pietz

8 ms·

61.
by pietz 10mo ago
I hear you. This boils town to personal opinion. I would have preferred to use an existing standard than introducing yet another one. The custom sections aren't something I've ever seen or needed anyway.
62.
by pietz 10mo ago
A bit of a missed opportunity not to use the JSON Resume schema for this. https://jsonresume.org/schema
63.
by pietz 10mo ago
That's a fair point and yet I deeply believe Codex is better here. After finishing a big task, I used two fresh instances of Claude and Codex to review it. Codex finds more issues in ~9 out of 10 cases. While I prefer the way Claude s
64.
by pietz 10mo ago
Interesting. Flash suggests more power to me than Mini. I never use gpt-5-mini in the UI whereas Flash appears to be just as good as Pro just a lot faster.
65.
by pietz 10mo ago
Now do the same with Rust, build a Python wrapper and we went full circle :)
66.
by pietz 10mo ago
People buy F150s because they find them cool and not because they actually need the space. Your Corolla could make deliveries around town in roughly the same time, while being cheaper and easier compared to introducing a new expensive car.
67.
by pietz 10mo ago
You start by underselling what can be done with Postgres and then follow up with the upper end of requirements that most projects won't need. My argument is exactly what you conveniently left out: The big bulk between the two.
68.
by pietz 10mo ago
It seems to me like many developers are moving from native vector stores back to Postgres with pgvector. It's easier to integrate, cheaper to set up, familiar to work with, oftentimes faster, while not hitting any limitations for most
69.
by pietz 10mo ago
They are definitely ahead in multi modality and I'd argue they have been for a long time. Their image understanding was already great, when their core LLM was still terrible.
70.
by pietz 11mo ago
It means half of the internet isn't (wasn't) working.
71.
by pietz 11mo ago
I imagine Anthropic employs a lot of talent from China. Beyond the political, they should be fairly certain to publish these claims to avoid an internal shit storm.
72.
by pietz 11mo ago
Can you go into a bit more detail why the two approaches are so different in your opinion? I don't think I agree and I want to understand this argument better.
73.
by pietz 11mo ago
Since they also limited the usage of codex CLI quite a bit, this might help. I really like the gpt-5-codex model. It has the most impressive dynamic reasoning effort I've seen. Responding instantly to simple questions while thinking ve
74.
by pietz 1y ago
Thanks for sharing the font story! That made me quite a bit happier than it should have.
75.
by pietz 1y ago
Cannot confirm this. Both sound like prompting issues. - Depends on your use case to let the model understand when and when not to use tools - gpt-5 s VERY persistent and often searches more than 10 times in a single run depending on the re
76.
by pietz 1y ago
Oh, it's nothing fancy I'm afraid. I have an AGENTS.md file in my vault that provides a bit of context that this is not about coding, but about serving as my Obsidian helper. I ask it to check the .obsidian folder to check my plug
77.
by pietz 1y ago
It has been wild watching deta over the years. They didn't pivot, they completely reinvented themselves. Twice. I loved their first cloud offering, which they sadly abandoned. Then they launched Space, which was kinda cool, but mostly
78.
by pietz 1y ago
Makes sense, I appreciate the comment. Well written article. Subscribed.
79.
by pietz 1y ago
I enjoyed reading this, but it seems overly complex and at least slightly flawed. Why not embed all tweets, cluster them with an algorithm of your choice and have an LLM provide names for each cluster? Cheaper, better clusters and more accu
80.
by pietz 1y ago
My biggest RAG learning is to use agentic RAG. (Sorry for buzzword dropping) - Classic RAG: `User -> Search -> LLM -> User` - Agentic RAG: `User <-> LLM <-> Search` Essentially instead of having a fixed loop, you provid
81.
by pietz 1y ago
But the model is like 18 months old. and recently we've seen big leaps on MTEB. Not sure how well those translate to reality, but I'm a little surpised this wasn't worth looking into.
82.
by pietz 1y ago
I find it interesting that so many services and tools were investigated except for embedding models. I would have thought that's one of the biggest levers.
83.
by pietz 1y ago
I threw in the first image/table into Gemini 2.5 Pro letting it choose the output format and it looks like it extracted the data just fine. It decided to represent the checkboxes as "checked" and "unchecked" because
84.
by pietz 1y ago
Maybe my imagination is limited or our documents aren't complex enough, but are we talking about realistic written documents? I'm sure you can take a screenshot of a very complex spreadsheet and it fails, but in that case you alre
85.
by pietz 1y ago
Could you provide an example that fails? I'm interested in this.
86.
by pietz 1y ago
We ran a small experiment internally on this and it looked like Gemini is better at handwriting recognition than I am. After seeing what it parsed, I was like "oh yeah, that's right". I do agree that instead of saying "S
87.
by pietz 1y ago
My impression is that OCR is basically solved at this point. The OmniAI benchmark that's also referenced here wasn't updated with new models since February 2025. I assume that's because general purpose LLMs have gotten better
88.
by pietz 1y ago
I think with gpt-5-mini and now Haiku 4.5, I’d phrase the question the other way around: what do you need the big models for anymore? We use the smaller models for everything that’s not internal high complexity tasks like coding. Although t
89.
by pietz 1y ago
Give me a 30min uncut video of it doing chores around the house to make me a believer.
90.
by pietz 1y ago
Simon went quickly from not agreeing with common buzz words to making up his own.

More ›