Brief Details: A distilled version of Deepseek-R1 model optimized for GGUF format, featuring NSFW content filtering and roleplay capabilities.
Brief-details: Qwen2-VL-2B-Instruct GGUF is a vision-language model with various quantization options, optimized for CPU/GPU inference using llama.cpp, supporting image analysis and description tasks.
Brief-details: A DistilBERT-based model for detecting and classifying stereotypes in text across gender, race, profession, and religion categories, with both stereotype and anti-stereotype detection capabilities.
Brief Details: Falcon3-10B-Base is a 10B parameter foundation model supporting 4 languages with 32K context length, excelling in reasoning and math tasks.
Brief-details: A specialized LoRA model for generating Pokémon trainer sprites in pixel art style, offering 768x768 and 512x512 versions with downscaling capabilities for authentic 96x96 sprites
Brief Details: RWKV-4-Novel is a specialized language model series focused on novel generation, featuring variants for Chinese and English content, built on the innovative RWKV architecture that combines RNN efficiency with transformer-like performance.
BRIEF-DETAILS: An anime-styled model combining MouseyMix textures with realistic proportions, featuring enhanced backgrounds and lighting through multiple versions (V1-V5)
Brief-details: XLM-RoBERTa-large model fine-tuned for multilingual zero-shot classification, achieving 93.7% accuracy on XNLI-es and strong ANLI performance
BRIEF-DETAILS: Distilled version of BART-MNLI with 12 encoder and 9 decoder layers, achieving 89.56% matched accuracy while maintaining high performance through selective layer copying.
Brief-details: Sentence-transformer model based on RoBERTa-large, maps text to 768D vectors. Trained on ANLI/MNLI/SNLI datasets for semantic similarity tasks.
BRIEF DETAILS: PTT5: Portuguese-optimized T5 model with 220M parameters, trained on BrWac corpus. Features Portuguese vocabulary and supports both PyTorch/TensorFlow.
BRIEF DETAILS: GottBERT_base_last is a German RoBERTa model with 125M parameters, trained on OSCAR dataset. Achieves strong performance on NER (87.48% F1) and text classification tasks.
Brief-details: Amharic RoBERTa language model for NLP tasks, specializing in masked word prediction and natural language processing for the Amharic language
Brief Details: BioBERT-based NER model fine-tuned on NCBI disease dataset (793 PubMed abstracts). Specialized in identifying disease mentions in medical text.
BRIEF DETAILS: Chinese RoBERTA model fine-tuned for Named Entity Recognition (NER) on CLUENER2020 dataset. Optimized for identifying entities like companies and addresses in Chinese text with high accuracy.
Brief Details: Distilled Chinese GPT-2 model (6-layer) trained on CLUECorpusSmall dataset. Optimized for Chinese text generation with reduced parameters while maintaining performance.
BRIEF-DETAILS: GPT2-based Chinese lyrics generation model trained on 150k lyrics. Pre-trained using UER-py framework with 100k steps. Capable of continuing lyrical phrases in Chinese.
BRIEF DETAILS: GPT2-based Chinese couplet generator trained on 700k couplets. Pre-trained using UER-py framework with 25k steps. Specializes in generating matching Chinese poetry couplets.
Brief-details: Chinese RoBERTa variant (4-layer, 512-hidden) optimized for efficiency. Part of UER's miniature series, trained on CLUECorpusSmall with strong performance on Chinese NLP tasks.
Brief Details: Brazilian Portuguese financial sentiment analysis model based on BERTimbau, optimized for analyzing financial texts and market sentiment
Brief-details: Fine-tuned T5-3B model for text-to-SQL translation achieving 75.5% accuracy with PICARD decoding, trained on Spider dataset for zero-shot database queries.
