017c2202dc549bcefa3674569b26a52 | 15,978 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class RohanShampoo(Optimizer):
"""Self-contained RohanShampoo optimizer with eigenvalue ma... | {
"id": "1717c616-52cf-4bc3-92b2-ccc18fbea869",
"created_at": "2026-08-09 00:24:40.580829+00:00",
"db_md5": "cdf0a0613dfadfee4470decbcf01ee24",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "c038a7ce-29b0-4a30-92a1-667900c326ec",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 34.262760029172064,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.014999999888241291
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 40.05268478393555,
"example_count": 100... |
1719effb-b437-40a8-86dd-52efdbae48a3 | easy | karanganesan | 2026-08-05 11:28:12.307727+00:00 | succeeded | 89b1719513e10388c97d2d253a5750ed2cdda38f540dad031e7bb3d6bf74f064 | 22,098 | 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": "1719effb-b437-40a8-86dd-52efdbae48a3",
"created_at": "2026-08-05 11:28:12.307727+00:00",
"db_md5": "fef0820593bd1090a3ecf4675feecd78",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "796ef6ae-01c1-40b9-8982-07809d620f1b",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 10.722496509552002,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.011666666716337204
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 11.361918449401855,
"example_count": 10... |
17203513-aa9e-4d51-8463-f98f44ab2e35 | easy | sirish-gambhira | 2026-08-10 01:20:46.850327+00:00 | succeeded | 7bf30e19d90e7f6cadb4ad67d0b2cd51634a4fd2d29a507be6f840a0dd78f8c1 | 3,628 | null | """Basic single-pass Transformer with PyTorch AdamW."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL = 128
NUM_HEADS = 4
... | {
"id": "17203513-aa9e-4d51-8463-f98f44ab2e35",
"created_at": "2026-08-10 01:20:46.850327+00:00",
"db_md5": "eaac69fe345f82973fcf392c504a40ff",
"submitter": "Sirish Gambhira",
"github_login": "sirish-gambhira",
"run_id": "11f8594b-7c6f-4887-8bbf-4143b516ed8e",
"tier": "easy",
"dataset_id": "e5",
"stat... | {
"score": {
"mean_loss": 3.180167063296838,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 3.3679325142248864,
"example_count": 600... |
17251f22-27de-4f49-bfbe-1b17c91ea60b | easy | c-nnr | 2026-08-04 11:07:51.945327+00:00 | failed | bbe21690c82dbefac4a69e96fedefc0847bd72530449105beecc327b9081cee1 | 10,909 | null | """Tiny Recursive Model submission for One Layer Deeper Easy tiers.
The model keeps three sequence-shaped states:
x: an immutable encoding of the prompt
z: a latent reasoning state
y: an iteratively refined answer state
A single shared Transformer cell updates z several times and then y once.
Training uses fewe... | {
"id": "17251f22-27de-4f49-bfbe-1b17c91ea60b",
"created_at": "2026-08-04 11:07:51.945327+00:00",
"db_md5": "be042075346dbc1e5e46af18c6296954",
"submitter": "c-nnr",
"github_login": "c-nnr",
"run_id": "a1270348-2e56-48ed-b417-0601894454df",
"tier": "easy",
"dataset_id": "e1",
"status": "failed",
"st... | null
|
172facd4-73c7-4662-b70a-363d9d63f2cc | easy | IsaacNoel | 2026-08-13 02:45:17.134770+00:00 | succeeded | 24751229b1965f2dc407b66e485b681e0a4d01a3542453812a5d8f7ca69db30d | 10,576 | null | """Recurrent digit machine: one shared block applied exactly T times.
The prompt is parsed into two right-aligned digit registers — the running value
of x and the modulus N — each tagged with place value and variable identity.
One shared block maps ``(x, N)`` to a new x and is applied T times. Its output
is snapped b... | {
"id": "172facd4-73c7-4662-b70a-363d9d63f2cc",
"created_at": "2026-08-13 02:45:17.134770+00:00",
"db_md5": "c1e5a779df288534f08a3b90e5fd8925",
"submitter": "Isaac Noel",
"github_login": "IsaacNoel",
"run_id": "0f1b1254-9b49-418b-9da6-2ede7a998b6a",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.921475350856781,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.038333332166075706
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.8687976598739624,
"example_count": 100... |
173120d5-0700-4bf0-bcea-33b60c62f1b1 | easy | karanganesan | 2026-08-04 00:24:10.367839+00:00 | succeeded | fe34ae944919aaf38db99c75974df7803b880dfae0f5454822fe85864f7e8dcf | 8,612 | 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": "173120d5-0700-4bf0-bcea-33b60c62f1b1",
"created_at": "2026-08-04 00:24:10.367839+00:00",
"db_md5": "071b69ca4a8f05256511a0a1633eee0c",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "d1977adb-d81b-4337-9b63-1a42a06a18fc",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.0718289613723755,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.016666666604578495
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.0443766117095947,
"example_count": 10... |
17411bce-1e6e-441f-94e0-535aee0e191c | easy | Laz4rz | 2026-08-07 22:00:13.199026+00:00 | succeeded | d2c986f59da9d6e123af85f70737b7f681bc8615c08b583e174c4607f884f57f | 27,682 | null | """Independent Loopholing DLMS submission entrypoint.
Generated from the self-contained inspector split. This module does not import
or select the other implementation at runtime.
"""
from __future__ import annotations
import math
from typing import NamedTuple
import torch
import torch.nn.functional as F
from torch... | {
"id": "17411bce-1e6e-441f-94e0-535aee0e191c",
"created_at": "2026-08-07 22:00:13.199026+00:00",
"db_md5": "dfc41ce8636be50213057740c6b039d5",
"submitter": "Laz4rz",
"github_login": "Laz4rz",
"run_id": "dbb13059-1190-42d4-93b9-0d36f23af17d",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.802141547203064,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.019999999552965164
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8170366287231445,
"example_count": 100... |
17434863-f394-47f2-9203-0f78b5453cd7 | easy | CodeReclaimers | 2026-08-15 04:57:32.522639+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": "17434863-f394-47f2-9203-0f78b5453cd7",
"created_at": "2026-08-15 04:57:32.522639+00:00",
"db_md5": "fc08ccee7087a70aeefaaa03c48367d2",
"submitter": "CodeReclaimers",
"github_login": "CodeReclaimers",
"run_id": "caa4a279-125c-4e76-a838-0b4c5702bad6",
"tier": "easy",
"dataset_id": "e6",
"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": 60,
"exact_accuracy": 1.0,
"c... |
17452dfc-46c8-4a5b-8906-5b02245eb4b0 | easy | erdavis0 | 2026-08-22 02:08:58.925358+00:00 | succeeded | 19038191486b389bca73abae38e37218f120f288e18de2ef93b525dab864dba8 | 5,372 | null | """Second-order positive-feature linear attention."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
WIDTH = 64
HEADS = 4
EXPANSION ... | {
"id": "17452dfc-46c8-4a5b-8906-5b02245eb4b0",
"created_at": "2026-08-22 02:08:58.925358+00:00",
"db_md5": "d0f6481a29fcf3f19730b9386572efad",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "b65c7b6c-0f2e-49b0-ab5a-812dc55ea851",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 8.041155338287354,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0533783829305321
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 7.442058563232422,
"example_count": 60,
... |
174a7d5f-0087-4677-9efc-639d32bde581 | easy | oupadhyay | 2026-08-15 22:15:23.714657+00:00 | succeeded | d1bdbac1d0c1f1ff08df85bc75dc73ece9f423bbae698626a8390eb2462a9e86 | 3,997 | null | """Conditional optimizer phase 2: lr10_warm16."""
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.voca... | {
"id": "174a7d5f-0087-4677-9efc-639d32bde581",
"created_at": "2026-08-15 22:15:23.714657+00:00",
"db_md5": "d408b0ab1ad8087efc6a60b62df537dd",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "80174902-9237-478e-b57d-2bf7eee97d83",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.173934353505799,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.004166666691501935
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.146421205837096,
"example_count": 600,... |
174cfc9b-828f-4c01-8b48-45da8130accd | easy | DDanlov | 2026-08-08 10:24:41.022466+00:00 | succeeded | c9cc2a6293f4861a66738294793809c7dce1fd8534b2b063f7a40e8f63bf0ddd | 15,561 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class RohanShampoo(Optimizer):
"""Self-contained RohanShampoo optimizer with eigenvalue ma... | {
"id": "174cfc9b-828f-4c01-8b48-45da8130accd",
"created_at": "2026-08-08 10:24:41.022466+00:00",
"db_md5": "add31ca3022aa8ac444e9948520e428f",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "bea92a88-231b-418a-9d35-cb788e814ff7",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 74.37129872199844,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 74.41135635290445,
"example_count": 600,
"exact... |
17505212-5d2d-43eb-833c-f16259ca80c1 | easy | AFeinfield | 2026-08-18 23:25:03.513418+00:00 | succeeded | 6c5f7c4bad49ab564201978e76676988e06915c5cfb4e82554b5fa645ac428ab | 13,527 | null | """073: BILINEAR multiply + K=5 agreement heads.
The multiply sublayer becomes W((Ah)*(Bh)) — a native product primitive.
Digit products and angle-doubling are both bilinear; linear+pointwise
stacks can only approximate them piecewise. Base: 065 champion.
Original 065 notes follow.
Looped factored block (055) + K=5 o... | {
"id": "17505212-5d2d-43eb-833c-f16259ca80c1",
"created_at": "2026-08-18 23:25:03.513418+00:00",
"db_md5": "f9b9e82f3ee1df05a0e3ebbe8048fd1b",
"submitter": "AFeinfield",
"github_login": "AFeinfield",
"run_id": "6b4f29e1-c2a0-454f-bdf8-7033b5b225db",
"tier": "easy",
"dataset_id": "e2",
"status": "succ... | {
"score": {
"mean_loss": 11.779081344604492,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006041666609235108
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 12.989192008972168,
"example_count": 30... |
17511288-a382-4418-8083-75fb21e8d2ef | easy | c-nnr | 2026-08-04 13:28:03.141667+00:00 | succeeded | afe83584883674d83791684afda0011605b2a62e8967d01903973f56d8c63113 | 6,843 | null | """One Layer Deeper E1 final-15 experiment: Uncalibrated AdamW with warmup and cosine decay to avoid the late E2 memorisation phase."""
from __future__ import annotations
import math
import torch
import torch.nn.functional as F
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
... | {
"id": "17511288-a382-4418-8083-75fb21e8d2ef",
"created_at": "2026-08-04 13:28:03.141667+00:00",
"db_md5": "33a25d2f80d502262e33fbbaaed8e568",
"submitter": "c-nnr",
"github_login": "c-nnr",
"run_id": "b116d726-a8f5-4d8d-90a8-40361eda0511",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded",
... | {
"score": {
"mean_loss": 2.253780961036682,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.006666666828095913
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.300870656967163,
"example_count": 100,... |
175bc4fd-509c-42c0-8a97-e8c24a5d53b2 | easy | adambrzosko | 2026-08-30 16:24:42.934785+00:00 | succeeded | 83c3692179e7ca53ccc37fcf55881f351e964e4dfd682319d31cf83834a7c104 | 13,477 | 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": "175bc4fd-509c-42c0-8a97-e8c24a5d53b2",
"created_at": "2026-08-30 16:24:42.934785+00:00",
"db_md5": "275879d29e00493d7eaf939645f66031",
"submitter": "Adam Brzosko",
"github_login": "adambrzosko",
"run_id": "ea56b483-d461-412f-8efb-b56fb5e6d077",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.4770520461188714,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007083333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4660581580191985,
"example_count": 60... |
17639dfe-2a75-4855-94c4-01be31f60020 | easy | DDanlov | 2026-08-13 22:39:59.990345+00:00 | failed | 7297c8773e46e5209668fac60ad9a99038c3d25a2a2c2d9061a59967da74c7c6 | 22,143 | null | import torch
import torch.nn as nn
import torch.nn.functional as F
import math
from dataclasses import dataclass
from typing import Optional
from torch.optim import Optimizer
try:
from client import Submission, assert_model_state, OptimizerBundle
except ImportError:
Submission = None
assert_model_state = N... | {
"id": "17639dfe-2a75-4855-94c4-01be31f60020",
"created_at": "2026-08-13 22:39:59.990345+00:00",
"db_md5": "a6862fc3e25614109bde491876eb0056",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "1b7ea200-0188-4476-a9f1-e7c76caa907e",
"tier": "easy",
"dataset_id": "e5",
"status": "failed",
... | null
|
1763e173-b095-435f-a5ba-7814633a61c4 | easy | viridale | 2026-08-02 21:03:52.933373+00:00 | succeeded | fc2ce47faf4f935fa4df29dcf99c057c0dff2d4100b1647b30f585e444018481 | 67,145 | null | """Minimum-depth sequential expansion with learned staged carry.
hypothesis: v113 bounds family-8 depth but redundantly rescans all rejected
lower catalogues on updates2/3. Carry the learned top512 from each stage,
score those active complete rules on the current rows, and exhaustively
score only the newl... | {
"id": "1763e173-b095-435f-a5ba-7814633a61c4",
"created_at": "2026-08-02 21:03:52.933373+00:00",
"db_md5": "3548734bfdb46acf325b1f845619a11a",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "6372281c-f21b-4a5a-a025-4f3d52a139b0",
"tier": "easy",
"dataset_id": "e3",
"status": "succe... | {
"score": {
"mean_loss": 0.003014546214626436,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.00301454615599774,
"example_count": 800,
"... |
17661778-2380-4a55-81f6-19bad9cb5e07 | easy | TZZheng | 2026-08-07 04:38:59.915415+00:00 | succeeded | 265a7fbe6ae333832871e6c3ee3b7b5437f8e445627c63f5a4e8ccf8535ec2ef | 12,532 | null | """Learned recurrent modular-transition model.
The mutable digit state is initialized from x. A learned square-oriented
self-mixing network and an N-conditioned reduction network form one transition,
and exactly the same transition parameters are reused T times. No arithmetic
operation is hard-coded; only the public p... | {
"id": "17661778-2380-4a55-81f6-19bad9cb5e07",
"created_at": "2026-08-07 04:38:59.915415+00:00",
"db_md5": "76399259d7eeead0d8d5fdef15042700",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "3538a97d-66d9-4e1a-8436-211e9fcc93a8",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 10.00603461265564,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.06000000052154064
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 12.636055946350098,
"example_count": 100,... |
176717d4-6ef2-4348-8710-617244416d2e | easy | jordanrubin | 2026-08-11 01:50:30.896987+00:00 | succeeded | 20ede0b8469ed8f8ba7c78afc041bb710120ff9302be464571f14aee504195f7 | 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": "176717d4-6ef2-4348-8710-617244416d2e",
"created_at": "2026-08-11 01:50:30.896987+00:00",
"db_md5": "ff731c6828416ec56c3f69d90e8018d4",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "00fcdf06-dc07-46ed-96a1-3bd7e4d1156a",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.175448556248798,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007499999832361937
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.1651392827654097,
"example_count": 600... |
176cba0d-1e9d-40eb-8966-60abb5786f24 | easy | erdavis0 | 2026-08-23 15:29:48.587966+00:00 | succeeded | 7487528cfcbfc8b77a9d57a5e4603bd2ec4dedcf3635c4134191d420fcfa6d27 | 16,327 | null | """Full-tape learned field computation with global orthogonal scattering.
Opaque public categories are independently right-aligned into fixed N, X, and
T tapes. Separate interaction, carry, and modulus workspaces react through
one shared pointwise learned cell. Random learned strict-skew generators
produce global pl... | {
"id": "176cba0d-1e9d-40eb-8966-60abb5786f24",
"created_at": "2026-08-23 15:29:48.587966+00:00",
"db_md5": "28351bed687d09c10c79258f413c9db0",
"submitter": "Ethan",
"github_login": "erdavis0",
"run_id": "2c21d3e4-c531-4aa9-9b09-c43c0a37c5fd",
"tier": "easy",
"dataset_id": "e6",
"status": "succeeded",... | {
"score": {
"mean_loss": 1.8769747614860535,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05698198452591896
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.862117886543274,
"example_count": 60,
... |
177d28d3-5fab-461e-80e6-2692357305cd | easy | karanganesan | 2026-08-04 00:28:14.808624+00:00 | succeeded | 1d1c39ad640fb349eca567bfb30e286ec7627e67759856f09269730292a953f3 | 8,655 | 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": "177d28d3-5fab-461e-80e6-2692357305cd",
"created_at": "2026-08-04 00:28:14.808624+00:00",
"db_md5": "b4d87b0b48717578aa45ac956f415394",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "d3e9f495-d894-4c67-8da3-f175b281ebac",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 8.76064157485962,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.031666665337979794
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 8.42754077911377,
"example_count": 100,
... |
17808ef3-74b1-41a6-b609-42cec343d5c8 | easy | DDanlov | 2026-08-08 10:53:36.701002+00:00 | succeeded | f89d7d37d64932d0fbe4fd060e36152b50d82ab35061820f1f1332b301721e64 | 15,561 | null | from __future__ import annotations
import sys
import os
import math
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim.optimizer import Optimizer
VOCAB_SIZE = 17
DIGIT_OFFSET = 7
class RohanShampoo(Optimizer):
"""Self-contained RohanShampoo optimizer with eigenvalue ma... | {
"id": "17808ef3-74b1-41a6-b609-42cec343d5c8",
"created_at": "2026-08-08 10:53:36.701002+00:00",
"db_md5": "d821e6a4829e0e9b5e56019fdc79c665",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "68ad7550-6d1d-4bf6-a9d3-ed5c974dc663",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 98.4606735574011,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.000833333358168602
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 97.65385809928313,
"example_count": 600,
... |
178129af-0038-4e55-af7f-b9831ebd6fcc | easy | TZZheng | 2026-08-07 05:00:42.987631+00:00 | succeeded | 88834df04dbaaceac70196d6df7c55677ea33628f04fb3a2cd06b7f9205e8fa1 | 12,528 | null | """Learned recurrent modular-transition model.
The mutable digit state is initialized from x. A learned square-oriented
self-mixing network and an N-conditioned reduction network form one transition,
and exactly the same transition parameters are reused T times. No arithmetic
operation is hard-coded; only the public p... | {
"id": "178129af-0038-4e55-af7f-b9831ebd6fcc",
"created_at": "2026-08-07 05:00:42.987631+00:00",
"db_md5": "4a46d582290690dab8f82723b9976514",
"submitter": "TZZheng",
"github_login": "TZZheng",
"run_id": "c151abba-05cb-41a5-9d43-b454c9a52bc0",
"tier": "easy",
"dataset_id": "e1",
"status": "succeeded"... | {
"score": {
"mean_loss": 7.822463274002075,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.044999998062849045
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 9.744863510131836,
"example_count": 100,... |
17875634-c831-4943-b475-2a8c17b6e54b | easy | gauravmishra | 2026-08-10 08:27:24.603183+00:00 | succeeded | cc6f97457401ed2a9259cef30f0e53d1cc7a6eed390efb678aa36d829014bbb8 | 9,565 | null | """E44: historical K13-S385 with evaluator-owned two-pass SAM-AdamW.
Architecture parent SHA-256:
42dc862219390b8300e98260725c6c2621b27aae32f750f81902dea4bf95889d
Optimizer parent SHA-256:
6d5a70f506789d4c4bbc7a68efc29b772a942073b80bd4dd833b081f34e80209
"""
from __future__ import annotations
import torch
import torc... | {
"id": "17875634-c831-4943-b475-2a8c17b6e54b",
"created_at": "2026-08-10 08:27:24.603183+00:00",
"db_md5": "009f19532e0090a358d0d7649ab6f9c3",
"submitter": "Gaurav Mishra",
"github_login": "gauravmishra",
"run_id": "6acda56f-cddc-4762-ae2b-95ead5b16103",
"tier": "easy",
"dataset_id": "e5",
"status": ... | {
"score": {
"mean_loss": 2.2537446727302948,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.007500000024835269
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2852520429500016,
"example_count": 60... |
178cc404-4155-41c2-bf74-d9aad44dd75b | easy | shreyash-chonkie | 2026-09-01 03:38:08.178422+00:00 | succeeded | a31e7251ed1e2910090d4a522bf15f374e699b4d7d9202b356aa481ccebe51da | 20,246 | null | """Learned token scanner feeding a random recurrent path reservoir."""
from __future__ import annotations
from itertools import permutations
import torch
from torch import Tensor, nn
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
TokenLossBatch,
assert_model_s... | {
"id": "178cc404-4155-41c2-bf74-d9aad44dd75b",
"created_at": "2026-09-01 03:38:08.178422+00:00",
"db_md5": "1f1fc0968b24c6f2eccbda4e932aef45",
"submitter": "Shreyash",
"github_login": "shreyash-chonkie",
"run_id": "07d09a28-082c-450b-891c-6662246c467c",
"tier": "easy",
"dataset_id": "e10",
"status": ... | {
"score": {
"mean_loss": 24.65699005126953,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.05050504952669144
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 25.883277893066406,
"example_count": 125,... |
1793c871-0101-481b-a56e-683994f87fa5 | easy | 0xDEADD06 | 2026-08-12 14:40:02.434617+00:00 | failed | 9a080bebdd7e7ca83543233aac4c35aaa04361a7b0cbb2ef261d10cabf021bef | 5,831 | null | """ParaLSTM: recurrent LSTM with weight-tied passes for algorithmic tasks."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn, Tensor
from benchmark import (
ModelSpec,
OptimizerBundle,
OptimizerSpec,
Submission,
assert_model_state,
)
D_MODEL... | {
"id": "1793c871-0101-481b-a56e-683994f87fa5",
"created_at": "2026-08-12 14:40:02.434617+00:00",
"db_md5": "5fc0c8c33df74483ea2bb5ad9c624e8e",
"submitter": "0xDEADD06",
"github_login": "0xDEADD06",
"run_id": "346746d1-2d2e-45fc-bb63-3e567a780fbf",
"tier": "easy",
"dataset_id": "e5",
"status": "failed... | null
|
1796df32-faa6-4688-b98f-2d020ac76094 | easy | khushidahi | 2026-08-21 20:48:43.522394+00:00 | succeeded | 3d6b0ca0df77f6d195c0210c8eee33f92d67b119e3f4f91dbcd9b73bab959bd5 | 28,186 | null | """R362: one shared ordered decimal codebook across input and output.
This candidate is deliberately diagnostic. It trains on evaluator-owned final
labels at several depths so that one shared transition must compose. The model
never converts N or X to native numbers and never implements multiplication or
modular red... | {
"id": "1796df32-faa6-4688-b98f-2d020ac76094",
"created_at": "2026-08-21 20:48:43.522394+00:00",
"db_md5": "49cc307a03533e3a768755a226bf0f35",
"submitter": "khushidahi",
"github_login": "khushidahi",
"run_id": "f9ebe4fe-2310-490b-b438-8c8133f37a43",
"tier": "easy",
"dataset_id": "e5",
"status": "succ... | {
"score": {
"mean_loss": 2.8363346327283594,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.027083333333333334
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.8638253650323158,
"example_count": 60... |
1797a910-fb3c-4420-8571-15eedbb4ede8 | easy | chad-atexpedient | 2026-08-20 15:15:03.606212+00:00 | succeeded | 40fb1e2d738c0a25e82069ce26c4d84bf909a1674603e0c85451b26423e7ac74 | 23,938 | null | """X110: separate sum- and difference-routed pair interactions.
X105 routes both X-X and X-N interactions by place addition. Addition is the
right geometry for multiplication, but comparing a value against a modulus is
naturally aligned by place offset. X110 preserves the X-X anti-diagonal sum route
and adds a separat... | {
"id": "1797a910-fb3c-4420-8571-15eedbb4ede8",
"created_at": "2026-08-20 15:15:03.606212+00:00",
"db_md5": "7e093890efae64244aa5c78247f0291b",
"submitter": "chad-atexpedient",
"github_login": "chad-atexpedient",
"run_id": "778e2dc3-8bea-4df0-931d-b390a9ee5432",
"tier": "easy",
"dataset_id": "e5",
"st... | {
"score": {
"mean_loss": 2.39826118707144,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.005000000012417634
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.4824620961073802,
"example_count": 600,... |
179886b1-8294-466a-9d9d-f94cd03bf9a5 | easy | Mantissagithub | 2026-08-19 10:53:37.266057+00:00 | succeeded | b7337c62ebf9ad33c06fa353272e0c971ce80efd3318bba0b8f7cc31dcf40765 | 29,614 | null | """attempt_035 + a contentless T field (depth enters only as iteration count).
ADDENDUM (attempt_042_gated): attempt_040_residual with the cross-iteration
carry scaled by a single learned scalar gate, sigmoid(carry_logit), init 0.5.
This is one scalar parameter (+1 element against the model-state ceiling) and
one mult... | {
"id": "179886b1-8294-466a-9d9d-f94cd03bf9a5",
"created_at": "2026-08-19 10:53:37.266057+00:00",
"db_md5": "944e0b9d2d555ab917996ae68f46c271",
"submitter": "Pradheep P",
"github_login": "Mantissagithub",
"run_id": "c75bed17-14ad-4894-ac6e-06f224182b86",
"tier": "easy",
"dataset_id": "e5",
"status": "... | {
"score": {
"mean_loss": 2.6763793200685466,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.01708333334575097
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.593895487721191,
"example_count": 600,... |
179c6c01-b5aa-41ee-bb08-c443cc8f4b40 | easy | pepinu | 2026-08-21 06:35:08.195091+00:00 | succeeded | 8f2604323b0935ddcc7d49ce89cf4a60dda7735abe7afeae832b0240f7769930 | 40,010 | null | """Squaring mod N as a learned permutation on CRT components.
WHY THIS SHAPE. (Z_N, x) is a commutative monoid, so every finite-dimensional
linear representation splits into multiplicative characters and squaring is
diagonal there -- the component basis is the only one in which this operation is
shallow. Two consequen... | {
"id": "179c6c01-b5aa-41ee-bb08-c443cc8f4b40",
"created_at": "2026-08-21 06:35:08.195091+00:00",
"db_md5": "67bdec213264108b780b80da455f8b04",
"submitter": "pepinu",
"github_login": "pepinu",
"run_id": "0322c5f2-55af-47e9-9003-7e297a11769e",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded",
... | {
"score": {
"mean_loss": 0.010853210578552885,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.01086920937063849,
"example_count": 600,
"... |
17a7da96-f8ec-4c28-b0a2-7a0f5da7efc3 | easy | sapient-sapiens | 2026-08-09 14:34:48.261061+00:00 | succeeded | 6bf4d4fd7ddeead05c0b9110c34eb0b021b8eb987a28340ca44a2306cd6bbc25 | 11,016 | 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": "17a7da96-f8ec-4c28-b0a2-7a0f5da7efc3",
"created_at": "2026-08-09 14:34:48.261061+00:00",
"db_md5": "68da4a5fff2061e1ef98397d7ade0cac",
"submitter": "Ertondy",
"github_login": "sapient-sapiens",
"run_id": "579ef785-477d-40a1-9dbe-253d8a3795aa",
"tier": "easy",
"dataset_id": "e4",
"status": "su... | {
"score": {
"mean_loss": 2.2467999626061164,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.0003125
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.249164586317854,
"example_count": 1200,
... |
17b7c7d6-baf0-42bc-bb77-7aa17438a3db | easy | karanganesan | 2026-08-04 02:05:49.841456+00:00 | succeeded | c45f3fdbd3f2b919b60b1248a6543b7605186f42953d5f8a39397bced14985bc | 11,749 | 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": "17b7c7d6-baf0-42bc-bb77-7aa17438a3db",
"created_at": "2026-08-04 02:05:49.841456+00:00",
"db_md5": "4848e0614a0ce13cc9f6abe91797fce7",
"submitter": "Karan Ganesan",
"github_login": "karanganesan",
"run_id": "4ad6bad5-b45d-4a10-b8fa-6d227c53d913",
"tier": "easy",
"dataset_id": "e1",
"status": ... | {
"score": {
"mean_loss": 2.315798044204712,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.026666666381061077
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.413492202758789,
"example_count": 100,... |
17bb002f-666b-488a-9452-09c164a97174 | easy | DDanlov | 2026-08-20 12:22:12.997388+00:00 | succeeded | e45a2bbb7bb84a5d4db3f294ac5edf0547acad2d5477044b262bbe86dc3e2d44 | 20,958 | null | from __future__ import annotations
import math
import time
from typing import Any, Optional, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import Optimizer
try:
from benchmark.api import ModelSpec, OptimizerBundle, OptimizerSpec, Submission, assert_model_state
except Imp... | {
"id": "17bb002f-666b-488a-9452-09c164a97174",
"created_at": "2026-08-20 12:22:12.997388+00:00",
"db_md5": "f118f2c8d8e16a5f1dca4dcdfe51238e",
"submitter": "DDanlov",
"github_login": "DDanlov",
"run_id": "fcb85e50-d9ae-4573-a896-410bcfcb2979",
"tier": "easy",
"dataset_id": "e5",
"status": "succeeded"... | {
"score": {
"mean_loss": 2.345453574917724,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.012083333345750968
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.3410296589804336,
"example_count": 600... |
17bd3476-7c4f-480e-81f2-475d7b8109e5 | easy | jordanrubin | 2026-08-13 18:21:46.281110+00:00 | succeeded | e2245c74dfcfe7d00f8aeee61890a07c07a0d89977c245e3844890ce7f77b424 | 27,175 | null | """Weight-tied MLP-cell loop over soft digit states (mlploop).
Cell = MLP over [expected digits, pairwise digit products, RNS residue
simplices (fixed differentiable mixing over Z_p), N digits]. The cell that
learns one-step modular squaring from direct pairs (52% unseen at 14k rows,
day-20 screen) inside the exact-T ... | {
"id": "17bd3476-7c4f-480e-81f2-475d7b8109e5",
"created_at": "2026-08-13 18:21:46.281110+00:00",
"db_md5": "3f710550312ea469d13b008d83980242",
"submitter": "Jordan Rubin",
"github_login": "jordanrubin",
"run_id": "1c11ed22-99cf-4344-9acc-9cb8bfbfc88f",
"tier": "easy",
"dataset_id": "e5",
"status": "s... | {
"score": {
"mean_loss": 2.8398660081881375,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.03166666588125129
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 2.2247891210921678,
"example_count": 600... |
17c2890e-4b26-4240-91e7-bd744abedf0a | easy | oupadhyay | 2026-08-26 15:35:03.933122+00:00 | failed | 084e40da66c31c25d535c05b52f4194759790275d73a6b8b52f5301f39fce464 | 4,046 | null | """Literal polynomial cell."""
from __future__ import annotations
import torch
import torch.nn.functional as F
from torch import nn
from benchmark import OptimizerBundle,Submission,TokenLossBatch,assert_model_state
D,H=48,24
class C:
def __init__(s,vocab_size,max_seq_len):s.vocab_size,s.max_seq_len=vocab_size,max_seq_... | {
"id": "17c2890e-4b26-4240-91e7-bd744abedf0a",
"created_at": "2026-08-26 15:35:03.933122+00:00",
"db_md5": "dc12998ad4986f721c6b96dd163f541d",
"submitter": "Ojasw Upadhyay",
"github_login": "oupadhyay",
"run_id": "4c026c1c-3474-4f55-8949-d9e1d53394d6",
"tier": "easy",
"dataset_id": "e5",
"status": "f... | null
|
17ceb480-8ae9-48f5-b64b-2cafc0d51bb7 | easy | sankalp1999 | 2026-08-04 17:13:15.549006+00:00 | succeeded | 3b79fc21de9e0100fcd186c5a038abe5e87d8989a431eb06875bdc25e9f9d319 | 34,659 | null | """p2_program: identify f by hypothesis filtering, with a neural fallback.
The diagnosis this is built on
------------------------------
p1_generic memorises 14.4k examples of x^2 mod N to train loss 0.000 and scores
0% held out, at every width from 12.6M to 402M, at every LR that does not
diverge, on 16k / 50k / 150k... | {
"id": "17ceb480-8ae9-48f5-b64b-2cafc0d51bb7",
"created_at": "2026-08-04 17:13:15.549006+00:00",
"db_md5": "1d834f16da97d60fe96aa31d6bab0bff",
"submitter": "Sankalp ",
"github_login": "sankalp1999",
"run_id": "0d83f15b-b6e7-4bb4-ac1f-a9ad3588a496",
"tier": "easy",
"dataset_id": "e5",
"status": "succe... | {
"score": {
"mean_loss": 0.0,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 1.0
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 0.0,
"example_count": 600,
"exact_accuracy": 1.0,
"... |
17d794bc-d715-4c88-8491-6c3a4b105c37 | easy | viridale | 2026-08-13 10:28:40.589658+00:00 | succeeded | bb8a7a3d1264b7ccedee4a6840cedebd883739a7a4f97eea3d4b897a181951a1 | 24,571 | null | """Product-matrix carry algebra with a native digit coordinate.
hypothesis: v545's one four-state monoid was noncommutative but narrow; v549's wider
affine monoid was coordinatewise and geometry-specific. Four independent learned
4x4 matrix monoids provide sixteen interacting control coordinates at tractable
... | {
"id": "17d794bc-d715-4c88-8491-6c3a4b105c37",
"created_at": "2026-08-13 10:28:40.589658+00:00",
"db_md5": "b51bdcbba42fa9f6840ccbcf8d9aa9c6",
"submitter": "priormancer",
"github_login": "viridale",
"run_id": "b29f718f-186a-484a-a2e1-a43c0199f5e8",
"tier": "easy",
"dataset_id": "e1",
"status": "succe... | {
"score": {
"mean_loss": 1.8311987519264221,
"primary_metric": "mean_exact_accuracy",
"num_measurements": 2,
"mean_exact_accuracy": 0.026666666381061077
},
"seeds": [
{
"seed": 74,
"evaluation": {
"ood": {
"loss": 1.7800687551498413,
"example_count": 10... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.