Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code: DatasetGenerationError
Exception: CastError
Message: Couldn't cast
step: int64
phase: string
domain: string
head: string
loss: double
task_loss: double
grad_norm: double
vram_gb: double
best_loss: double
eta_hours: double
elapsed_s: double
timestamp: string
task: string
to
{'step': Value('int64'), 'loss': Value('float64'), 'task_loss': Value('float64'), 'phase': Value('string'), 'task': Value('string'), 'domain': Value('string'), 'grad_norm': Value('float64'), 'vram_gb': Value('float64'), 'elapsed_s': Value('float64'), 'timestamp': Value('string')}
because column names don't match
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1872, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 289, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 124, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2272, in table_cast
return cast_table_to_schema(table, schema)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2218, in cast_table_to_schema
raise CastError(
datasets.table.CastError: Couldn't cast
step: int64
phase: string
domain: string
head: string
loss: double
task_loss: double
grad_norm: double
vram_gb: double
best_loss: double
eta_hours: double
elapsed_s: double
timestamp: string
task: string
to
{'step': Value('int64'), 'loss': Value('float64'), 'task_loss': Value('float64'), 'phase': Value('string'), 'task': Value('string'), 'domain': Value('string'), 'grad_norm': Value('float64'), 'vram_gb': Value('float64'), 'elapsed_s': Value('float64'), 'timestamp': Value('string')}
because column names don't match
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1347, in compute_config_parquet_and_info_response
parquet_operations = convert_to_parquet(builder)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 980, in convert_to_parquet
builder.download_and_prepare(
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 884, in download_and_prepare
self._download_and_prepare(
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 947, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1739, in _prepare_split
for job_id, done, content in self._prepare_split_single(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1925, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.exceptions.DatasetGenerationError: An error occurred while generating the datasetNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
step int64 | loss float64 | task_loss float64 | phase string | task string | domain string | grad_norm float64 | vram_gb float64 | elapsed_s float64 | timestamp string |
|---|---|---|---|---|---|---|---|---|---|
1 | 14.393731 | 11.275101 | foundation | cog_D2 | bpe | 15.065992 | 5.536942 | 26.023984 | 2026-04-08T14:17:18.222866 |
2 | 5.978209 | 5.978208 | foundation | math_D1 | math | 551,945.5 | 5.54673 | 48.250006 | 2026-04-08T14:17:41.025610 |
3 | 11.601094 | 8.514552 | foundation | cog_D2 | bpe | 202,212.625 | 5.54673 | 73.141066 | 2026-04-08T14:18:05.946752 |
4 | 11.574977 | 8.490288 | foundation | cog_D3 | bpe | 9.635531 | 5.54673 | 99.442276 | 2026-04-08T14:18:32.267890 |
5 | 11.570036 | 8.487131 | foundation | cog_D2 | bpe | 9.612004 | 5.54673 | 126.723572 | 2026-04-08T14:18:59.575004 |
6 | 11.542425 | 8.459512 | foundation | cog_D2 | bpe | 9.666193 | 5.54673 | 153.961528 | 2026-04-08T14:19:26.787863 |
7 | 11.48914 | 8.404629 | foundation | language | bpe | 9.56272 | 5.54673 | 180.718482 | 2026-04-08T14:19:53.542782 |
8 | 11.458981 | 8.376014 | foundation | language | bpe | 48,291.609375 | 5.54673 | 207.148545 | 2026-04-08T14:20:19.978939 |
9 | 11.395897 | 8.314148 | foundation | language | bpe | 9.503506 | 5.54673 | 234.150749 | 2026-04-08T14:20:46.977576 |
10 | 11.306026 | 8.224536 | foundation | language | bpe | 9.363232 | 5.54673 | 261.035598 | 2026-04-08T14:21:13.862362 |
11 | 11.216825 | 8.137567 | foundation | cog_D1 | bpe | 9.350598 | 5.54673 | 287.983653 | 2026-04-08T14:21:40.821148 |
12 | 11.070868 | 7.990856 | foundation | cog_D2 | bpe | 9.284871 | 5.54673 | 314.987768 | 2026-04-08T14:22:07.820182 |
13 | 10.967323 | 7.891262 | foundation | language | bpe | 9.13363 | 5.54673 | 342.030278 | 2026-04-08T14:22:34.869818 |
14 | 10.769574 | 7.695646 | foundation | language | bpe | 9.059166 | 5.54673 | 369.135917 | 2026-04-08T14:23:01.962640 |
15 | 10.63722 | 7.565178 | foundation | cog_D2 | bpe | 8.771568 | 5.55872 | 396.257699 | 2026-04-08T14:23:29.079019 |
16 | 10.424549 | 7.35401 | foundation | language | bpe | 8.611031 | 5.55872 | 423.386631 | 2026-04-08T14:23:56.220632 |
17 | 10.331953 | 7.265524 | foundation | cog_D2 | bpe | 7,048.64209 | 5.55872 | 449.880404 | 2026-04-08T14:24:22.705893 |
18 | 10.215185 | 7.150193 | foundation | language | bpe | 2,691.442383 | 5.55872 | 475.751416 | 2026-04-08T14:24:48.582473 |
19 | 9.982019 | 6.920879 | foundation | language | bpe | 7.850101 | 5.55872 | 502.895061 | 2026-04-08T14:25:15.725588 |
20 | 3.194425 | 3.194423 | foundation | math_D2 | math | 1,186.448364 | 5.55872 | 529.265877 | 2026-04-08T14:25:42.086377 |
21 | 9.70642 | 6.649512 | foundation | language | bpe | 7.505064 | 5.55872 | 556.215019 | 2026-04-08T14:26:09.054269 |
22 | 9.581362 | 6.528383 | foundation | language | bpe | 7.39989 | 5.55872 | 583.331295 | 2026-04-08T14:26:36.168744 |
23 | 9.38299 | 6.33442 | foundation | cog_D1 | bpe | 764.443298 | 5.55872 | 609.326276 | 2026-04-08T14:27:02.152723 |
24 | 9.219893 | 6.175097 | foundation | language | bpe | 7.074155 | 5.55872 | 636.245511 | 2026-04-08T14:27:29.077077 |
25 | 9.078879 | 6.036811 | foundation | language | bpe | 6.838331 | 5.55872 | 663.254786 | 2026-04-08T14:27:56.081725 |
26 | 8.859949 | 5.822114 | foundation | language | bpe | 6.69445 | 5.55872 | 690.2989 | 2026-04-08T14:28:23.129510 |
27 | 8.656173 | 5.622917 | foundation | language | bpe | 601.330994 | 5.55872 | 716.774685 | 2026-04-08T14:28:49.601018 |
28 | 8.586067 | 5.55686 | foundation | language | bpe | 446.07547 | 5.55872 | 743.230093 | 2026-04-08T14:29:16.056233 |
29 | 8.393439 | 5.367965 | foundation | language | bpe | 5.817588 | 5.55872 | 770.194806 | 2026-04-08T14:29:43.024418 |
30 | 8.263997 | 5.247897 | foundation | language | bpe | 5.669382 | 5.55872 | 797.136974 | 2026-04-08T14:30:09.971884 |
31 | 8.171625 | 5.157629 | foundation | cog_D2 | bpe | 356.328857 | 5.55872 | 823.518442 | 2026-04-08T14:30:36.346192 |
32 | 8.102732 | 5.094116 | foundation | language | bpe | 5.28403 | 5.55872 | 850.544952 | 2026-04-08T14:31:03.377006 |
33 | 7.941405 | 4.939985 | foundation | cog_D3 | bpe | 5.100594 | 5.55872 | 877.664862 | 2026-04-08T14:31:30.502121 |
34 | 7.721476 | 4.727292 | foundation | language | bpe | 407.893707 | 5.55872 | 904.275015 | 2026-04-08T14:31:57.106641 |
35 | 7.62527 | 4.63847 | foundation | cog_D2 | bpe | 4.750548 | 5.55872 | 931.401455 | 2026-04-08T14:32:24.229287 |
36 | 7.490486 | 4.510187 | foundation | language | bpe | 4.611003 | 5.55872 | 958.285521 | 2026-04-08T14:32:51.115796 |
37 | 7.319392 | 4.350196 | foundation | language | bpe | 4.475295 | 5.55872 | 985.456416 | 2026-04-08T14:33:18.287749 |
38 | 7.248524 | 4.288671 | foundation | language | bpe | 4.32087 | 5.55872 | 1,012.393615 | 2026-04-08T14:33:45.219002 |
39 | 7.038265 | 4.088513 | foundation | language | bpe | 4.164541 | 5.55872 | 1,039.423097 | 2026-04-08T14:34:12.257200 |
40 | 6.933413 | 3.994433 | foundation | cog_D1 | bpe | 4.096105 | 5.55872 | 1,066.640662 | 2026-04-08T14:34:39.472732 |
41 | 2.184479 | 2.184475 | foundation | math_D2 | math | 329.252014 | 5.55872 | 1,093.106615 | 2026-04-08T14:35:05.926683 |
42 | 6.670658 | 3.755095 | foundation | cog_D2 | bpe | 4.024798 | 5.55872 | 1,120.132973 | 2026-04-08T14:35:32.970836 |
43 | 6.574663 | 3.671324 | foundation | language | bpe | 3.950286 | 5.55872 | 1,147.361751 | 2026-04-08T14:36:00.191387 |
44 | 6.465172 | 3.576617 | foundation | cog_D2 | bpe | 3.878572 | 5.55872 | 1,174.472934 | 2026-04-08T14:36:27.299330 |
45 | 6.290162 | 3.416647 | foundation | language | bpe | 3.84665 | 5.55872 | 1,201.326799 | 2026-04-08T14:36:54.155371 |
46 | 2.295437 | 2.295433 | foundation | math_D1 | math | 222.957245 | 5.55872 | 1,227.797631 | 2026-04-08T14:37:20.618629 |
47 | 6.082046 | 3.237749 | foundation | cog_D1 | bpe | 3.809242 | 5.55872 | 1,254.938543 | 2026-04-08T14:37:47.776157 |
48 | 5.987242 | 3.158971 | foundation | language | bpe | 3.73201 | 5.55872 | 1,281.727154 | 2026-04-08T14:38:14.553438 |
49 | 5.840088 | 3.03182 | foundation | language | bpe | 3.683343 | 5.55872 | 1,308.747196 | 2026-04-08T14:38:41.582180 |
50 | 5.688198 | 2.900123 | foundation | cog_D2 | bpe | 3.647856 | 5.55872 | 1,335.891475 | 2026-04-08T14:39:08.720308 |
51 | 5.620642 | 2.847663 | foundation | language | bpe | 291.802582 | 5.55872 | 1,362.411911 | 2026-04-08T14:39:35.241133 |
52 | 5.508452 | 2.756604 | foundation | language | bpe | 3.577532 | 5.55872 | 1,389.466541 | 2026-04-08T14:40:02.298474 |
53 | 5.452821 | 2.718495 | foundation | language | bpe | 3.456722 | 5.55872 | 1,416.492708 | 2026-04-08T14:40:29.321944 |
54 | 5.358399 | 2.639658 | foundation | cog_D2 | bpe | 223.29834 | 5.55872 | 1,442.462838 | 2026-04-08T14:40:55.292183 |
55 | 5.177473 | 2.484287 | foundation | language | bpe | 3.46989 | 5.55872 | 1,469.469495 | 2026-04-08T14:41:22.300414 |
56 | 2.221123 | 2.221117 | foundation | math_D2 | math | 183.626251 | 5.55872 | 1,495.941398 | 2026-04-08T14:41:48.761435 |
57 | 5.063032 | 2.402569 | foundation | cog_D2 | bpe | 3.335264 | 5.55872 | 1,522.977428 | 2026-04-08T14:42:15.806573 |
58 | 2.065715 | 2.065712 | foundation | math_D2 | math | 176.648453 | 5.55872 | 1,549.50153 | 2026-04-08T14:42:42.321594 |
59 | 4.888708 | 2.271163 | foundation | cog_D3 | bpe | 2.995616 | 5.55872 | 1,576.561528 | 2026-04-08T14:43:09.390909 |
60 | 4.808334 | 2.211668 | foundation | language | bpe | 3.000092 | 5.55872 | 1,603.67393 | 2026-04-08T14:43:36.512690 |
61 | 4.782877 | 2.200411 | foundation | language | bpe | 2.760939 | 5.55872 | 1,630.832035 | 2026-04-08T14:44:03.670786 |
62 | 4.71143 | 2.149776 | foundation | language | bpe | 2.791388 | 5.55872 | 1,658.004379 | 2026-04-08T14:44:30.841654 |
63 | 4.601563 | 2.070397 | foundation | language | bpe | 339.199982 | 5.55872 | 1,683.84736 | 2026-04-08T14:44:56.672352 |
64 | 4.522365 | 2.018022 | foundation | language | bpe | 2.667801 | 5.55872 | 1,710.797601 | 2026-04-08T14:45:23.636330 |
65 | 4.520829 | 2.023156 | foundation | cog_D2 | bpe | 2.429053 | 5.55872 | 1,737.963228 | 2026-04-08T14:45:50.791035 |
66 | 4.38259 | 1.928275 | foundation | language | bpe | 2.337807 | 5.55872 | 1,764.875493 | 2026-04-08T14:46:17.702089 |
67 | 2.004086 | 2.004084 | foundation | math_D1 | math | 269.749084 | 5.55872 | 1,789.790487 | 2026-04-08T14:46:42.611245 |
68 | 4.309349 | 1.901803 | foundation | language | bpe | 2.920581 | 5.55872 | 1,816.918177 | 2026-04-08T14:47:09.752140 |
69 | 4.277479 | 1.881119 | foundation | language | bpe | 2.053432 | 5.55872 | 1,843.793907 | 2026-04-08T14:47:36.622281 |
70 | 4.195809 | 1.835732 | foundation | language | bpe | 2.6392 | 5.55872 | 1,870.845971 | 2026-04-08T14:48:03.675221 |
71 | 4.106848 | 1.782329 | foundation | language | bpe | 1.933949 | 5.55872 | 1,897.905662 | 2026-04-08T14:48:30.738374 |
72 | 4.085266 | 1.78445 | foundation | language | bpe | 2.369529 | 5.55872 | 1,925.09869 | 2026-04-08T14:48:57.927401 |
73 | 4.059068 | 1.780413 | foundation | language | bpe | 1.932692 | 5.55872 | 1,951.94334 | 2026-04-08T14:49:24.768333 |
74 | 3.934036 | 1.701195 | foundation | language | bpe | 1.840266 | 5.55872 | 1,978.982208 | 2026-04-08T14:49:51.820466 |
75 | 3.900321 | 1.697533 | foundation | cog_D3 | bpe | 1.729512 | 5.55872 | 2,006.094725 | 2026-04-08T14:50:18.932615 |
76 | 3.861178 | 1.684433 | foundation | language | bpe | 1.797878 | 5.55872 | 2,033.141368 | 2026-04-08T14:50:45.969889 |
77 | 3.816336 | 1.674152 | foundation | language | bpe | 100.762039 | 5.55872 | 2,059.233644 | 2026-04-08T14:51:12.072427 |
78 | 3.844666 | 1.713973 | foundation | language | bpe | 3.230531 | 5.55872 | 2,086.449851 | 2026-04-08T14:51:39.276660 |
79 | 3.768594 | 1.673069 | foundation | language | bpe | 1.894403 | 5.55872 | 2,113.290331 | 2026-04-08T14:52:06.116101 |
80 | 3.705286 | 1.649837 | foundation | language | bpe | 78.64106 | 5.55872 | 2,139.624261 | 2026-04-08T14:52:32.458471 |
81 | 3.721405 | 1.688796 | foundation | language | bpe | 160.776794 | 5.55872 | 2,165.175354 | 2026-04-08T14:52:58.013637 |
82 | 3.763832 | 1.7525 | foundation | cog_D2 | bpe | 5.042962 | 5.55872 | 2,192.138256 | 2026-04-08T14:53:24.971663 |
83 | 3.710308 | 1.717309 | foundation | language | bpe | 3.895301 | 5.55872 | 2,219.125037 | 2026-04-08T14:53:51.951891 |
84 | 3.590942 | 1.628826 | foundation | cog_D2 | bpe | 74.936104 | 5.55872 | 2,245.501992 | 2026-04-08T14:54:18.328752 |
85 | 3.594478 | 1.654166 | foundation | language | bpe | 99.002007 | 5.55872 | 2,270.916594 | 2026-04-08T14:54:43.744951 |
86 | 3.622346 | 1.714081 | foundation | language | bpe | 4.738483 | 5.55872 | 2,297.90253 | 2026-04-08T14:55:10.729397 |
87 | 3.596311 | 1.694313 | foundation | language | bpe | 4.002959 | 5.55872 | 2,324.816863 | 2026-04-08T14:55:37.646039 |
88 | 3.549018 | 1.656596 | foundation | cog_D2 | bpe | 2.746698 | 5.55872 | 2,351.849981 | 2026-04-08T14:56:04.680902 |
89 | 3.459421 | 1.602425 | foundation | cog_D2 | bpe | 2.303792 | 5.55872 | 2,378.883314 | 2026-04-08T14:56:31.713929 |
90 | 3.504732 | 1.652285 | foundation | language | bpe | 2.817028 | 5.55872 | 2,405.880471 | 2026-04-08T14:56:58.706898 |
91 | 3.455418 | 1.631007 | foundation | language | bpe | 3.001804 | 5.55872 | 2,432.745635 | 2026-04-08T14:57:25.581311 |
92 | 3.389996 | 1.598796 | foundation | language | bpe | 72.905159 | 5.55872 | 2,458.792972 | 2026-04-08T14:57:51.628450 |
93 | 3.388363 | 1.61796 | foundation | cog_D3 | bpe | 2.816233 | 5.55872 | 2,485.86012 | 2026-04-08T14:58:18.696564 |
94 | 3.346949 | 1.595958 | foundation | language | bpe | 2.237379 | 5.55872 | 2,512.897045 | 2026-04-08T14:58:45.722589 |
95 | 3.325164 | 1.58757 | foundation | language | bpe | 1.931317 | 5.55872 | 2,539.71945 | 2026-04-08T14:59:12.543975 |
96 | 3.287006 | 1.581173 | foundation | cog_D2 | bpe | 2.273898 | 5.55872 | 2,566.710051 | 2026-04-08T14:59:39.546862 |
97 | 2.163351 | 2.16335 | foundation | math_D1 | math | 75.748039 | 5.55872 | 2,592.661876 | 2026-04-08T15:00:05.482769 |
98 | 3.350524 | 1.661838 | foundation | language | bpe | 2.843746 | 5.55872 | 2,619.692293 | 2026-04-08T15:00:32.522141 |
99 | 3.24283 | 1.600684 | foundation | language | bpe | 77.074593 | 5.55872 | 2,645.777784 | 2026-04-08T15:00:58.605873 |
100 | 3.222349 | 1.597344 | foundation | cog_D2 | bpe | 62.517479 | 5.55872 | 2,672.203335 | 2026-04-08T15:01:25.030286 |
End of preview.
GLADIUS Research Dataset
Research papers, architecture documentation, training analysis, and experimental results for the GLADIUS cognitive kernel and WYRM model family.
GLADIUS is a novel transformer architecture featuring Synthase depth attention, PUP (Propagated Uncertainty Principle), SLA² specialist heads, and memory-augmented cognition. WYRM is the product model — currently training at 476M parameters on Kaggle T4.
Organization: Artifact Virtual
Structure
Architecture & Core
| Directory | Contents |
|---|---|
architecture/ |
Core specs, anti-collapse patches, trajectory docs |
src/ |
Synthase kernel source (Python) |
training-logs/ |
Raw training telemetry and loss data |
eval-results/ |
Evaluation outputs and benchmarks |
analysis/ |
Training analysis, convergence studies |
dashboards/ |
Visualization dashboards |
data/ |
Tokenizer files, data configs |
trajectory/ |
Architecture evolution and roadmap |
reference/ |
Reference materials |
Research Papers (papers/)
Uranium Series — Novel Contributions
| Paper | Topic |
|---|---|
1-bit-intelligence.md |
Binary weight learning — gradients are optional |
progressive-expansion.md |
Warm-starting larger models from smaller checkpoints |
ghost-protocol.md |
Autoregressive self-poisoning in agentic systems |
cell-division.md |
Architecture scaling via biological cell division |
uranium-layer7-gateway-halflife.md |
Cross-modal invariant boundaries and cognitive half-lives |
the-impulse-problem.md |
Impulse dynamics in gradient-free learning |
impulse-paper-notes.md |
Impulse paper working notes |
Architecture Analysis
| Paper | Topic |
|---|---|
gladius-moda-depth-attention.md |
MoDA (Mixture of Depth Attention) mechanism |
gladius-cognition-awakening.md |
Cognition module activation analysis |
gladius-forward-pass-map.md |
Complete forward pass trace |
gladius-invariant-deep-analysis.md |
Cross-modal invariant analysis |
gladius-habitat-paper.md |
Training habitat and curriculum design |
gladius-spectre-cycle.md |
Loss oscillation patterns |
gladius-time-series-definitive.md |
Time series integration via surgical head swap |
gladius-progressive-expansion.md |
Progressive expansion experiments |
GLADIUS-COMPLETE-FINDINGS.md |
Comprehensive findings summary |
gladius-day30-definitive-paper.md |
Day 30 milestone — definitive architecture paper |
Training Reports
| Paper | Topic |
|---|---|
gladius-distillation-gpt2-report.md |
GPT-2 distillation results |
gladius-atp-analysis-report.md |
ATP (All-Time Performance) analysis |
gladius-mnist-dissection.md |
MNIST task dissection |
dragon-autopsy-v2.2.md |
Training collapse autopsy |
collapse-raw-data.md |
Raw collapse event data |
static-analysis.md |
Static architecture analysis |
v2.1-entropy-fix-notes.md |
Entropy fix notes |
IEEE-Format Papers (papers/ieee/)
| Paper | Topic |
|---|---|
gladius-adaptive-cognitive-model.md |
Adaptive cognitive model |
gladius-resonance-architecture.md |
Resonance architecture |
gladius-cross-modal-invariant.md |
Cross-modal invariant learning |
gladius-cross-modal-geometry.md |
Cross-modal geometry |
gladius-distillation-edge.md |
Edge distillation |
gladius-hatchling-training.md |
Hatchling training curriculum |
gladius-broadcast-experiment.md |
Broadcast activation experiments |
gladius-multi-script-attention.md |
Multi-script attention patterns |
gladius-time-series-implantation.md |
Time series surgical implantation |
gladius-vlm-feeding-pipeline.md |
Vision-language feeding pipeline |
lattice-clock-temporal-quantization.md |
Temporal quantization |
muonclip-orthogonal-optimization.md |
MuonCLIP orthogonal optimization |
gladius-multi-tokenizer.md |
Multi-tokenizer architecture |
gladius-pup-uncertainty.md |
PUP uncertainty propagation |
gladius-training-report.md |
Training methodology report |
Research Reports (reports/)
| Report | Topic |
|---|---|
moda-v2-synthase-design.md |
MoDA v2 Synthase design spec |
atp-resonance-analysis.md |
ATP resonance analysis |
depth-attention-survey-2026.md |
Depth attention survey |
gladius-geometry-derivation.md |
Geometric derivations |
golden-ratio-investigation.md |
Golden ratio in attention |
golden-ratio-paper.md |
Golden ratio formal paper |
allspark-numbers.md |
AllSpark numerical analysis |
allspark-numbers-verified.md |
AllSpark verified results |
stress-test-report.md |
Architecture stress tests |
WYRM v29 — Current Training Run
- Architecture: 476M params · 1024d / 24L / 32H / 4096FFN
- Innovations: Synthase depth attention, PUP uncertainty, SLA² specialists, memory
- Platform: Kaggle T4 (16GB VRAM)
- Corpus: 9GB scientific text (QM, GR, calculus, topology, genomics, arXiv)
- Progress: Step 2,241 / 15,000
Artifact Virtual — 2026
- Downloads last month
- 896