Instructions to use botezx/flagtldr with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Grok
How to use botezx/flagtldr with Grok:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
File size: 1,861 Bytes
043267b | 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 | """Quick eval of a backend on a labelled JSONL (text, label) with the example labelset.
uv run python scripts/eval.py --backend rules
uv run python scripts/eval.py --backend basal --basal-url http://127.0.0.1:8000
"""
import argparse, asyncio, json, time
from pathlib import Path
from xaddons_classifier.backends import get_backend
from xaddons_classifier.spec import LabelSet, Post
ROOT = Path(__file__).resolve().parents[1]
async def main():
ap = argparse.ArgumentParser()
ap.add_argument("--backend", default="rules")
ap.add_argument("--basal-url", default="http://127.0.0.1:8000")
ap.add_argument("--labelset", default=str(ROOT.parent / "shared" / "labelset.example.json"))
ap.add_argument("--data", default=str(ROOT / "data" / "eval.sample.jsonl"))
ap.add_argument("--no-examples", action="store_true", help="zero-shot: drop examples from the labelset")
a = ap.parse_args()
ls = LabelSet.model_validate_json(Path(a.labelset).read_text())
if a.no_examples:
ls.examples = []
rows = [json.loads(l) for l in Path(a.data).read_text().splitlines() if l.strip()]
kw = {"url": a.basal_url} if a.backend == "basal" else {}
be = get_backend(a.backend, **kw)
posts = [Post(id=str(i), text=r["text"]) for i, r in enumerate(rows)]
t0 = time.time()
res = await be.classify(ls, posts)
dt = time.time() - t0
ok = 0
for i, r in enumerate(rows):
p = res[str(i)]
ok += p.label == r["label"]
mark = "✓" if p.label == r["label"] else "✗"
print(f"{mark} gold={r['label']:<13} pred={p.label:<13} conf={p.confidence:.2f} {r['text'][:60]}")
print(f"\nbackend={a.backend} examples={'no' if a.no_examples else len(ls.examples)} accuracy={ok}/{len(rows)} = {ok/len(rows):.2f} time={dt:.1f}s ({dt/len(rows)*1000:.0f} ms/post)")
asyncio.run(main())
|