VibeThinker-1.5B is a 1.5-billion-parameter dense language model launched by Weibo AI. It is fine-tuned based on Qwen2.5-Math-1.5B and is specifically designed for mathematical and algorithmic coding problems. Trained using the 'Spectrum to Signal Principle' framework, it outperforms larger models in multiple math competition tests. The training cost is approximately $7,800, and it supports an output of up to about 40k tokens.

Natural Language ProcessingTransformersMultiple Languages

This is a static quantized version of the PRIME-RL/P1-30B-A3B model, a large language model with 30 billion parameters, optimized specifically for fields such as physics, reinforcement learning, and competition reasoning, and supports English and multilingual processing.

Natural Language ProcessingTransformersMultiple Languages

This model is specifically fine-tuned for SemEval 2025 Task3: Mu-SHROOM competition to identify hallucinated text segments in large language model outputs.

Natural Language ProcessingTransformers

ALMA-13B-R is an advanced translation model based on large language models, fine-tuned using Contrastive Preference Optimization (CPO), capable of matching or even surpassing GPT-4 or WMT competition winners.

Natural Language ProcessingTransformers