I'm not an AI doomsdayer. I'm not too concerned about AI replacing people on a mass scale. Although I do run into people from time to time whom perhaps I wouldn't mind being replaced by AI 😀
Between PIP certificates, PHP and LiteLLM (again 😡), it's a busy day in the security department.
Luckily, none of this effects txtai directly. Just make sure you read the following articles.
discuss.python.org/t/informa…
apnews.com/article/php-websi…
docs.litellm.ai/blog/securit…
Always great to come across unexpected use cases for txtai.
frontiersin.org/journals/mar…
If someone told me that coding agents create each PR in anticipation of a follow on PR to drive up token counts, it wouldn't surprise me.
I've never been shy to share that NeuML is just 1 person. Now with the power of open source, it has a 35-60+ person engineering team behind it. Whether that's people, people with Claude Code Agents or Agents, it doesn't matter to me. txtai keeps improving "for free".
Out of all courses I took in College, Signals and Systems has the principles that I've leaned the most on in machine learning.
Fourier Transforms and Convolution just seems to be something I often come back to when solving hard problems.
David Mezzetti retweeted
⚕️🧬🔬 PubMedBERT Embeddings gets over 1 million downloads a month and has been cited in over 50 articles over the last year!
Google Scholar: scholar.google.com/scholar?s…
Model: huggingface.co/NeuML/pubmedb…
David Mezzetti retweeted
💥 txtai v9.14.0 is out!
This release adds new ANN backends for RaBitQ and Zvec Sparse, native support for ONNX vectors along with a bonanza of improvements and bug fixes. There were 25 contributors with 20 new ones. These contributors added solid new features and are experts in their field. What a great and growing community!
Release Notes: github.com/neuml/txtai/relea…
GitHub: github.com/neuml/txtai
David Mezzetti retweeted
Long scrolling window of dependencies stink if you don't need them! Did you know that txtai supports a minimal no-dependency install?
neuml.github.io/txtai/instal…
Jev has been great for the AI space: a grounded, realistic approach to solving real-world problems that minimizes compute.
Meanwhile, the big LLM vendors are going to religious experts and scholars trying to get them to say AI has a soul...
With the advent of Jev, it brings me back to a recent model I built, H.G. BERT.
The idea was freezing the knowledge of 1899 into an encoder only model. If you make decisions with the ideas of the past, are there new ideas that are possible?
huggingface.co/blog/NeuML/hg…
David Mezzetti retweeted
Why the large uptick in txtai usage, PRs and overall interest? Well the "Why txtai?" page that's been around forever covers it best.
- Up and running in minutes with pip or Docker
- Built-in API makes it easy to develop applications using your programming language of choice
- Run local - no need to ship data off to disparate remote services
- Work with micromodels all the way up to large language models (LLMs)
- Low footprint - install additional dependencies and scale up when needed
- Learn by example - notebooks cover all available functionality
neuml.github.io/txtai/why/
The AI space needs better leadership. It's literally an SNL skit at this point.
youtube.com/watch?v=-Nvne3Lz…
The AI encoder-only space is 🔥 right now. I'm cooking something that you'll want to see. Brand new architecture. Stay tuned.
As I now get 5+ new PRs a day for txtai, with multiple new contributors a week, it has me wondering what's changed? Is txtai AI-agent friendly? Is it the fact that I merge PRs in? Are more people using txtai and running into edge cases? All of the above?
Jev has been great in bringing a renewed interest to encoder-only models. But once again me in 2020.
When ChatGPT was released in 2022 it took the air out of anything that wasn't Generative AI.
reddit.com/r/MachineLearning…
![From the MachineLearning community on Reddit: [P] Zero-shot classification in Python, JavaScript,...](https://pbs.twimg.com/card_img/2115247331935457281/s4laKQlg?format=jpg&name=800x419)
We're about to enter a new age where parameters can be controlled and small models with way less parameters than you think can be useful. It's not always going to be bigger is better.
