supra-50m-instruct-fp16-mlx
MLX quantization of SupraLabs/Supra-50M-Instruct for Apple Silicon.
Variant: BFloat16 (lossless reference)
Disk size: 201 MB
Quantized by: sahilchachra
Benchmark results
Evaluated on Apple M4 Pro with MLX. Model loaded once; performance and quality measured in a single pass.
Performance
| This model | FP16 baseline | |
|---|---|---|
| Decode tok/s (avg, long traces) | 1270.13 | 1270.13 |
| Peak memory (GB) | 0.223 | 0.223 |
| Disk size (MB) | 201 | 201 |
Quality
| Benchmark | This model | FP16 baseline | n |
|---|---|---|---|
| IFEval (instruction following) | 15.9% | 15.9% | 44 |
| Alpaca-cleaned (instruct F1 vs reference) | 36.2 | 36.2 | 50 |
Context scaling (decode tok/s)
| Context length | Decode tok/s |
|---|---|
| ~128 tokens | 1294.4 |
| ~256 tokens | 1274.3 |
| ~512 tokens | 1277.6 |
| ~1024 tokens | 1234.2 |
Usage
pip install mlx-lm
from mlx_lm import load, generate
model, tokenizer = load("sahilchachra/supra-50m-instruct-fp16-mlx")
response = generate(model, tokenizer, prompt="Your prompt here", max_tokens=256, verbose=True)
All variants in this collection
| Model | Variant |
|---|---|
| sahilchachra/supra-50m-instruct-8bit-mlx | Affine int8 |
| sahilchachra/supra-50m-instruct-optiq-5bpw-mlx | OptiQ mixed-precision (target 5.0 bpw) |
Notes
- Requires Apple Silicon (M1 or later) with MLX
- Benchmarks run on Apple M4 Pro, 24 GB unified memory
- License: see SupraLabs/Supra-50M-Instruct for the original model's license
Original model
See SupraLabs/Supra-50M-Instruct for full model details and intended use.