Brief Details: Spanish GPT-2 small model trained on Wikipedia data, featuring 70-hour training on 4 GPUs. Optimized for Spanish text generation with Apache 2.0 license.
Brief Details: DeiT-tiny is a lightweight Vision Transformer with 5M params, achieving 72.2% ImageNet accuracy. Efficient training approach for vision tasks.
BRIEF DETAILS: PoolFormer-based vision model with 56.2M params, trained on ImageNet-1k. Efficient MetaFormer architecture for image classification and feature extraction.
Brief-details: CogVideoX-2b is an open-source text-to-video diffusion model offering 720x480 video generation at 8fps, optimized for low VRAM usage starting from 4GB with FP16 precision.
BRIEF-DETAILS: ConViT Base model with 86.5M parameters for image classification, trained on ImageNet-1k. Features soft convolutional inductive biases for improved vision transformer performance.
Brief Details: Spanish BERT model trained on large Spanish corpus using Whole Word Masking, achieving SOTA performance on various Spanish NLP benchmarks.
Brief-details: A powerful sentence transformer model that maps text to 1024-dimensional vectors, optimized for semantic search and similarity tasks with 55K+ downloads.
BRIEF DETAILS: A voice gender classification model leveraging ECAPA-TDNN architecture, achieving 98.7% accuracy on VoxCeleb1, with 15.5M parameters and MIT license.
BRIEF DETAILS: Vision-friendly transformer model with 40.2M parameters, trained on ImageNet-1k. Optimized for image classification with efficient architecture and 224x224 input size.
Brief-details: Llama-3 based 8B parameter model optimized for German/English, featuring two-stage DPO fine-tuning and strong multilingual capabilities
Brief-details: Japanese to English translation model using Marian-NMT architecture, achieving 39.1 BLEU score on Tatoeba test set. Popular with 56K+ downloads.
Brief-details: Greek speech recognition model based on wav2vec2-large-xlsr-53, achieving 11.62% WER and 3.36% CER on Common Voice test set. Ideal for ASR tasks.
Brief-details: RoBERTa-based spam detection model with 125M parameters. Achieves 99.06% accuracy for text classification. MIT licensed, trained on merged spam datasets.
Brief-details: Efficient image classification model with 4.46M params, trained on ImageNet-1k using RMSProp. Features hardware-efficient architecture designed through neural architecture search.
Brief-details: ReXNet-100 is a lightweight CNN with 4.8M params, optimized for efficient image classification on ImageNet-1k with 77.8% top-1 accuracy.
Brief Details: Inception-v3 image classification model with 23.9M params, trained on ImageNet-1k. Features 299x299 input size with 5.7 GMACs compute.
Brief Details: Microsoft's BioGPT - A specialized biomedical language model achieving SOTA results on medical NLP tasks with 78.2% accuracy on PubMedQA
Brief-details: HerBERT is a BERT-based Language Model trained specifically for Polish text, featuring MLM and SSO objectives with dynamic word masking and 50k vocabulary.
Brief-details: High-quality multilingual text-to-speech model supporting various English accents (US, UK, Indian, Australian) plus Spanish, French, Chinese, Japanese, and Korean voices.
Brief-details: A powerful 334M parameter text embedding model optimized for retrieval tasks, achieving SOTA performance with 55.98 NDCG@10 on MTEB benchmark.
Brief Details: FBNet-based image classification model with 5.6M params, trained on ImageNet-1k using RMSProp. Efficient architecture for 224x224 images.
