Instructions to use pratikgx/Clairic-Flash-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use pratikgx/Clairic-Flash-MLX with MLX:
# Make sure mlx-vlm is installed # pip install --upgrade mlx-vlm from mlx_vlm import load, generate from mlx_vlm.prompt_utils import apply_chat_template from mlx_vlm.utils import load_config # Load the model model, processor = load("pratikgx/Clairic-Flash-MLX") config = load_config("pratikgx/Clairic-Flash-MLX") # Prepare input image = ["http://images.cocodataset.org/val2017/000000039769.jpg"] prompt = "Describe this image." # Apply chat template formatted_prompt = apply_chat_template( processor, config, prompt, num_images=1 ) # Generate output output = generate(model, processor, formatted_prompt, image) print(output) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
File size: 2,735 Bytes
b729e44 | 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 69 70 71 72 73 74 75 76 77 78 79 | """Command line for Clairic on MLX."""
from __future__ import annotations
import argparse
import json
from pathlib import Path
from clairic_mlx.pipeline import build_demo
from clairic_mlx.weights import convert, load
SAMPLE_STATE = """Coffee House
12 Main Street
Latte 4.50
Muffin 3.25
Subtotal 7.75
Tax 0.64
Total 8.39
Date: 2026-03-14
"""
SAMPLE_FIELDS = {
"merchant": {"type": "text", "instructions": "The merchant name."},
"total": {"type": "number", "decimals": 2},
"issued": {"type": "date", "instructions": "The receipt date."},
"items": {
"type": "list",
"items": "text",
"max_items": 2,
"instructions": "The purchased item names, in order.",
},
}
def main(argv: list[str] | None = None) -> None:
parser = argparse.ArgumentParser(description="Run Cloudflare Clairic on Apple MLX")
sub = parser.add_subparsers(dest="command", required=True)
demo = sub.add_parser("demo", help="Run the receipt example on a tiny random model")
demo.add_argument("--state", default=SAMPLE_STATE)
demo.add_argument("--fields", type=Path, help="JSON object of field specs")
extract = sub.add_parser("extract", help="Extract fields with a local Clairic snapshot")
extract.add_argument("--model", required=True, help="Hugging Face id or local snapshot")
extract.add_argument("--state", required=True)
extract.add_argument("--fields", type=Path, required=True)
extract.add_argument("--quant", type=int, choices=(4, 8))
extract.add_argument("--reference-date")
extract.add_argument("--quantiles", action="store_true")
exported = sub.add_parser("convert", help="Write MLX weights, optionally quantized")
exported.add_argument("--model", required=True)
exported.add_argument("--out", type=Path, required=True)
exported.add_argument("--bits", type=int, choices=(4, 8))
args = parser.parse_args(argv)
if args.command == "demo":
fields = json.loads(args.fields.read_text()) if args.fields else SAMPLE_FIELDS
print(json.dumps(build_demo().extract(args.state, fields), indent=2, ensure_ascii=False))
return
if args.command == "extract":
fields = json.loads(args.fields.read_text())
extractor = load(args.model, quant_bits=args.quant)
response = extractor.extract(
args.state,
fields,
reference_date=args.reference_date,
include_quantiles=args.quantiles,
)
print(json.dumps(response, indent=2, ensure_ascii=False))
return
convert(args.model, args.out, bits=args.bits)
print(f"Wrote MLX weights to {args.out}")
if __name__ == "__main__":
main()
|