Text Classification
Transformers
Safetensors
English
nli
cross-encoder
qwen3.5
reranker
image-text-to-text
Instructions to use AlexWortega/openjev with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use AlexWortega/openjev with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="AlexWortega/openjev")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("AlexWortega/openjev", device_map="auto") - Notebooks
- Google Colab
- Kaggle
File size: 1,694 Bytes
26de23c | 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 | """Install NVIDIA's CUDA compiler redistributables into a private directory.
No driver/system modifications. Archives are hash-verified against NVIDIA's
versioned redistribution manifest before extraction.
"""
import argparse
import hashlib
import json
from pathlib import Path
import tarfile
import urllib.request
ap = argparse.ArgumentParser()
ap.add_argument("root")
ap.add_argument("--version", default="12.6.3")
a = ap.parse_args()
root = Path(a.root).resolve()
root.mkdir(parents=True, exist_ok=True)
url = "https://developer.download.nvidia.com/compute/cuda/redist/"
manifest = json.load(urllib.request.urlopen(url + f"redistrib_{a.version}.json"))
archives = root.parent / "cuda_archives"
archives.mkdir(exist_ok=True)
for package in ("cuda_nvcc", "cuda_cudart", "cuda_cccl", "cuda_nvrtc"):
info = manifest[package]["linux-x86_64"]
archive = archives / Path(info["relative_path"]).name
if not archive.exists():
urllib.request.urlretrieve(url + info["relative_path"], archive)
if hashlib.file_digest(archive.open("rb"), "sha256").hexdigest() != info["sha256"]:
raise RuntimeError(f"Checksum mismatch: {archive}")
with tarfile.open(archive) as source:
for member in source.getmembers():
parts = Path(member.name).parts
if len(parts) < 2:
continue
member.name = str(Path(*parts[1:]))
source.extract(member, path=root, filter="data")
print(f"Installed {package}: {info['sha256']}", flush=True)
(root / f"redistrib_{a.version}.json").write_text(json.dumps(manifest, indent=2))
lib64 = root / "lib64"
if not lib64.exists():
lib64.symlink_to("lib", target_is_directory=True)
|