Evidence ledger // 0-bro

M0 // character-model toy

01Ternary QAT

0.072 gap

final loss gap at 4,000 steps: fp32 finishes at 0.015 and ternary QAT at 0.086, with 35% zero weights.

Run: python3 mind-1b/proto/toy_ternary_lm.py --steps 4000

Limit. This trains on the text of its own source file. It establishes that this quantization recipe can converge in a tiny deterministic setting; it does not establish language-model parity at useful scale.

Read the script on GitHub

M3 scout // operation-count check

02Surprise gate

7.7%

of dense MACs on a generated 512-dimensional, 2,000-step stream at redundancy 0.95 and threshold 0. The relative RMS error is 4.14e-6, not mathematically zero.

Run: python3 mind-1b/proto/surprise_gating.py

Default result sweep
Redundancy	Threshold	MAC ratio	Relative RMS error
0.85	0.00	16.7%	0.000007
0.85	0.10	15.5%	0.014681
0.85	0.40	12.2%	0.115351
0.95	0.00	7.7%	0.000004
0.95	0.10	7.1%	0.016392
0.95	0.40	5.2%	0.121315

Limit. This counts operations for an exact delta-coded linear matvec on a generated stream. It does not measure language-model quality, kernel latency, or training savings.

Read the script on GitHub / Read the original lab note

M4 scout // synthetic routing rule

03Phase routing

1.90/ 8

active regions per token in the default eight-domain demo: 23.75% of dense region compute, with zero trainable router parameters and 12 settling steps.

Run: python3 mind-1b/proto/phase_binding.py

Default routing output
Input kind	Active regions	Relative error
Simple, one domain	1.65	0.257
Composite, two domains	2.37	0.466
Dense all-region baseline	8.00 (4.2× compute)	0.914

Limit. The script seeds each region's prototype and expert matrix from the synthetic task's ground-truth centers and rules. It tests the routing rule, not learned specialization or a fair trained-router comparison.

Read the script on GitHub

M2 // synthetic continual-memory check

04Sleep cleanup

81.4%

mean sign recall after eight cycles with cleanup, versus 34.4% for the naive raw-magnitude retention control.

Run: python3 mind-1b/proto/sleep_cleanup.py

20,000 synthetic synapses; eight tasks; 400 signal synapses per task; seed 0
Cycle	With cleanup	Naive control
1	89.2%	49.8%
2	88.5%	44.1%
4	85.2%	39.1%
6	82.3%	35.6%
8	81.4%	34.4%

Limit. Neither arm is a language model. The control is a naive raw-magnitude retention strategy, not an otherwise-identical model "without sleep." This result earns a larger experiment; it does not solve catastrophic forgetting.

Read the script on GitHub / Read the original lab note

Configuration calculator

05Memory budget

221 MB

packed non-embedding ternary core: 885.08M weights at two bits each. This is not the complete 1.016B-parameter configuration.

Run: python3 mind-1b/calc/params.py and python3 mind-1b/calc/memory_budget.py

Current default configuration
Measure	Value	Scope
Total parameters	1.016B	Untied embeddings plus core
Training estimate	15.76 GB/GPU	QAT, batch 8, sequence 2,048, checkpointing
Wake-mode session	1.50 GB	Core, fp16 embeddings, int8 plastic tags, and 64k-token ring

Limit. These are arithmetic planning estimates. They are not measured device memory or throughput, and the int8 plastic-tag assumption remains a design choice to test.

Parameter calculator / Memory calculator / Design specification

What earns the next claim

06The gates still open

The next material claim needs a 100M PyTorch model, matched dense baselines, held-out language tasks, measured latency and memory, ablations for each novel mechanism, and a published failure report if the gate does not pass. MatFormer elastic width is still gated because its toy run carries a notable tax. Until then, this page remains a ledger of small checks.

The claim, the command, and the limit.