Instructions to use noahmclaughlin/Mixed-Messenger-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use noahmclaughlin/Mixed-Messenger-4B with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download noahmclaughlin/Mixed-Messenger-4B --local-dir Mixed-Messenger-4B
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
File size: 1,146 Bytes
bfee871 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 | {
"name": "Mixed-Messenger-4B",
"kind": "ensemble",
"base_repository": "mlx-community/Qwen3-4B-Instruct-2507-4bit",
"base_revision": "50d427756c6b1b2fe0c0a10f67fbda1fc8e82c1b",
"weights": "softmax-head.npz",
"weights_sha256": "ed30965b7e41a823d5c9dd19c7638f2ddef88d9bdb7b3e74106deb114f9521d1",
"adapters": [
"adapter"
],
"base_prior_weight": 0.5,
"calibrated_weight": 0.5,
"features": [
"benchmark_layer18_mean",
"benchmark_layer36_mean"
],
"artifacts": {
"adapter/adapter_config.json": "4f7c90e032dc7d08692a8b21df5be8ce73c54e4ef05ceae713dfea7ec93646bd",
"adapter/adapters.safetensors": "70bfcc189507b62d5d874d79e28094c4dd8deca15fe872b780eabe9b26309579",
"multiview-head.npz": "c10c7842a362ac2c1309142ec5be9d459cf9979ff834a6d71a24b8923afcc775",
"softmax-head.npz": "ed30965b7e41a823d5c9dd19c7638f2ddef88d9bdb7b3e74106deb114f9521d1"
},
"selection": "lora-30+multiview",
"description": "Equal probability average of a reference-trained two-view residual readout and a probability-LoRA softmax predictor. Original Qwen greedy probabilities feed the residual readout. No outcome lookup."
}
|