BARBET
OPENFORMOSA · PUBLIC RESEARCH RELEASE

Barbet 1B
Structured SFT

A text backbone for structured editing and evidence-grounded decisions.

One full SFT epoch on Traditional Chinese task data. Explore the exact requests, generated answers, measured gains, and observed failures before building on the checkpoint.

v0.1.1 · 2026.09.22 · Experimental
1.12BParameters
980,000Training examples
214/408Whole-answer exact
408/408Parseable responses
01 / Overview

Text in. Structured decisions out.

Work with ASR transcripts, TTS written and spoken forms, OCR candidates, and translation revisions. Requests include explicit task contracts and visible reference evidence. Audio encoders, image encoders and speech waveform generation are not included.

ASRTTS textOCRTranslation
02 / Predictions

Read what the model actually said.

Eight inputs fixed before inference on 19 September 2026. Four structured requests contain reference text; four plain instructions test a different use pattern. These are recorded outputs, not a live demo or a benchmark.

Complete input

Raw model output

Reference answer
03 / Evaluation

The gains and the cost.

Identical fixed synthetic test: 1,020 answer-loss examples and 408 generations per model. Whole-answer exact includes every field, including synthetic confidence. Parse success is not answer correctness.

DomainBase NLLSFT NLLSFT exactParse rate
ASR1.69940.002927/48100%
TTS1.64050.002719/32100%
OCR1.62310.002624/40100%
MT1.56860.002413/32100%

General-text retention

Existing 6,955-row development corpus. Bits per byte (BPB) measures text prediction loss; lower is better. These are relative loss changes, not accuracy-point changes.

CategoryBaseSFTBPB change
zh_tw_zh1.13991.1756+3.13%
math0.68900.8007+16.21%
ja_ko0.96800.9833+1.58%
english_general0.86150.8727+1.30%
code0.71540.9111+27.36%
multilingual1.13741.1679+2.69%
04 / Quick start

Run a complete structured request.

The loader preserves the original BF16 / FP32 parameter mix. The example’s assistant reference is never included in the prompt.

TERMINAL
git clone https://github.com/voidful/barbet-structured-sft.git
cd barbet-structured-sft
python -m venv .venv
source .venv/bin/activate
python -m pip install torch==2.11.0 packaging ninja setuptools wheel
python -m pip install --no-build-isolation -r requirements-inference.txt
python scripts/infer.py --example ocr_correction --device cuda:0

Tested: Python 3.12 · PyTorch 2.11 · CUDA 13.0 · H200. Native Mamba packages require a compatible CUDA build environment. Short CPU inference is supported.

05 / Documentation

Source, evidence, and reproducibility.

Training covered every one of the 980,000 examples once, without truncation. Public inference and synthetic-data evaluation are documented. Exact retraining additionally needs the separately controlled base checkpoint; the historical retention corpus is not distributed.

License status

Apache-2.0 covers the project code, released SFT weights, documentation and bundled examples. Third-party components and the separately published full training dataset retain their own terms. LICENSE ↗