# Multi-task Optimization

Ling is a large-scale Mixture of Experts (MoE) language model open-sourced by InclusionAI. The Lite version features 16.8 billion total parameters with 2.75 billion activated parameters, demonstrating exceptional performance.

OLMo 2 1B Instruct Edition is a post-training variant of the OLMo-2-0425-1B-RLVR1 model, optimized through supervised fine-tuning, DPO training, and RLVR training to achieve state-of-the-art performance across multiple tasks.

OLMo 2 1B is a post-training variant of the allenai/OLMo-2-0425-1B-RLVR1 model, undergoing supervised fine-tuning, DPO training, and RLVR training, aiming to achieve state-of-the-art performance across multiple tasks.

DeepSeek-R1 is the first-generation inference model launched by DeepSeek. Through large-scale reinforcement learning training, it performs excellently in mathematics, code, and reasoning tasks.

A text embedding model based on the ModernBERT pre-trained encoder, supporting long text processing up to 8192 tokens, with excellent performance on evaluation tasks such as MTEB, LoCO, and COIR.

Ruri is a model specialized in Japanese text embedding, capable of efficiently calculating sentence similarity and extracting text features.

SILMA-9B-Instruct-v1.0 is a 9-billion-parameter open-source Arabic large language model that excels in Arabic tasks, built on the Google Gemma architecture.

Beyonder-4x7B-v2 is a large language model based on the Mixture of Experts (MoE) architecture, consisting of 4 expert modules, each specializing in different domains such as dialogue, programming, creative writing, and mathematical reasoning.

A Hindi pre-trained language model based on the ELECTRA architecture, outperforming multilingual BERT on various Hindi NLP tasks