LLM Theory
Published:
Awesome LLM theory articles I chanced upon
- State of LLMs in 2023 by Hyung Won Chung
- Transformer Math 101 by Eleuther AI.
- Transformer Inference Arithmetic by Kipply.
You May Also Enjoy
Suppose that every device takes in a batch of tensors where the tensors across devices are of different sizes, will 3D parallelism still work?
2 minute read
Published:
As I’m learning more about 3D parallelism, I wonder - suppose that every device takes in a batch of tensors where the tensors across devices are of different sizes, will 3D parallelism still work? Turns out, it works for data and pipeline parallelism, but tensor parallelism will need some work.
Simple tricks to improve Retrieval Augmented Generation (RAG) systems
2 minute read
Published:
Motivation
Retrieval Augmented Generation (RAG) is a framework for retrieving facts from an external knowledge base to ground large language models (LLMs) on the most accurate, up-to-date information. RAG is increasingly popular in industry as it’s simple to implement yet powerful. Here I’ll share some tricks to improve RAG systems.
Given the same compute budget, which is better: parameter efficient finetuning (PEFT) or full finetuning?
less than 1 minute read
Published:
Motivation
If we are allowed to train till convergence, we know that full finetuning is better than parameter efficient finetuning (PEFT). But what if we have a fixed compute budget? Given a fixed budget, PEFT can go through significantly more tokens. Will full finetuning still be better than PEFT?
Simple solution to training causal LMs with seq2seq objective
2 minute read
Published:
Motivation
How do we train causal language models (e.g. Alpaca, LLaMA, gpt-neox-20b…) with seq2seq objective? This goal is important because we want to instruction-tune our causal LMs, especially since Alpaca is the best open model at time of writing.