---
title: "Discussion on \"DeepSeek LLM: Scaling Open-Source Language Models with Longtermism 논문 리뷰\" | Hashnode"
description: "Discussion on \"DeepSeek LLM: Scaling Open-Source Language Models with Longtermism 논문 리뷰\". DeepSeek AI는 영어와 중국어 분야에서 우수한 성능을 보이는 모델을 선보이는 동시에, 하이퍼파라미터와 모델·데이터 간 연산량 할당에 관한 정교한 스케일링 법칙도 제시했습니다."
url: "https://hashnode.com/posts/deepseek-llm-scaling-open-source-language-models-with-longtermism/67e2c222dc0e88ba125abf9f"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [DeepSeek LLM: Scaling Open-Source Language Models with Longtermism 논문 리뷰](https://bits-bytes-nn.hashnode.dev/deepseek-llm-scaling-open-source-language-models-with-longtermism?utm_source=hashnode&utm_medium=feed)

DeepSeek AI는 영어와 중국어 분야에서 우수한 성능을 보이는 모델을 선보이는 동시에, 하이퍼파라미터와 모델·데이터 간 연산량 할당에 관한 정교한 스케일링 법칙도 제시했습니다.

Jan 4, 2024 · 21 min read · bits-bytes-nn.hashnode.dev  
Updated Apr 8, 2025  
Read the article

-   [#deepseek-model](https://hashnode.com/tag/deepseek-model)
-   [#directpreferenceoptimization](https://hashnode.com/tag/directpreferenceoptimization)
-   [#scalinglaws](https://hashnode.com/tag/scalinglaws)
-   [#multisteplearningratescheduler](https://hashnode.com/tag/multisteplearningratescheduler)
-   [#gqa](https://hashnode.com/tag/gqa)
-   [#rotarypositionalembeddings](https://hashnode.com/tag/rotarypositionalembeddings)
-   [#swiglu](https://hashnode.com/tag/swiglu)
-   [#prenormalization](https://hashnode.com/tag/prenormalization)
-   [#large-language-models-llms](https://hashnode.com/tag/large-language-models-llms)
-   [#sft](https://hashnode.com/tag/sft)

## Written by

[Jonas Kim](https://hashnode.com/@bits-bytes-nn)

@bits-bytes-nn · Seoul

  
[View profile](https://hashnode.com/@bits-bytes-nn)

♾️ Life is complex. It has both real and imaginary parts.

Sr. Data Scientist at AWS

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
