metadata
language:
- ru
pipeline_tag: text-classification
tags:
- emotion
- multi-label
- bert
- lora
- rubert-large
license: mit
datasets:
- seara/ru_go_emotions
metrics:
- f1
- accuracy
library_name: peft
RuBERT-large Emotion Detection (LoRA Fine-tuned)
Model trained by Ark2016. Base model: ai-forever/ruBert-large.
Metrics
- F1 Micro: 0.57
- Accuracy: 0.45
Usage
from peft import PeftModel, PeftConfig
from transformers import AutoModelForSequenceClassification, AutoTokenizer
import torch
repo_id = "Ark2016/ruBert-large-emotions-lora"
model = PeftModel.from_pretrained(AutoModelForSequenceClassification.from_pretrained("ai-forever/ruBert-large", num_labels=28, problem_type="multi_label_classification"), repo_id)
tokenizer = AutoTokenizer.from_pretrained(repo_id)