I’m looking for the "state of the art" in 2026. Specifically:

LLM4Rec: Beyond just using LLMs for feature engineering—who is doing generative recommendation well?

Retrieval vs. Ranking: Any new breakthroughs in the "Two-Tower" paradigm or vector database integration?

Real-world Scale: Papers that address the latency/cost trade-offs of these newer, heavier models.

What has been the most influential paper you’ve read recently that changed how you think about discovery?

Comments URL: https://news.ycombinator.com/item?id=46692368

Points: 15

# Comments: 1

His refreshingly honest take is very relatable, humorous and encouraging.

I can highly recommend it if you’re interested in prayer life (and how to use powerful formulations in letters)

One example is a GPT version [1] fine tuned on texts of Sefaria [2].

The initiator was in direct contact with Sam Altman to kickstart it. (Personal communication).

He talks about this publicly [3]

[1]: https://github.com/Sefaria/AppliedAI [2]: https://www.sefaria.org/texts [3]: https://m.youtube.com/watch?v=tZkJl2fk0rc

https://news.ycombinator.com/item?id=33047634

What makes DVC especially useful for MLOps? Aren’t MLFlow or W&B solving that in a way that’s open source (the former) or just increases the speed and scale massively ( the latter)?

Disclaimer: I work at W&B.

[1]: https://m.youtube.com/watch?v=qNXebAQ6igs

[1]: https://huggingface.co/Aleph-Alpha/Pharia-1-LLM-7B-control

I’ll be visiting San Francisco in about 4 weeks and am hoping to get some book recommendations. Here’s what I’m looking for:

1. *For my kids (ages 3 & 5):* I’m particularly interested in science books for young children that are only available or easiest to find in the US. I’d love to bring back some unique and educational reads that might not be readily available in Germany.

2. *For myself:* I’m a machine learning engineer with interests in linguistics, philosophy, theology, natural history, and epistemology. If you know of any extraordinary books in these areas, especially those that are hard to find or are particularly renowned in the US, I’d love to hear your suggestions.

3. *Must-Reads:* Finally, if there are any other books—across any genre—that you consider a must-read, I’m all ears. Whether it’s fiction, non-fiction, or anything in between, I’m looking to expand my library with some great reads.

Also, any tips on where to buy these books in SF would be greatly appreciated!

Thanks in advance for your recommendations!

Comments URL: https://news.ycombinator.com/item?id=41212701

Points: 1

# Comments: 2

I’m going to watch finding NeMo now

https://targum.video/

Comments URL: https://news.ycombinator.com/item?id=35670591

Points: 1

# Comments: 1

Somehow, this time feels to me like the early days of computer science, when Don Knuth was barely known and a Turing award was only known to Turing award winners. I met Don Knuth in Palo Alto in March and we talked about LLMs. His take: „Vint Cerf told me he was underwhelmed when he asked the LLM to write a biography on Vinton Cerf.“

There are also tools being built and released for Prompt engineering [1]. Full transparency: I work at W&B

LangChain and other connecting elements will vastly increase the usability and combinations of different tools.

[1]: https://wandb.ai/wandb/wb-announcements/reports/Introducing-...

based on Whisper [0] and Whisper.cpp [1], I created a comparison of transcription performance (quantitative metrics such as relative speed).

You can find the code in the Colab [2] and a blog post [3] containing a how-to guide and visualizations.

In the future, I'd love to add WER evaluation and visualizations based on ground-truth data.

Bonus: Normally you would log these results from Python to Weights & Biases, but there is a way to log even from C++ / the cli by using `subprocess`

Would love to know what you think of this comparison and what features / attributes you would like to see in a more sophisticated comparison.

Thanks!

[0]: https://news.ycombinator.com/item?id=32927360 [1]: https://news.ycombinator.com/item?id=33877893 [2]: https://colab.research.google.com/drive/1mXZUdIbvdNVOFRJaIhW... [3]: https://wandb.ai/hans-ramsl/gradient-dissent-transcription/r...

Comments URL: https://news.ycombinator.com/item?id=34789339

Points: 5

# Comments: 0

[0]: https://github.com/ggerganov/whisper.cpp#whispercpp

It comes with a smart versioning approach, checks the Δ based on the checksum and has a feature to visualize the lineage.

You can also use your existing object store and link it for very large / sensitive data.[2]

Disclaimer: I work at W&B.

[1]: https://docs.wandb.ai/guides/data-and-model-versioning/model... [2]: https://docs.wandb.ai/guides/artifacts/track-external-files#...

> The truth in fact is that C. P. Snow got it wrong by at least an order of magnitude — there are many more than two cultures. I think a lot of you know the Apple Macintosh ads telling us to “think different,” but people already do. From my own corner of the academic world, I know for example that physicists think different from mathematicians; mathematicians who do algebra think different from mathematicians who do geometry; both kinds of mathematicians think different from computer scientists who work on algorithms; and so on and so forth. People often decry this lack of unity in the knowledge of the world, but let’s face it: People are different. Vive la différence.

Fun fact: PG mentions the two cultures while writing about Knuth [1]

[0]: http://web.stanford.edu/group/cslipublications/cslipublicati...

[1]: http://www.paulgraham.com/knuth.html?viewfullsite=1

[2]: https://en.wikipedia.org/wiki/Things_a_Computer_Scientist_Ra...

I do it to encourage young people to use software to make a change in their and other’s lives.

Currently making $500 with 90min/week effort. I already got the syllabus material ready so I do not necessarily have to invest time here.

Right now they’re writing an exam.

Comments URL: https://news.ycombinator.com/item?id=29680957

Points: 50

# Comments: 40