Brief-details: A specialized climate-focused NLP model with 82.3M parameters, fine-tuned on DistilRoBERTa for detecting climate-related content in text paragraphs.
Brief Details: ALBERT-based Chinese POS tagging model, part of CKIP NLP tools suite. Optimized for traditional Chinese text processing with tiny parameter configuration. GPL-3.0 licensed.
BRIEF-DETAILS: BERT large Japanese model with 24 layers and 1024-dim hidden states, trained on Wikipedia using whole word masking and Unidic tokenization
Brief Details: ALBERT base model for Traditional Chinese NLP tasks, developed by CKIPLAB. Supports word segmentation, POS tagging, and NER. Uses BERT tokenizer with GPL-3.0 license.
Brief Details: CamemBERT-large: A powerful 337M-parameter French language model based on RoBERTa, trained on 135GB of CCNet text data for advanced NLP tasks.
Brief-details: Multilingual BERT-based model trained on 11 African languages with 126M parameters. Optimized for NLP tasks like classification and NER.
BRIEF-DETAILS: ALBERT-based NER model for Traditional Chinese text processing, developed by CKIP Lab. Features token classification capabilities and PyTorch integration. GPL-3.0 licensed.
Brief-details: A fine-tuned Wav2Vec2-Large-XLSR model for Basque speech recognition, achieving 12.44% WER on Common Voice test set, built on XLSR-53 architecture.
Brief-details: Clinical diagnosis prediction model based on BioBERT, fine-tuned for ICD9 code prediction from medical admission notes. Supports 9,237 label predictions with specialized medical training.
BRIEF DETAILS: GPT2 model fine-tuned for Persian language generation with 256 token context window, using sentence piece tokenization and special token handling for non-Persian characters.
Brief-details: VetBERT is a specialized BERT model pretrained on 15M+ veterinary clinical records, fine-tuned for veterinary medicine NLP tasks with Bio+Clinical BERT initialization.
Brief Details: BlueBERT specialized for biomedical NLP, trained on PubMed abstracts with 4000M words. Large-scale BERT variant (24 layers, 1024 hidden) optimized for medical text processing.
BRIEF DETAILS: Lightweight 8-layer BERT model (42.8M params) specialized for biomedical text mining, offering 3x faster processing than BERT-base with comparable performance.
Brief Details: Korean intent classification model based on KLUE RoBERTa-small, fine-tuned on 3i4k dataset with 68.1M params, achieving 90% accuracy across 7 intent classes.
Brief Details: BERT-based emotion classification model trained on Go Emotions dataset achieving 96.1% accuracy. Supports multi-label classification for emotional analysis.
Brief Details: KcBERT-base is a Korean BERT model trained on news comments data with 110M parameters, optimized for informal text and social media content
Brief-details: A Chinese RoBERTa model using WECHSEL initialization for efficient cross-lingual transfer, achieving 78.32% NLI and 80.55% NER scores.
Brief Details: StructBERT large model extending BERT architecture with language structure pre-training, 340M parameters, optimized for GLUE benchmarks.
Brief Details: Russian song lyrics generator based on ruGPT-3, fine-tuned on genius.com lyrics. 176M parameters, supports multiple Russian artists including Oxxxymiron and Morgenshtern.
Brief-details: AraGPT2-Mega: A 1.51B parameter Arabic language model trained on 77GB of text data. Features advanced text generation capabilities with TPU optimization.
Brief Details: HeBERT is a Hebrew BERT-based language model trained on 10.6GB of data, specialized in polarity analysis and emotion recognition with robust sentiment classification capabilities.
