Downloads · 30 days
3
6% of all-time downloads
S7-relevance-rubert-20260823
DanielNRU/S7-relevance-rubert-20260823
S7-relevance-rubert-20260823 is a text classification model from DanielNRU. Use it when you need a label for a piece of text. It is set up for pytorch.
Бинарная классификация релевантности русскоязычных текстов для домена S7.
All-time downloads
50
Public
Parameters
356M
2.8 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin1.4 GB · 50%
At a glance
- Task
- Text Classification
- Library
- pytorch
- Access
- Public
- Created
- Aug 23, 2026
- Updated
- Aug 23, 2026
- SHA
- 86d569a4
Try a prompt
Base models
Model card
From the Hugging Face model README
S7-relevance-rubert-20260823
Бинарная классификация релевантности русскоязычных текстов для домена S7.
Model details
- Base model:
ai-forever/ruRoBERTa-large - Architecture:
RuBERTBinaryClassifier - Best validation F1:
0.978417 - Test F1:
0.971429 - Test precision:
0.951049 - Test recall:
0.992701 - Decision threshold:
0.5 - Maximum sequence length:
512 - Training class ratio (relevant:irrelevant):
3.0:1
Usage
from huggingface_hub import hf_hub_download
from transformers import AutoTokenizer
import torch
repo = "DanielNRU/S7-relevance-rubert-20260823"
tokenizer = AutoTokenizer.from_pretrained(repo)
model = RuBERTBinaryClassifier("ai-forever/ruRoBERTa-large", hidden_size=512)
weights = hf_hub_download(repo_id=repo, filename="model.safetensors" if (output_dir / "model.safetensors").exists() else "pytorch_model.bin")
model.load_state_dict(torch.load(weights, map_location="cpu"))
model.eval()
Limitations
Модель предназначена для S7-домена. При изменении распределения данных рекомендуется повторно проверить threshold и метрики.
- Task
- Text Classification
- Library
- pytorch
- Languages
- ru
- Created
- Aug 23, 2026
- Updated
- Aug 23, 2026