---
title: "Anni Huang (@HuangAnni) | Hashnode"
description: "LLM Training Operation Specialist @ ByteDance | Ex-Master of IT in Business (AI track) at SMU. Read the latest articles by Anni Huang on Hashnode."
url: "https://hashnode.com/@HuangAnni"
---

@HuangAnni

# Anni Huang

@HuangAnni · Singapore · Joined July 2022

LLM Training Operation Specialist @ ByteDance | Ex-Master of IT in Business (AI track) at SMU

### [LLM for Coding Benchmarks and Datasets](https://huanganni.hashnode.dev/llm-for-coding-benchmarks-and-datasets?utm_source=hashnode&utm_medium=feed)

LiveCodeBench, SWEBench, Aider Polyglot, BBH, HumanEval, MBPP, Common Crawl (Time Span, Dataset Size, Data Format, LLM Testing Capability)

1 min read · [huanganni.hashnode.dev](https://huanganni.hashnode.dev/llm-for-coding-benchmarks-and-datasets?utm_source=hashnode&utm_medium=feed)

[Aug 14, 2025](https://hashnode.com/posts/llm-for-coding-benchmarks-and-datasets/689d49de1d52dcaf7fd402a5)

### [How to change a base model to a reasoning model?](https://huanganni.hashnode.dev/how-to-change-a-base-model-to-a-reasoning-model?utm_source=hashnode&utm_medium=feed)

Turning a base model into a reasoning model is essentially a post-training + data problem

2 min read · [huanganni.hashnode.dev](https://huanganni.hashnode.dev/how-to-change-a-base-model-to-a-reasoning-model?utm_source=hashnode&utm_medium=feed)

[Aug 14, 2025](https://hashnode.com/posts/how-to-change-a-base-model-to-a-reasoning-model/689d41e765bbb43b312c3b4c)

### [DeepSeek GRPO Explanation (Why we need it? How does it work? What are the findings?)](https://huanganni.hashnode.dev/deepseek-grpo-explanation-why-we-need-it-how-does-it-work-what-are-the-findings?utm_source=hashnode&utm_medium=feed)

Explained with a simple mathematical formula for the loss function

2 min read · [huanganni.hashnode.dev](https://huanganni.hashnode.dev/deepseek-grpo-explanation-why-we-need-it-how-does-it-work-what-are-the-findings?utm_source=hashnode&utm_medium=feed)

[Aug 13, 2025](https://hashnode.com/posts/deepseek-grpo-explanation-why-we-need-it-how-does-it-work-what-are-the-findings/689cc76e82e17fe4eeba6653)

### [Beyond Pre-training: The Power of RLHF in LLM Alignment](https://huanganni.hashnode.dev/beyond-pre-training-the-power-of-rlhf-in-llm-alignment?utm_source=hashnode&utm_medium=feed)

If LLM is compared to a person, then the early training is like general education, and the later training is like vocational training.

2 min read · [huanganni.hashnode.dev](https://huanganni.hashnode.dev/beyond-pre-training-the-power-of-rlhf-in-llm-alignment?utm_source=hashnode&utm_medium=feed)

[Aug 13, 2025](https://hashnode.com/posts/beyond-pre-training-the-power-of-rlhf-in-llm-alignment/689cc42ea3cc115547a50c37)

### [LeetCode Daily Challenge-3477. Fruits Into Baskets II(Easy, O(n²))](https://huanganni.hashnode.dev/leetcode-daily-challenge-3477-fruits-into-baskets-iieasy-on?utm_source=hashnode&utm_medium=feed)

Unavoidable O(n²) - Optimal Solution in Python, Java, Golang, C, C#, C++

4 min read · [huanganni.hashnode.dev](https://huanganni.hashnode.dev/leetcode-daily-challenge-3477-fruits-into-baskets-iieasy-on?utm_source=hashnode&utm_medium=feed)

[Aug 5, 2025](https://hashnode.com/posts/leetcode-daily-challenge-3477-fruits-into-baskets-iieasy-on/68917ae6f2dbd648272e4d13)
