http://c0ldstudy.github.io/ A minimal, responsive and feature-rich Jekyll theme for technical writing. 2026-03-27T19:16:07-07:00 Jiacen (Jason) Xu http://c0ldstudy.github.io/ Jekyll © 2026 Jiacen (Jason) Xu /assets/img/favicons/favicon.ico /assets/img/favicons/favicon-96x96.png 2023-11-10T00:00:00-08:00 2024-03-24T16:03:27-07:00 http://c0ldstudy.github.io/posts/LLM_Deploy/ Jiacen (Jason) Xu

In the blog, I would like to talk about my exploiting of Large Language Deployment in a linux server. TL; DR: I use LMDeploy to deploy the large language models like llama2 with the TurboMind backend and gradio frontend. 1.0 Preparation Use the following command to install lmdeploy. pip install lmdeploy # install git lfs following the link (https://github.com/git-lfs/git-lfs/blob/main/I...

2023-10-21T00:00:00-07:00 2023-11-20T09:51:07-08:00 http://c0ldstudy.github.io/posts/GSA/ Jiacen (Jason) Xu

Title: Understanding and Bridging the Gap Between Unsupervised Network Representation Learning and Security Analytics Author: Jiacen Xu, Xiaokui Shu and Zhou Li Conference: 2024 IEEE Symposium on Security and Privacy (SP) Paper: link Code: https://github.com/C0ldstudy/Argus Bibtex: @INPROCEEDINGS {, author = {J. Xu and X. Shu and Z. Li}, booktitle = {2024 IEEE Symposium on Secu...

2023-06-10T00:00:00-07:00 2023-06-10T00:00:00-07:00 http://c0ldstudy.github.io/posts/LLM_Paper_Summary/ Jiacen (Jason) Xu

1. Attention is All You Need Paper/TF_code/Torch_code Main Idea The paper is the first to propose a parallel Transformer in the sequence transduction models. Key insight The problems of the RNN (LSTM, GRU) model: Need to calculate $h_t$ by $h_{t-1}$ which cannot be executed in parallel. While Transformer reduces the sequential computation Model Structure: Position Encoding: they use sin...

2023-05-20T00:00:00-07:00 2023-05-20T00:00:00-07:00 http://c0ldstudy.github.io/posts/Transformer_Related_Implementation/ Jiacen (Jason) Xu

1. Transformer[1] I start it by following the video[1] from Andrej Karpathy and check the code from the repo. In the video, he shows how to build a GPT by oneself step by step which also shows the way to build a NLP model in a high level. I summarize these steps and add some useful tips that mentioned in the video. 1.1 Data Preprocessing Tokenizer 1.2 Bigram Model The code is here. High-leve...

2023-04-25T00:00:00-07:00 2023-05-08T21:20:38-07:00 http://c0ldstudy.github.io/posts/ChatGPT_Prompt/ Jiacen (Jason) Xu

The note is for the course: DeepLearning.AI ChatGPT Prompt Engineering Course. ChatGPT Prompt Engineering for Developers Course link: https://learn.deeplearning.ai/chatgpt-prompt-eng/lesson/1/introduction 1 Basic concepts: Base LLM: Predict the next word based on the training data. Instruction Tuned LLM: Fine-tune on instructions and good attempts at following instructions. 2 Prompting ...