Brief-details: A multilingual translation model supporting English to Romance languages (French, Spanish, Portuguese, Italian, Romanian, etc.) with strong BLEU scores for Latin translation (50.1).
Brief-details: BERT-based embedding model with 109M parameters optimized for English text similarity and retrieval tasks, achieving strong performance on MTEB benchmarks
Brief Details: OPT-2.7B is Meta AI's open-source language model with 2.7B parameters, trained for text generation and designed as a more accessible alternative to GPT-3.
Brief Details: GPTBigCode model for code generation in Python/Java/JavaScript. 1.12B params, trained on GitHub data. Strong performance on code completion tasks.
Brief Details: Indonesian RoBERTa-based sentiment classifier with 124M params, achieving 94.36% accuracy on evaluation. MIT licensed, specialized for Indonesian text analysis.
Brief-details: A powerful 4.15B parameter multimodal LLM combining vision and language capabilities, excelling in document understanding, OCR, and visual reasoning tasks
BRIEF DETAILS: A compact Russian conversational language model with 107M parameters, trained on social media data. Offers 2x faster inference than base model.
Brief-details: Multilingual BERT model fine-tuned for efficient prompt compression, featuring 177M parameters and task-agnostic capabilities across languages.
Brief-details: Multilingual speech recognition model for 6 Ugandan languages, built on Facebook's MMS-1B-ALL with code-switching support and wav2vec2 architecture.
Brief-details: XLNet large-cased model trained on BookCorpus and Wikipedia, featuring generalized autoregressive pretraining for advanced language understanding tasks. MIT licensed.
Brief Details: SOLAR-10.7B uncensored GGUF model - High-performance instruction-tuned LLM with 10.7B parameters, optimized for various quantization levels and GPU acceleration
Brief Details: Greek text summarization model based on mT5-small, optimized for abstractive summarization of Greek language content with 36.8K downloads.
Brief-details: polyBERT is a specialized chemical language model for polymer informatics, mapping PSMILES strings to 600D fingerprints, built on DeBERTa-v2 architecture
Brief-details: A powerful speaker verification model using x-vector embeddings, trained on VoxCeleb datasets. Achieves 3.2% EER, built with SpeechBrain for speaker identification tasks.
Brief-details: Robust speech recognition model with 315M parameters, fine-tuned on LibriSpeech. Handles multi-domain audio processing with 16kHz sampling rate support.
Brief-details: A pre-trained language model for dense passage retrieval using a bottleneck architecture, achieving 43.8 MRR@10 on MS-MARCO with efficient representation learning.
Brief Details: Llama-3-Open-Ko-8B is an 8B parameter Korean language model based on Meta's Llama-3, trained on 60GB+ text with 17.7B+ tokens
Brief Details: CodeGen-350M-mono is a 350M parameter AI model specialized in program synthesis, pre-trained on Python code and optimized for generating executable code from natural language prompts.
Brief-details: A powerful vision-language model by Qwen supporting multilingual image understanding, caption generation, and visual QA with state-of-the-art performance across multiple benchmarks.
Brief-details: KcELECTRA-base is a Korean ELECTRA model trained on user-generated content, optimized for noisy text processing with 109M parameters and MIT license
Brief-details: Efficient T5-based time series forecasting model with 47.7M params, offering zero-shot predictions and up to 250x faster inference than original Chronos models.
