custom
code
sovereign-compute
File size: 2,978 Bytes
ef6eb55
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
# PAX-Coder Package Manifest

Package: `pax-coder`
Version: `1.0.0`
Release date: 2026-08-18
Repository: `SNAPKITTYWEST/pax-coder`

## Package Identity

PAX-Coder is the institutional package for proof-carrying GPU kernel generation
around the PAX axiom basis, Lean 4 proof modules, CUDA/PTX implementation
surfaces, Futhark functional references, and training-data export.

## Contents

| Path | Package role |
| --- | --- |
| `README.md` | Institutional entry point |
| `ABOUT.md` | Short project overview |
| `LICENSE.tri` | License structure |
| `VERSION` | Version marker |
| `CHANGELOG.md` | Release history |
| `RELEASE_NOTES.md` | Current release notes |
| `PAX/` | Lean 4 proof-module surfaces |
| `src/` | CUDA/PTX/Futhark source surfaces |
| `backends/` | License-policy backend |
| `docs/` | Institutional, user, architecture, and GTM documentation |
| `demo/` | Demonstration package |
| `export_training_data.py` | Training-data exporter |
| `train.py` | QLoRA training script |
| `run_training.sh` | Training launcher |
| `Modelfile` | Ollama packaging template |
| `MODEL_CARD.md` | Model-card draft |
| `DATASET_CARD.md` | Dataset-card draft |
| `SOVEREIGN_NODE_KEY.md` | Node-key and seal policy |
| `CONTRIBUTING.md` | Contribution guidance |

## Release Gates

The package may be published as an institutional repository release when:

- Version files and release notes are present.
- README states the PAX axiom basis and governance rules.
- License text matches `LICENSE.tri`.
- GitHub About metadata and topics identify the institutional scope.
- Release notes do not overclaim artifact-specific runtime verification.
- `python export_training_data.py` completes in the release environment.

Generated kernels require additional artifact-specific gates:

- Lean/Lake proof check under the declared PAX axiom basis.
- CUDA/PTX compiler output for the target architecture.
- Runtime comparison against a functional reference on the target hardware.
- License path selection and node-key/seal policy, when production use applies.

## GitHub Topics

Recommended repository topics for v1.0.0:

- `pax-coder`
- `formal-verification`
- `lean4`
- `cuda`
- `ptx`
- `futhark`
- `gpu-kernels`
- `proof-carrying-code`
- `tensor-cores`
- `ampere`
- `deepseek-coder`
- `qlora`
- `sovereign-compute`
- `verified-kernels`
- `model-training`

## Release Artifact

The GitHub release should use tag `v1.0.0`.

Release assets are the automatic source archives generated by GitHub unless a
separate model artifact, GGUF file, dataset export, or signed binary package is
explicitly attached later.

## v1.0.0 Packaging Evidence

Observed on Windows:

```text

Total unique examples: 10

train: 9 examples

val: 0 examples

test: 1 examples

```

Generated `build/` outputs are package build products and are not part of the
source release unless explicitly attached as release assets.