# LLM_Homework

**Repository Path**: AndyWu93/llm_homework

## Basic Information

- **Project Name**: LLM_Homework
- **Description**: llm homework anf notebooks for llm research
- **Primary Language**: Python
- **License**: Apache-2.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2025-09-21
- **Last Updated**: 2025-09-21

## Categories & Tags

**Categories**: Uncategorized

**Tags**: llm

## README

# 作业20250920

## Complection API

- presence_penalty
- best_of
- top_p
- [How to count tokens with tiktoken](https://cookbook.openai.com/examples/how_to_count_tokens_with_tiktoken)

在notebook上把所有case跑一遍，加深对token的印象

## 仅保留3次对话，超过3次删除

## 仅保留3次对话，3次之前的做概要总结，然后在message中将3次前删除，并加上概要总结后的消息

> pip install -i <https://pypi.mirrors.ustc.edu.cn/simple/> "pandas<2.0.0" "numpy<2.0.0"

# 20250920

- 研究自己组的pipeline对应的prompts

- 将arrow文件的数据导出json格式

- pipeline出来的内容均为中文