Text Classification
MLX
Safetensors
English
qwen3_5
qwen3.5
decision
unofficial-port
experimental
4bit
4-bit precision
Instructions to use cowWhySo/jeeves-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use cowWhySo/jeeves-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir jeeves-mlx cowWhySo/jeeves-mlx
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
File size: 2,250 Bytes
6f4220a | 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 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 | {
"variant": "jeeves-mlx-4bit",
"identity": "c4d0ea30e584dbf0d1e672c59c2e00036cb6a1a7025833ef49c01c63432b18fa",
"conversion": "weights_converted_and_structurally_checked",
"format_head_checks": {
"status": "passed",
"format_cases": 4,
"original_head_dtypes": {
"q.weight": "torch.float32",
"q.bias": "torch.float32",
"k.weight": "torch.float32",
"k.bias": "torch.float32"
},
"exported_head_dtype": "float32",
"temperature": 1.8589280843734741,
"max_head_probability_error": 0.0,
"scope": "Original encoder token/readout parity, safe head serialization, and original PyTorch head vs MLX head on synthetic hidden states. Not backbone parity."
},
"cpu_smoke": {
"status": "passed",
"identity": "c4d0ea30e584dbf0d1e672c59c2e00036cb6a1a7025833ef49c01c63432b18fa",
"device": "cpu",
"cases": [
{
"id": "case0",
"probabilities": [
0.016998127102851868,
0.9830018877983093
],
"elapsed_seconds": 4345.674271831
}
],
"fixture_sha256": "2887c941fd5a497ae51343a9e2382ea1d198ab5eb140f4e8c8fd37c54be4ebf0",
"scope": "Small no-thinking execution smoke test; not a benchmark."
},
"bf16_backbone_parity": {
"status": "small_fixture_check_passed",
"cases": [
{
"id": "case0",
"maximum_probability_delta": 0.00010848045349121094,
"minimum_hidden_cosine": 0.9999632468948669,
"top_option_agreement": true,
"within_diagnostic_tolerances": true
}
],
"probability_atol": 0.02,
"minimum_cosine": 0.995,
"scope": "Heuristic diagnostic tolerances, not bitwise equivalence or general accuracy evidence."
},
"apple_metal_validation": "not_run",
"task_accuracy_calibration_benchmark": "not_run",
"speculative_drafters": "not_ported",
"greedy_generation_validation": "not_run",
"quantization_vs_mlx_bf16": {
"cases": [
{
"id": "case0",
"maximum_probability_delta": 0.016297996044158936,
"top_option_agreement": true
}
],
"status": "measured_not_quality_certified"
},
"overall_status": "converted_and_small_cpu_diagnostics_passed_not_benchmarked",
"diagnostic_failures": []
}
|