-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathNOTICE
More file actions
73 lines (61 loc) · 3.14 KB
/
Copy pathNOTICE
File metadata and controls
73 lines (61 loc) · 3.14 KB
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
nullpii
Copyright 2026 nullpii contributors
This product includes software developed by the nullpii contributors.
Licensed under the Apache License, Version 2.0 (see LICENSE).
------------------------------------------------------------------------
Third-party attributions
------------------------------------------------------------------------
This product depends on the following third-party components.
Run `npm run license-check` for the full machine-readable list.
Runtime (npm):
- @anush008/tokenizers MIT (NAPI bindings to HF tokenizers)
- any-ascii ISC (Unicode → ASCII transliteration)
- debug MIT
- onnxruntime-node MIT (peer, optional)
- commander MIT (CLI)
- chalk MIT (CLI)
- cli-progress MIT (CLI)
Build & dev:
- typescript Apache-2.0
- vitest MIT
- @vitest/coverage-v8 MIT
- @biomejs/biome MIT or Apache-2.0
- madge MIT
- license-checker BSD-3-Clause
- lefthook MIT
- tsx MIT
Research kit (`packages/eval/`, Python — gitignored, not published):
- huggingface-hub Apache-2.0
- transformers Apache-2.0
- gliner Apache-2.0
- torch BSD-3-Clause
- onnx Apache-2.0
- onnxruntime MIT
- datasets Apache-2.0
Model artifacts (downloaded at runtime from HuggingFace Hub — NOT bundled
in the npm package; see README "First-call download"):
Base model (Apache-2.0):
- urchade/gliner_multi_pii-v1 ........ GLiNER head on Microsoft
mDeBERTa-v3 base.
(Zaratiana et al., NAACL 2024.)
Published weights:
- lBroth/nullpii ..................... ONNX + tokenizer + GLiNER config
(~1.2 GB FP32). See the
HuggingFace model card for
licence, intended use, and
upstream attributions.
Bench / eval datasets (used only to measure F1, never bundled in the
npm package): see `packages/eval/datasets/README.md`. Each external
dataset retains its own licence.
License posture:
- 100% permissive runtime tree (MIT / Apache-2.0 / BSD / ISC / CC0).
Zero LGPL, GPL, AGPL, or SSPL components reachable from the published
npm package. Verified by `npm run license-check` in CI.
- The npm package ships only TypeScript/JS source + compiled `dist/`.
Model weights are fetched on first use from HuggingFace and are NOT
part of the npm artifact; their licence is governed by the
`lBroth/nullpii` model card (see above).
- Research kit (packages/eval/, gitignored) is not published. Its
Python deps include MPL-2.0 components (`certifi`, `tqdm` is
dual-licensed MIT). MPL-2.0 is file-level copyleft and imposes
no obligation on this project's own source.