Hacker Newsnew | past | comments | ask | show | jobs | submit | Arcuru's commentslogin

demonstrate != explain?


Lots of GDP entries there. Uh. Like terabytes of it downloaded.

I would quite like to think that GDP references... https://en.wikipedia.org/wiki/GirlsDoPorn - where "six people involved with the website were charged on counts of sex trafficking by force, fraud, and coercion".

A little yikes on that card today. Dude was ripping hundreds of videos of assault and coerced rape pornography for training.


> starting from a strong open-source foundation and investing $40 million to train Thomson

Sounds like they spent $40 million finetuning an open weight model on their own data? I wonder what they built on.


From the HF link posted above it's Qwen3.6-35B-A3B.

https://huggingface.co/thomsonreuters/Thomson-1.0-Small


Thanks! And from the Technical Report linked in there it looks like their Large unreleased model is a trained Qwen3.5-397B


Per Business Insider [1], it was built on Qwen.

[1] https://www.businessinsider.com/thomson-reuters-builds-ai-mo...


Anecdotal, but from my personal usage I found GLM-5.3 was not as capable at performing autonomous tasks as Opus/Sol. Certainly competitive with the Sonnet/Terra level, but not with the Frontier.

I got their lowest subscription tier and burned a week of quota on trialling it.


Be nice, maybe this person isn't very tech-savvy and doesn't know how to setup their own blog?

It's not like they are the highly technical founder of a very successful, privacy-focused messaging app....


I think it's a design mockup, they didn't implement it yet.


So this is a new design, but it's at the design proposal stage and not yet implemented?

They've been talking about redesigning Thunderbird for years now, but I'm honestly not sure I've noticed any difference at all in my actual use.


Since I don't see it mentioned yet, https://blessed.rs seems to be pretty popular for finding common/recommended crates in the Rust ecosystem.


Awesome! I've been waiting for them to start scaling ternary models for over a year[1]. Excited to try it out, typical Qwen 27B is too heavy for me to run on my local hardware at reasonable speeds.

[1] https://jackson.dev/post/dont-sleep-on-bitnet/


Same here. I’m excited to have a model that might be usable on a 16 GB laptop.


I've continued working on Eidetica, my decentralized database project. I recently added support for a client/server architecture so that it can be transparently run as a local daemon for background sync and sharing the local storage with multiple users. I've been making progress on integrating blob storage next, as well as scoping out WASM based "lenses" for handling decentralized version/schema updates. https://github.com/arcuru/eidetica

I've primarily been testing it by building out my AI tool chaz into an Eidetica-native AI Agent framework for decentralized Agent sessions. It's working surprisingly well, it maps pretty well onto the storage model and it's uncovering issues with Eidetica I need to fix (which was always my primary reason for building it anyways). https://github.com/arcuru/chaz

Separately I'm building OptiMap, a SIMD-accelerated hashmap repo that explores the design space for hashmaps and benchmarks different approaches. This is mostly for my own learning but I'll eventually turn into a blog post. https://github.com/arcuru/optimap


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: