This release is a pre-release and may not be stable for production use.
OVOS Model2Vec Intent Pipeline
An intent-matching pipeline plugin for OpenVoiceOS, built on Model2Vec static embeddings. It classifies an utterance against the intents that loaded skills registered at runtime, either through a pre-trained classifier head or through the skills' own example utterances with no training step. Use it as a semantic fallback behind exact matchers such as Adapt and padacioso.
Install
uv pip install --prerelease=allow ovos-m2v-pipeline
Configure
{
"intents": {
"ovos-m2v-pipeline": {
"model": "OpenVoiceOS/ovos-m2v-intents-multilingual"
},
"pipeline": [
"ovos-padacioso-pipeline-plugin-high",
"ovos-m2v-pipeline-high",
"ovos-m2v-prototype-pipeline-medium"
]
}
}
intents.pipeline replaces the whole pipeline list, not only its m2v
entries; see the pipeline order
for where the other plugins a deployment installs belong in it.
ovos-padacioso-pipeline-plugin-high runs first: a frozen classifier only
answers with a label it was trained on, so an exact template line of a
skill it never saw is claimed earlier, and correctly, by an exact matcher.
See the pipeline order
for the full reasoning.
Tiers
| Entry-point suffix | Method | Answers from |
|---|---|---|
-high |
match_high() |
the trained head, or the skill's own examples for the standalone prototype plugin |
-medium |
match_medium() |
same, at a lower threshold |
-low |
match_low() |
the skill's own examples by default (low_tier: "prototype") |
Documentation
| Page | Covers |
|---|---|
| Configuration | Every config key, its default and its type |
| OVOS Pipeline Plugin | Entry points, tiers, modes, caching |
| Models | Published model ids and picking one |
| Training | Building the corpus and fitting a model |
| Label scheme | Label format, families, dedup and renames |
| Pre-release quirks | Behaviour changes by pre-release version, reset at each stable release |
Training your own model starts at Training.
License
Apache 2.0. See LICENSE.
Credits
First built by TigreGótico for OpenVoiceOS under the ILENIA project, and extended through the NGI0 Commons Fund.
2 maintainers
Release files for ovos-m2v-pipeline 0.28.0a1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| ovos_m2v_pipeline-0.28.0a1.tar.gz | 181.2 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| ovos_m2v_pipeline-0.28.0a1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 246.7 kB
Release files / ovos_m2v_pipeline-0.28.0a1.tar.gz
| Download URL | ovos_m2v_pipeline-0.28.0a1.tar.gz |
|---|---|
| Size | 181.2 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
7df30b36383424199bfed9f302d51cc373e3ff2e561296cd14c90ccd46df1a32
|
|
BLAKE2b-256 checksum How to use checksums |
50012b62dd1edf48a4bd8984448d1c04d113866d9cd7c8b385b184d17f8bd06d
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Release files / ovos_m2v_pipeline-0.28.0a1-py3-none-any.whl
| Download URL | ovos_m2v_pipeline-0.28.0a1-py3-none-any.whl |
|---|---|
| Size | 65.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
7c2768198812b06e4d60a4728d89b1c8943e2931301178728e22de08fbc09e19
|
|
BLAKE2b-256 checksum How to use checksums |
5bcfb9a5588d84c9b935696a2cb6e0c0a3d50fad22712594f3f789869d3aa3c9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
