# Low-parameter efficiency

A compact and efficient speech-language model designed for Automatic Speech Recognition (ASR) and Automatic Speech Translation (AST), featuring a two-stage design for processing audio and text

A multilingual visual retriever combining Qwen2 vision model with stella_en_1.5B_v5, ranked first among models with ≤2B parameters in Vidore benchmark

TIGER is a lightweight speech separation model that achieves efficient audio processing through frequency band segmentation and multi-scale feature extraction

A fine-tuned Llama 3.2 (3B) model specialized in generating concise summaries of conversations.

A pre-trained Amharic language model based on the bert-medium architecture, with 40.5 million parameters trained on 290 million tokens, achieving performance comparable to larger multilingual models.

This is a 42M-parameter Llama 2 architecture float32 precision model trained on the TinyStories dataset, suitable for simple text generation tasks.

A 15-million-parameter Llama 2 architecture model trained on the TinyStories dataset

This is a Hugging Face transformers-compatible version conversion of the 125M parameter dense model from Artetxe et al.'s paper