Let me tell you a tale – a tale of magic and power. Ok, maybe there isn’t any real magic to this other than the magic described by Arthur Clarke in this quote: “Any sufficiently advanced technology is indistinguishable from magic.” My reference to power is in the sense of what you as a third… Continue reading a tale of complexity
Category: Research
chatGPT code hallucinations
By this point, many people understand the concept of an AI “hallucination”. This is the term that has come to describe incorrect information stated as facts in the output of a chatGPT prompt. For example, if you ask “Who is Brian Toone?” to chatGPT, you get the response below which has some correct information, but… Continue reading chatGPT code hallucinations
Llama, Llama! (Red Pajama)
This was a fun book we used to read to our kids … but in other exciting news, Meta approved my research request for access to the LLaMA pre-trained large language model. 4/18 5:27pm. I started using the llama download script to download all the data, but the download.sh script doesn’t work by default on… Continue reading Llama, Llama! (Red Pajama)
Win11 officeGPT
Through my efforts to install the Stanford Alpaca system on my M1 mac studio that I dubbed homeGPT, I discovered that some of the required packages are configured to only use CUDA … in other words I need an NVIDIA graphics card. One of my research computers from school is a Windows 11 system with… Continue reading Win11 officeGPT
homeGPT, cont’d
Stanford Alpaca takes the LLaMA pre-trained model released by the Meta AI and “fine-tunes” the model to respond better to natural language questions and prompts (exactly what chatGPT does). Note, the screenshot above is from LLaMA, not from Alpaca. This shows you the value that alpaca adds as the output from the example LLaMA is… Continue reading homeGPT, cont’d
ChatGPT programming
I’m still waiting to hear back from Meta (Facebook) about my application for the llama dataset. I can install all the tools, but the pre-trained model which took months to train is behind an application form. It was leaked online through 4chan with someone then submitting a GitHub pull request to update the AI research… Continue reading ChatGPT programming
HomeGPT, part 1 – PyTorch
3/21 6:48AM. The genie truly is out of the bottle. I am in the process of installing LLaMA on my home mac studio. Here’s the play-by-play. LLaMA is based on PyTorch, so in this part, I install and activate the latest version of pytorch optimized for my M1 Ultra. Some notes … pyenv vs virtualenv… Continue reading HomeGPT, part 1 – PyTorch
Data integrity
There are two kinds of data integrity: there’s the data storage kind where you are concerned that data isn’t accidentally (or maliciously) modified or deleted once it is in the system. This is a problem I am going to tackle in the next few days replacing a failing SSD. But there’s also a second kind… Continue reading Data integrity
Server-Sent Events
One of the HTML5 features introduced that I have not needed until now is something called “server-sent events“. It’s a strange name which I think is derived from the HTTP content-type text/event-stream which must be set on the server side script generating the events. This isn’t quite the same as web push notifications which are… Continue reading Server-Sent Events
MySQL 8.0
I’ve already run into a number of issues with MySQL 8.0. Most of them have been related to the GIS features (e.g., all functions are now named ST_, geometries must exactly match the column type, etc…). One new issue I ran into today is that my spatial indices are being ignored because I didn’t create… Continue reading MySQL 8.0