DeepSeek-R1-8B-Unified-IdemFormer

Description

DeepSeek-R1-8B-Unified-IdemFormer is an open-weight text-generation model with roughly 8B parameters published by aecetin on the Hugging Face Hub. It has 471 downloads.

Categories

• Research

Tags

• deepseek_r1_unified_idemformer
• deepseek-r1
• reasoning
• universal-transformer
• weight-shared-attention
• 128k-context
• 32x-context-folding
• tarski-invariance

Capabilities

• Text generation
• 8B parameters