Brief-details: 4x_NMKD-Siax_200k is an AI upscaling model designed for image super-resolution, offering 4x upscaling capabilities with NMKD architecture trained on 200k iterations.
BRIEF-DETAILS: Portuguese legal BERT model fine-tuned from neuralmind/bert-base-portuguese-cased, achieving 0.644 loss, optimized for legal text processing
Brief Details: BLOOM intermediate checkpoints - Large-scale multilingual model with 176B parameters, trained on 46+ languages. Designed for research and non-commercial use.
BRIEF-DETAILS: State-of-the-art English-to-Vietnamese neural machine translation model developed by VinAI, published in INTERSPEECH 2022.
Brief-details: State-of-the-art Vietnamese-to-English neural machine translation model by VinAI, published in INTERSPEECH 2022, focused on high-quality bidirectional translation
Brief-details: Chinese PEGASUS model (238M params) fine-tuned on 7 Chinese summarization datasets. Achieves ROUGE-1/2/L scores of 43.46/29.59/39.76 on LCSTS benchmark. Optimized for abstractive summarization.
Brief-details: Spanish medical NER model detecting UMLS entities (anatomy, chemicals, disorders, procedures) with 86.47% F1-score. Built for clinical trials text analysis.
BRIEF-DETAILS: Large-scale Mandarin speech recognition model (120M params) using Conformer-Transducer architecture. Achieves 5.3-5.7% WER on AISHELL-2. Supports 16kHz mono audio.
Brief-details: A U-Net based denoising diffusion model trained on Oxford Flowers 102 dataset, capable of generating flower images through iterative denoising of Gaussian noise. Optimized for educational purposes.
Brief-details: A fine-tuned Wav2vec2 model for Farsi ASR, trained on 108 hours of CommonVoice data with 5gram KenLM integration, achieving 6% WER on cleaned test sets.
Brief Details: Japanese DeBERTa(V2) model trained on Wikipedia and Aozora Bunko texts, optimized for NLP tasks like POS-tagging and parsing.
Brief Details: Unity ML-Agents Walker model - A PPO-based reinforcement learning agent trained to master bipedal locomotion in Unity's physics environment.
Brief-details: BERT model fine-tuned on QQP dataset achieving 91% accuracy, optimized for question pair similarity tasks with strong F1 score of 0.8788
GENA-LM: Pre-trained transformer model for long DNA sequences (up to 4500 nucleotides), utilizing BPE tokenization and trained on T2T human genome assembly
Brief-details: Fine-tuned Swin Transformer model for food image classification, achieving 92.14% accuracy on Food101 dataset. Based on microsoft/swin-base-patch4-window7-224.
BRIEF-DETAILS: Speech-to-speech translation model based on HuBERT architecture, specialized in multilingual audio processing with 1000 discrete speech units
Brief-details: NEZHA is a Chinese language understanding model that provides neural contextualized representations, utilizing BERT tokenization with specialized NEZHA architecture.
Brief-details: NEZHA is a Chinese language understanding model that uses neural contextualized representations, utilizing BERT tokenization with specialized architecture for Chinese text processing.
Brief Details: A 3B parameter GGUF-quantized LLM based on Llama-3.2, fine-tuned by Nous Research for improved reasoning, conversation, and function calling capabilities.
Brief Details: A machine learning model by rohan57, hosted on HuggingFace. Limited public information available. Requires further documentation for comprehensive understanding.
Brief-details: Qwen2.5-Coder-14B-Instruct-AWQ is a 4-bit quantized code-specialized LLM with 14.7B parameters, supporting 128K context length and optimized for code generation and reasoning.
