LLM Reference

DeepSeek 7B Chat

About

DeepSeek LLM 7B Chat is a 7-billion parameter large language model created by DeepSeek AI, designed primarily for conversational applications. It is trained on an extensive dataset consisting of 2 trillion tokens in both English and Chinese, enabling it to generate coherent and contextually relevant text responses. Initially developed from scratch, the model underwent additional fine-tuning with instruction data to enhance its conversational skills, thus distinguishing it from the DeepSeek LLM 7B Base model. DeepSeek also offers larger models such as the DeepSeek LLM 67B Chat. Although it excels in natural language understanding and generation, the model may encounter limitations in domain-specific knowledge and may not perform optimally with complex or low-quality visual content, as seen with its multimodal counterpart, DeepSeek VL 7B Chat.

Capabilities

MultimodalFunction CallingTool UseJSON Mode

Specifications

FamilyDeepSeek
Released2023-11-29
Parameters7B
ArchitectureDecoder Only
Specializationgeneral