DeepSeek LLM is an advanced language model available in both 7 billion and 67 billion parameters. Both a chat and base variation are available. (DeepSeek LLM 是一个高级语言模型，提供 70 亿和 670 亿参数两种版本。都提供 chat 和 base 两种变体。)

• Superior General Capabilities: DeepSeek LLM 67B Base outperforms Llama2 70B Base in areas such as reasoning, coding, math, and Chinese comprehension. (卓越的通用能力：DeepSeek LLM 67B Base 在推理、编码、数学和中文理解等领域优于 Llama2 70B Base。)

• Proficient in Coding and Math: DeepSeek LLM 67B Chat exhibits outstanding performance in coding (using the HumanEval benchmark) and mathematics (using the GSM8K benchmark). (精通编码和数学：DeepSeek LLM 67B Chat 在编码（使用 HumanEval 基准）和数学（使用 GSM8K 基准）方面表现出色。)

References (参考资料)

GitHub (代码仓库)

HuggingFace