Viewing profile — viktour19

US healthcare is expensive because prices aren’t real and complexity is profitable. Hospitals and pharma have monopoly power, insurers add massive administrative drag, and nobody s…

The world’s first hydrogen exploration company was founded in Mali. No one knows for a fact what the development pathway is. Solar might turn out to be inconsequential. https://new…

The effective parameters of the model are the parameters of the original model + lora parameters i.e lora updates only lora parameters, and full finetuning updates only original mo…

> LoRA and full fine-tuning, with equal performance on the fine-tuning task, can have solutions with very different generalization behaviors outside the fine-tuning task distributi…

Diffusion models are already being evaluated using pretrained SSL models à la CLIP Score [1]. So it makes sense that one would incorporate that directly into training the model fro…

What are details of the technique?

It's great how we went from "wait.. this model is too powerful to open source" to everyone trying to shove down their 1% improved model down the throats of developers

If hallucination is inevitable? What should developers do? Design user experiences that align users with this behaviour! Relatedly, I built a game to demonstrate how one might cali…

Calibration game is a game to get better at identifying hallucination in LLMs. Prompts and hallucination labels (using ChatGPT) are obtained from Alpaca and HaluEval. You can use y…

Combine this with something like Copilot and you get the 10x programmer :)

Yes please!

I love this.

I wonder how much of the annoyance with the term "Imbox" has to do with the dichotomy between System 1 and System 2 thinking. When I see "Inbox", I put almost no thought into what …

Interesting. I wonder what the dataset used to train the machine learning model looks like. Books and conversation style query pairs?

This is awesome. some new ideas for my drone.