Skip to content
View soundstarrain's full-sized avatar

Block or report soundstarrain

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.

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
soundstarrain/README.md

Hi there, I'm SoundRain! 👋

I am a Computer Science sophomore deeply focused on Machine Learning, Large Language Models (LLMs), and Recommender Systems.

🔬 Research & Interests

  • LLMs & Alignment: Post-Training, Reinforcement Learning, NLP, and Machine Translation.
  • Recommender Systems: Graph Neural Networks (GNN), Denoising, and Contrastive Learning (e.g., InfoNCE).
  • Exploring: AI for Science (AI4S), Embodied AI (Robotics), and Speech Technologies.

💻 Tech Stack

  • Deep Learning: PyTorch
  • LLM Training: Unsloth, verl, LLaMA-Factory
  • Inference & Optimization: vLLM, SGLang, OpenAI Triton

📌 Featured Work & Research

  • Murasaki-project: An ACGN translation model with native CoT and Long Context support. Achieving SOTA performance in the ACGN translation and localization domain.
  • [Ongoing Research]: Currently working on specific research problems in LLM Machine Translation, Math Reasoning, and Recommender Systems.

📫 Connect


📊 My GitHub Stats

Pinned Loading

  1. Murasaki-project Murasaki-project Public

    基于 System 2 推理范式的 ACGN 特化翻译模型,原生支持 CoT 思维链与长上下文窗口。Murasaki: An ACGN translation model with native CoT (Chain-of-Thought) and Long Context support.

    22

  2. Murasaki-Translator Murasaki-Translator Public

    Murasaki 系列模型官方推理前端。原生支持CoT(思维链)与长上下文的次时代ACGN 翻译引擎,一键高质量翻译轻小说、字幕与游戏和漫画文本。

    TypeScript 121 3

  3. Murasaki-benchmark Murasaki-benchmark Public

    A Comprehensive Benchmark for Evaluating Japanese-to-Chinese ACGN Translation Quality

    Python 8

  4. LLM-Filter-Probe LLM-Filter-Probe Public

    一款针对 LLM 输入侧审查的精确逆向分析工具。自动定位 NewAPI、OneAPI 及任何实施基于字典规则进行 Prompt 过滤的 API 网关中的敏感关键词 | A precision reverse-engineering tool for LLM input censorship. Automatically pinpoints blocked keywords in NewAPI…

    Python 3 1