Brief-details: Spanish to Chinese translation model by Helsinki-NLP achieving 38.8 BLEU score, supports multiple Chinese variants including Mandarin, Cantonese, and Classical Chinese.
Brief-details: Estonian to English translation model using transformer architecture, achieving 30.3 BLEU score on newstest2018, developed by Helsinki-NLP.
Brief-details: Neural machine translation model for Bulgarian to Finnish conversion using transformer architecture. Achieves 23.7 BLEU score on JW300 dataset.
BRIEF DETAILS: Helsinki-NLP's English-to-Marathi translation model using transformer architecture. BLEU score: 22.0, built on OPUS dataset with apache-2.0 license.
Brief-details: A Hindi speech recognition model based on wav2vec2 architecture, achieving 17.4% WER on Common Voice 7.0, with Apache 2.0 license and comprehensive language support.
Brief Details: A specialized text classification model for detecting fiction/non-fiction in historical book titles, supporting 30 languages with 65.8M parameters.
Brief-details: JavaBERT is a 110M parameter BERT-based model specialized for Java code analysis, trained on ~3M Java files from GitHub, supporting masked language modeling tasks.
Brief-details: A Spanish GPT-2 model trained on 11.5GB of text data, including Wikipedia and literature, featuring a custom BPE tokenizer optimized for Spanish language generation.
BRIEF-DETAILS: Korean language chatbot model based on KoGPT2, trained on AIHub data. Specialized for youth interactions with custom tokenization and generation parameters.
Brief-details: A fine-tuned DistilBERT model for sentence compression achieving 89.12% accuracy, with strong F1 (0.8367) and precision (0.8495) scores. Apache 2.0 licensed.
Brief Details: Clinical BERT model for automated ICD-10 code prediction from medical text. Built on BERT/BioBERT architecture with MIMIC notes training.
Brief-details: RoBERTa-based detector model (356M params) fine-tuned to identify GPT-2 generated text with 95% accuracy. Built by OpenAI for synthetic text detection.
Brief-details: Two-stage Mistral-based model optimized for creative writing and roleplay, featuring high LoRA dropout training and TIES merge method. Built on Mistral-Nemo-Instruct-2407.
Brief-details: Advanced text-to-image SDXL model using v-prediction, trained on Danbooru and e621 datasets. Optimized for artistic generation with specific parameter requirements.
Brief-details: Quantized 14B parameter LLM with multiple GGUF variants optimized for different performance/quality tradeoffs. Features imatrix quantization for enhanced efficiency.
Brief Details: Lightweight 1.54B parameter LLM optimized for Russian/English text generation, featuring GGUF quantization for efficient deployment and strong performance metrics.
Brief-details: An 8B parameter conversational AI model available in multiple GGUF quantization formats, optimized for different performance/size tradeoffs
Brief-details: OLMo-2-1124-13B Instruction model optimized for GGUF format with multiple quantization options (13.7B params). Features high-quality compression variants suitable for different hardware configurations.
Brief-details: An 8.03B parameter GGUF model with multiple quantization options, optimized for efficient deployment and featuring imatrix quantization techniques for enhanced performance.
Brief-details: 8B parameter GGUF quantized model optimized for efficient inference, offering multiple compression variants from 2.1GB to 6.7GB with imatrix quantization.
Brief Details: A 14B parameter GGUF-quantized language model offering multiple compression variants, optimized for efficient deployment while maintaining quality.
