Zero-Shot Classification
MLX
Safetensors
qwen3_5
clef
cloudflare
systemone
structured-output
classification
multimodal
custom-code
decision-model
Eval Results
4-bit precision
Instructions to use mlx-community/clef-flash-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/clef-flash-4bit with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download mlx-community/clef-flash-4bit --local-dir clef-flash-4bit
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Commit History
-
Add Typed Decisions benchmark results
Verified signatureAdds this model's scores on [Typed Decisions](https://huggingface.co/datasets/LocalLLaMA/typed-decisions) as `.eval_results/typed-decisions.yaml`, so they are listed on the benchmark's Hub leaderboard. | Accuracy | KL from gold (lower is better) | Brier (lower is better) | ECE (lower is better) | |---|---|---|---| | 0.701 | 0.2126 | 0.1104 | 0.1204 | General model (any question schema at request time), not fitted by us on this benchmark's train split. Scored on the full 400-case test split (2,000 decisions): one request per case with all five questions, text only, the benchmark's own metrics, MLX on an Apple M3 Max. Once merged, the model appears on the leaderboard. Quantizations and other derived models are hidden by default; turn the base-model switch off to see them. While this PR is open the Hub marks the scores as community-provided. If you would rather not list the model, or a number looks wrong, close this PR or tell us and we will correct it. Scored by codelion (OptiQ).879c284
-
Add decision-model tag (#1)
- Add decision-model tag (3fc2a15ec29dc6a521f6dbc421870475d7f41301) Co-authored-by: Victor Mustar <victor@users.noreply.huggingface.co>3fb6bec
-
clef_mlx.py: use the repo name (not the cache snapshot hash) as the default model name
Verified signature140bf7e -
clef_mlx.py: default --model works inside Hub cache snapshots (symlinked files)
Verified signature686174e -
clef_mlx.py: add 'predict' CLI and 'serve' (local POST /v1/systemone server); card: CLI and server docs
Verified signature7c95239 -
Card: zero-shot-classification tag, RAM guide for all variants, limitations, pinned mlx-vlm; clef_mlx: warn on untested mlx-vlm/mlx-lm versions
Verified signature6822f0f -
Add Decision Index quality check; clef_mlx: strict no-truncation mode (truncate=False)
Verified signatured9ec324
-
Add MLX 4-bit conversion of Cloudflare/clef-flash with joint schema head and clef_mlx.py loader
Verified signature7cec35c -
initial commit
Verified signature3a3b4b2