Comments URL: https://news.ycombinator.com/item?id=48517272

Points: 3

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=46847714

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=45500485

Points: 464

# Comments: 230

Comments URL: https://news.ycombinator.com/item?id=45399267

Points: 3

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=45320890

Points: 3

# Comments: 1

Comments URL: https://news.ycombinator.com/item?id=45214209

Points: 54

# Comments: 9

Comments URL: https://news.ycombinator.com/item?id=45119002

Points: 1

# Comments: 0

It's basically an SQL wrapper around poppler.

Comments URL: https://news.ycombinator.com/item?id=44669393

Points: 6

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=44161042

Points: 1

# Comments: 0

> This great basin of water, the cradle of so many civilizations, now looks like a mirror of death. Let us not let our sea (mare nostrum) be transformed into a desolate sea of death (mare mortuum). Let us not allow this place of encounter to become a theatre of conflict. Let us not permit this “sea of memories” to be transformed into a “sea of forgetfulness”. Please brothers and sisters, let us stop this shipwreck of civilization!

> We are in the age of walls and barbed wire. To be sure, we can appreciate people’s fears and insecurities, the difficulties and dangers involved, and the general sense of fatigue and frustration, exacerbated by the economic and pandemic crises. Yet problems are not resolved and coexistence improved by building walls higher, but by joining forces to care for others according to the concrete possibilities of each and in respect for the law, always giving primacy to the inalienable value of the life of every human being

Worth reading in full https://www.vatican.va/content/francesco/en/speeches/2021/de...

You can tell that because simonw writes quite heavily-documented code an the logic is pretty straightforward, it helps the model a lot!

https://github.com/Florents-Tselai/Tutorial-Codebase-Knowled...

https://github.com/Florents-Tselai/Tutorial-Codebase-Knowled...

from ollama import chat, ChatResponse

def call_llm(prompt, use_cache: bool = True, model="phi4") -> str: response: ChatResponse = chat( model=model, messages=[{ 'role': 'user', 'content': prompt, }] ) return response.message.content

I frequently work with Postgres & Python and toggle between multiple versions of each. In Postgres, I keep a separate installation for each project, but I don't like git submodules and don't want to ship "pg_build.sh" scripts with the source code.

So, I built pgvenv as a Python shell package that, when installed, triggers a from-source installation of Postgres and installs it at venv's prefix. Could become handy at times I think.

You can roughly use it like this.

python3.11 -m venv ./venv

source ./venv/bin/activate

PGVERSION=17.4 pip install pgvenv --force-reinstall --no-cache-dir

initdb ./pgdata

postgres -D ./pgdata

Many things can go wrong with this in production, so should probably use for development only.

Comments URL: https://news.ycombinator.com/item?id=43726463

Points: 2

# Comments: 0

I run diofanti.org, a free, open-data platform that monitors government spending and operations in Greece. We transform raw public data into a tableau-like dashboard, providing citizens, journalists, and policymakers with transparency and actionable insights.

For the past five years, a Hetzner-hosted Postgres instance and Metabase have served us well. However, I’ve recently been working on adding an AI/Chat interface to enhance the user experience. I can handle the AI engineering, ML ops, and model pipeline setup myself. Initial prototyping has been promising, but the AI infrastructure costs are now growing—primarily for GPU and AI compute (inference and generation, but no fine-tuning yet).

Diofanti is a non-profit, and while we’re focused on transparency and civic engagement, we’re running into limitations around infrastructure costs. Traditional VC funding routes don’t apply since it's not monetizing, and big cloud providers aren’t a good fit either—most want you to join an accelerator that I don’t have the time to go through. Local (Greek) philanthropic organizations and fellowships have been frustratingly slow, bureaucratic, and not particularly helpful.

I’m looking for any suggestions or funding opportunities to help support our AI infrastructure, whether it’s grants, cloud credits, or connections with organizations that support civic tech or AI for good projects.

If anyone has leads or suggestions (or if you have any experience with similar projects), I’d greatly appreciate hearing from you. Feel free to reach out directly (contact info is on my website: tselai.com).

Comments URL: https://news.ycombinator.com/item?id=43620299

Points: 2

# Comments: 0

I’m curious to hear from others working with non-English languages—what challenges have you faced? Some areas of interest: - Models that are more open to switch language - Availability and quality of language-specific retrieval corpora - Differences in tokenization and embedding quality - Handling multilingual queries and responses - Any workarounds or best practices you’ve discovered

Would love to hear both success stories and pain points.

Comments URL: https://news.ycombinator.com/item?id=43579192

Points: 2

# Comments: 0