Cameron Wolfe

2 weeks ago | stackoverflow.blog | Cameron Wolfe

TL;DR: Self-supervised learning is a key advancement in deep learning that is used across a variety of domains. Put simply, the idea behind self-supervised learning is to train a model over raw/unlabeled data by making out and predicting portions of this data. This way, the ground truth “labels” that we learn to predict are already present in the data itself and no human annotation is required. Types of learning. Machine learning models can be trained in a variety of ways.

2 weeks ago | cameronrwolfe.substack.com | Cameron Wolfe

The recent release of Llama 4 [1] was far from perfect, but there is a lot that can be learned from this new generation of models. Put simply, Llama 4 is a massive pivot in Meta’s research direction. In response to increasing competition, Meta is reinventing the Llama series and clearly pushing to create a frontier-level LLM. Given that LLM development is an iterative process, such significant changes incur a lot of risk—there’s a huge chance that these models will perform poorly at first.

1 month ago | cameronrwolfe.substack.com | Cameron Wolfe

After the popularization of text-based large language models (LLMs), one of the most important questions within the research community was how we could extend such powerful models to understand other modalities of data (e.g., images, video or speech). Research on multi-modal LLMs is promising for several reasons:Improving model capabilities. Uncovering new sources of training data. Expanding the scope of problems that LLMs can solve.

1 month ago | ncmedicaljournal.com | Robert J. Rolfe |Colin M Smith |Cameron Wolfe

By Robert J. Rolfe, Colin M. Smith, Cameron R. Wolfe Rolfe, Robert J., Colin M. Smith, and Cameron R. Wolfe. 2021. "The Emerging Chronic Sequelae of COVID-19 and Implications for North Carolina." North Carolina Medical Journal 82 (1): 75-78. https:/​/​doi.org/​10.18043/​ncm.82.1.75.

2 months ago | stackoverflow.blog | Cameron Wolfe

There are many variants of LoRA you can use to train a specialized LLM on your own data. Here’s an overview of all (or at least most) of the techniques that are out there. LoRA models the update derived for a model’s weights during finetuning with a low rank decomposition, implemented in practice as a pair of linear projections. LoRA leaves the pretrained layers of the LLM fixed and injects a trainable rank decomposition matrix into each layer of the model.