Skip to content

Releases: IamBusy/OpenJev-Vision

Release list

OpenJev v0.3.2 — consistent model naming

OpenJev v0.3.2 — consistent model naming Pre-release
Pre-release

Choose a tag to compare

View all tags
@IamBusy IamBusy released this 19 Sep 10:54
v0.3.2
5a827c5

This release separates the public model name from internal experiment labels and software versions. The model is now OpenJev-0.6B across Hugging Face, documentation, API responses and download bundles.

  • Public model: IamBusy/OpenJev-0.6B.
  • Software version: 0.3.2. Trained model revision: 0.3.0.
  • Public SDK: from openjev import OpenJevModel.
  • Current-model CLI: openjev-model.
  • Bundle: OpenJev-0.6B.zip, with hashes in SHA256SUMS. Base Qwen weights are downloaded separately.

Load

pip install 'openjev[qwen] @ git+https://github.com/IamBusy/OpenJev.git@v0.3.2'
from openjev import OpenJevModel
model = OpenJevModel.from_pretrained(
    "IamBusy/OpenJev-0.6B"
)

Compatibility and validation

The former Hugging Face address redirects to the new name. Existing openjev-branch commands, implementation imports and historical releases remain usable. Local inference prefers artifacts/openjev-0.6b and can fall back to the earlier checkpoint directory.

The trained adapter, scalar head and calibration bytes are unchanged. This is a naming, interface and packaging update; the existing model limitations and experimental status still apply.

32 local tests passed. Canonical and legacy CLI predictions match, model identifiers are validated through the HTTP API, and local checkpoint fallback is verified. Replay tests now keep generated reports in a temporary workspace.

Naming and version rules · Model card · Measured results

Loading

OpenJev v0.3.0 — experimental shared-state decisions

OpenJev v0.3.0 — experimental shared-state decisions Pre-release
Pre-release

Choose a tag to compare

View all tags
@IamBusy IamBusy released this 19 Sep 09:37
v0.3.0
2bf9fc9

First public experimental release of OpenJev: local typed probabilistic decisions with Qwen3-0.6B, a shared state prefix, isolated candidate branches and a learned scalar scoring head.

Included

  • Reproducible data preparation, training, development-only checkpoint selection, calibration and evaluation.
  • English and Chinese documentation, model card, source attribution and contribution guidance.
  • CPU CI, cache output/gradient equivalence checks and safe checkpoint downloads.
  • Historical v0.1/v0.2 baselines and complete measured regressions.

Quick start

git clone https://github.com/IamBusy/OpenJev.git
cd OpenJev
uv sync --frozen --extra qwen --extra dev
uv run --no-sync openjev-branch download
uv run --no-sync openjev-branch predict --input examples/refund.json

The model ZIP contains the LoRA adapter, scalar head, calibration and provenance. Qwen base weights are downloaded separately from a pinned Hugging Face revision. No provider key is required for inference or frozen v0.3 data reconstruction. SHA256SUMS covers the attached model ZIP, wheel and source distribution.

Evidence and limitations

  • 31 local checks passed; source-only tests: 24 passed and 7 explicit integration skips. Public Linux CI passed.
  • All reconstructed public split hashes and the complete v0.3 manifest match the original experiment.
  • Fixed 60-question subset: v0.3 45/60, v0.2 39/60, DeepSeek 59/60.
  • Long synthetic state / 8 questions × 4 candidates: about 0.70 s versus v0.2's 1.38 s on the measured M3 Pro. Short single-question inputs are slower (104 ms versus 55 ms).
  • Some banking, access-policy and shipping tasks regress. Calibration is domain-specific. This is one seed with narrow data; architecture and data changed together.

OpenJev is independent of TypeSafe. It does not reproduce Jev's unpublished weights, architecture or RLCD and is not production-ready.

Full results · Model card · Reproduction guide

Loading