dmezzetti

8 ms·

61.
by dmezzetti 1y ago
I built TxtAI with this philosophy in mind: https://github.com/neuml/txtai
62.
by dmezzetti 1y ago
This thread is the best sales pitch for local / self-hosted models. With local, you have total control over when you decide to upgrade.
63.
by dmezzetti 1y ago
Impressive that this knocked GPT-5 from the top.
64.
by dmezzetti 1y ago
LLMs are basically all the same at this point. The margins are razor thin. The real take-off / winner-take-all potential is in retrieval and knowing how to provide the best possible data to the LLM. That strategy will work regardless o
65.
by dmezzetti 1y ago
As the margins shrink between the capabilities of each of these models, those who specialize in retrieval / context engineering will be the next frontier. Those who provide the most relevant information to a model will win the day.
66.
by dmezzetti 1y ago
It's always worth checking out the MTEB leaderboard: https://huggingface.co/spaces/mteb/leaderboard There are some good open models there that have longer context limits and fewer dimensions. The benchmarks a
67.
by dmezzetti 1y ago
It's interesting that anyone is surprised by this.
68.
by dmezzetti 1y ago
It's clear that what we currently call AI is best suited for augmentation not automation. There are a lot of productivity gains available if you're willing to accept that.
69.
by dmezzetti 1y ago
While all these architectures are innovative and have helped improve either accuracy or speed, the same fundamental problem of generating factual information still exists. Retrieval Augmented Generation (RAG), Agents and other similar metho
70.
by dmezzetti 1y ago
It's called "lowest cost technically acceptable". Publicly traded companies are driven by quarterly earnings and increasing net margins. You do that by selling products at the lowest cost possible where buyers will still buy
71.
by dmezzetti 1y ago
One additional option to consider is a local vector database with Wikipedia articles: https://huggingface.co/NeuML/txtai-wikipedia I've built this as a datasource for Retrieval Augmented Generation (RAG) but it ce
72.
by dmezzetti 1y ago
I'm the developer of txtai, a fairly popular open-source project. I don't use any AI-generated code and it's not integrated into my workflows at the moment. AI has a lot of potential but it's way over-hyped right now. Li
73.
by dmezzetti 1y ago
Very interesting work! If biomedical research and paper analysis is of interest to you, I've been working on a set of open source projects that enable RAG over medical literature for a while. PaperAI: https://github.com&#x2F
74.
by dmezzetti 1y ago
Good retrieval/search is the foundation of context. It's definitely garbage in - garbage out here otherwise. Search is far from a solved problem.
75.
by dmezzetti 1y ago
PaperAI is also an option if you prefer open-source: https://github.com/neuml/paperai Disclaimer: I'm the primary author of this project.
76.
by dmezzetti 1y ago
Software managers have long pushed "productivity" tools on developers. Most don't stick and this is likely similar. It's best to hire smart people and let them use whatever stack works best for them.
77.
by dmezzetti 1y ago
I'm not sure many seasoned developers are really using AI that much in their workflows.
78.
by dmezzetti 1y ago
There has long been ways to reduce writing boilerplate code with IDEs. AI code generation is just another tool and it will help enable competent people.
79.
by dmezzetti 1y ago
For full disclosure, I work on txtai one of the more popular AI frameworks out there. So this checks out :)
80.
by dmezzetti 1y ago
There are plenty of AI companies solving interesting problems and possibly worth it. But most problems are more simple than that and that hasn't changed.
81.
by dmezzetti 1y ago
This article is missing an even more important point: you don't always need to start with an LLM, plain old coding still solves a lot of problems.
82.
by dmezzetti 2y ago
One thing that I've found with this model is that it's not heavily censored. This is the biggest development to me, being unbiased. This could lead to more enterprise adoption. https://gist.github.com/davidmezzetti
83.
by dmezzetti 2y ago
Glad to see this getting visibility. Model2Vec is great and The Minish Lab is doing some excellent work!
84.
by dmezzetti 2y ago
Rust is a fine language and good for certain use cases. There is a segment of developers where the syntax and complexity is a bridge too far. The community tends to evangelize the greatness of Rust too much and be unrealistic with this real
85.
by dmezzetti 2y ago
How to make downsizing sound cool....
86.
by dmezzetti 2y ago
txtai automatically builds graphs using vector similarly as data is loaded. Another option is to use something like GLiNER and create entities on the fly. And then create relationships between those entities and/or documents. Or you ca
87.
by dmezzetti 2y ago
Why not semantic search was the bigger question.
88.
by dmezzetti 2y ago
Why?
89.
by dmezzetti 2y ago
Grobid is great. paperetl is the workhorse of the projects mentioned above. Good ole programming and multiprocessing to churn through data.
90.
by dmezzetti 2y ago
Excellent project. As mentioned in another comment, I've put together an embeddings database using the arxiv dataset ( https://huggingface.co/NeuML/txtai-arxiv ) recently. For those interested in the literature sear

More ›