Saina Helm 3 MLX

MLX build of run-saina/saina-helm-3-0.8b (model id saina-helm-3-0.8b) for Apple silicon. Helm 3, natively multimodal: it reads text, images and audio together in one pass and answers typed questions with a probability for every option. No text is generated.

FP16 weights, loaded with stock mlx-vlm, plus Helm's FP32 answer head. score.py runs the model the way Helm reads its prompt, builds the prompt with the saina package and applies the head. mlx_vlm.generate can load these weights but can't produce Helm's answers; use score.py.

Benchmarks

Saina Helm 3 scores 45.89 on the Decision Index 0.3 (out of 100, chance-corrected), measured on run-saina/saina-helm-3-0.8b with the saina runtime.

Usage

pip install "mlx-vlm>=0.7.8" "saina[tokenize,media]>=0.9.1"
hf download run-saina/saina-helm-3-0.8b-MLX --local-dir helm3-mlx

import sys; sys.path.insert(0, 'helm3-mlx')
from score import Helm

helm = Helm('helm3-mlx')

helm.ask('I was charged twice for my subscription this month.',
         [{'question': 'Which team should handle this?', 'choices': ['Billing', 'Technical', 'Other']},
          {'question': 'Is this urgent?', 'choices': ['No', 'Yes']}])

helm.ask('A chess puzzle from an online game.',
         [{'question': 'Black to move. Which move is best?', 'choices': ['Rf7', 'h6', 'Rxf1', 'a5']}],
         images=['board.png'])           # PNG, JPEG or WebP, up to 8 images

helm.ask('Inbound call to the bank: a short recording of English speech.',
         [{'question': 'Is this speech synthesized or voice-converted by a computer (a spoof)?',
           'choices': ['No', 'Yes']}],
         audios=['call.wav'])            # WAV, FLAC, MP3, Ogg or WebM, up to 10 s each, up to 4 clips
# -> one probability list per question, in the order of `choices`

Each question takes choices and an optional mode: single_label (default; the probabilities sum to 1) or multi_label (one independent probability per choice). All questions are answered in one pass.

Files

License

Apache 2.0. Built on Qwen/Qwen3.5-0.8B.