Last Updated on 2024-08-31 by Clay
Variational AutoEncoder (VAE) is an advanced variant of the AutoEncoder (AE). The architecture is similar to the original AutoEncoder, consisting of an encoder and a decoder.
Read More »[Machine Learning] Note Of Variational AutoEncoder (VAE)
Last Updated on 2024-08-30 by Clay
Currently, LLM services cover a wide range of fields, and Prompt Injection and Jailbreak threats to LLMs are growing by the day. A few months ago, a customer service LLM even provided incorrect information, leading to a loss of customer rights (although that wasn't caused by a prompt attack).
Microsoft's open-source BIPIA (Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Models) evaluation method, although tested six months ago without significant updates since, remains a simple and convenient testing method for the tasks I have at hand.
Read More »Evaluating LLM Defense Capabilities Using the Microsoft BIPIA Framework
Last Updated on 2024-08-28 by Clay
difflib is a module in the Python standard library used to compare differences between sequences (often text). Back when I was doing my thesis, I implemented this by hand. It's funny and a bit frustrating to realize now in my work that there's such a neat module for this.
Last Updated on 2024-08-27 by Clay
In Python class construction, the @property decorator is commonly used and has significant benefits. Its main purpose is to transform a class method into a read-only attribute, allowing users to retrieve computed results via attribute access.
Last Updated on 2024-08-26 by Clay
Newton's interpolation is a polynomial interpolation method that constructs a set of polynomial functions using multiple data points. A major advantage is that with the addition of new data, Newton's interpolation method does not require recalculations from scratch but can instead expand on the existing function.
Read More »Note Of Newton Polynomial
Last Updated on 2024-08-25 by Clay
On platforms like LeetCode, if one looks at the solutions provided by top coders after solving problems, one often encounters a peculiar piece of code (C++ only):
Read More »[C++] The meaning of ios_base::sync_with_stdio(false) and cin.tie(NULL) in programming competition
Last Updated on 2024-08-24 by Clay
ColBERT is an embedding model designed specifically for retrieval tasks, transforming the tokens of Queries and Documents into embeddings and computing the maximum similarity.
Read More »[Paper Reading] ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT
Last Updated on 2024-08-23 by Clay
Hadolint is a Dockerfile linter that helps you follow best practices and style guidelines when writing Dockerfiles.
Read More »[Linux] How To Use Hadolint To Avoid Terrible Dockerfile
Last Updated on 2024-08-22 by Clay
To this day, many AI applications and open-source projects are developed based on the HuggingFace transformers package. A large number of models and packages are written to be compatible with the transformers format, and even share the same functions and methods, which makes them more widely accepted.
Under this premise, I came across an open-source training framework that conveniently wraps the automatic reading of Transformer architectures. However, one unavoidable problem is I want to use my custom model for experiments. I tried several solutions, hoping that when using AutoModel.from_pretrained(), by simply providing the local path to my model, I could successfully use my custom model architecture. This article records the method that worked.
Last Updated on 2024-08-21 by Clay
When building deep learning models in PyTorch, adjusting the shapes of layers and input/output dimensions is something every AI engineer has to deal with. However, there is a small but interesting pitfall in the view() method of PyTorch:
RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.Read More »[Solved] RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.
