LLM cocktail
Recently, I found an interesting paper[2] from twitter[1]. Below are my notes.
Catastrophic Forgetting
The “pre-training and fine-tuning” paradigm i...
Recently, I found an interesting paper[2] from twitter[1]. Below are my notes.
The “pre-training and fine-tuning” paradigm i...
MLOps(Machine Learning Operations) is a practice that focuses on the deployment, management, and monitoring of machine learning models in production environments. It co...
In NLP, embedding is a key conpoent for many downstream tasks such as classification, NLI etc. A good loss function usually benefits for embedding of training. Recently, I have read two papers r...
In my previous blog, we introduced the semantic similarity of text about cosine distance. However, it’s only about how to mea...
Cosine similarity is commonly used for semantic similarity because it offers several advantages in measuring the similarity between vectors,...
As LLM(Large Language Model) has great performance of generation, code generation product like github copilot has become a powerful tool for programmer which can help them save ...
In my previous post about introduction of entity matching, tranditional entity matching process consists of data-processing, schema match...
My previous post-compress prompt has introduced the way of compression to reduce costs. Recently, I watched the Read More