Skip to content
View Ariasu123's full-sized avatar

Sponsoring

@vllm-project

Block or report Ariasu123

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Ariasu123/README.md

◇ Tech Stack

◈ Learning Journey

LLM — PyTorch, Transformers, Pretrain and SFT — Built Knowledge — GraphRAG, Vector Search and LLM wiki — Built Agent — MCP, Agent Skills and Tool Use — Building Infra — vLLM, CUDA and Distributed Systems — Exploring

▣ Featured Projects

Applications

A 54M Chinese LLM built from scratch with native PyTorch, from pretraining to full SFT.

Python PyTorch Pretrain SFT TinyLLM stars

A lightweight terminal coding agent with tools, MCP, hooks, and Docker sandboxing.

Python Agent MCP Docker Pion stars

A compact vLLM extension for speculative decoding, chunked prefill, fair scheduling, and prefix caching.

Python vLLM Speculative Decoding Prefix Cache Nano-vLLM-Ext stars

◫ Development Metrics

Automatically updated GitHub statistics

Pinned Loading

  1. Nano-vLLM-Ext Nano-vLLM-Ext Public

    Nano-vLLM 扩展:Draft–Target 投机解码、Chunked Prefill 与公平调度、Prefix Cache 增强

    Python

  2. Pion Pion Public

    参考 Pi coding agent 核心设计,基于 Python 重写的极简 CLI 编程智能体

    Python 1

  3. TinyLLM TinyLLM Public

    复刻MiniMind项目,从0到1构建了轻量级大语言模型

    Python

  4. harbor-agentic-eval-libyaml harbor-agentic-eval-libyaml Public

    基于 Harbor 的可复现 Agentic Coding 评测:为 LibYAML 实现解析器级资源限制

  5. AgentDisk AgentDisk Public

    多Agent 的本地磁盘分析与安全管理工具

    Go

  6. GraphRAG-Kit GraphRAG-Kit Public

    从文档到知识图谱的完整链路——解析、构建、增量更新、多模式检索

    Python