pilooch

Yes, full ft or lora https://github.com/NVIDIA-NeMo/Automodel/blob/main/docs/guid...

Yes, all my emails gyer sorted out by a finetuned gemma. There are turned into images passes to the model, as multimodal is so practical.

Reminded me of the recent, and excellent, Canadian tv series Empathy, with the main character is found in a garbage can by his adoptive parents.

AlphaEvolve couples map-elites with LLMs. It's an key step in machine learning, in the vein of DQN for reinforcement learning.

AE brings diversity from the genetic algorithms community to large scale optmized deep learning and RL models.

It is a mandatory step for moving forward. The approach is clean and simple, while generic.

The only caveats is the per optimization problem definition of the map élites dimensions. But surely, this will get tackled somehow over the next few years.

If you don't know about map-elites, go look up Jean-Baptiste Mouret' s work and talks, it's both very interesting and universal.

Slides, publications and tech reports, very handy for figures !

It's useful when using prism, and for exploratory research & code.

Revolting and so inevitable though I believe: we're sort of running these already in our minds, we'll be outrun here too.!

Good catch. Cancer treatment scheduling is hard as well as mixes need tombe prepared in advance and cancelles appointments are hard to fill.

Hello! Not commenting on content or functionality. Scheduling in AI is a very dense field. An a past researcher in AI decision making, I got confused by the 'Scheduling solved' slogan. FYI recent AI for scheduling include GNNs and RL applied to NP and P-space problems that plague many industries. A larger scope I believe from vela's (rightful) target, a bit confusing IMO. Good luck with your endeavor, all scheduling problems are beautiful :)

Try Flashback, it's darker but genius as well, maybe more approachable.

Synthetic data for human (machine) learning... We should spend more time outside, we will!

The 'Eclipse' album is a classic.

I like the glasses path, well I do wear glasses, but some elements remain unclear to me:

- are prescription glasses available for display ? I guess not ? - these glasses need to be online, I guess they do so with a phone and bluetooth connection nearby ? So that's the glasses, the band and the phone, oh and the glasses case, seems a lot to carry. - pedestrian navigation seems to be rolled out per city, so it's not like having gmaps available right out of the box.

Congrats, this solution resembles AlphaEvolve. Text serves as the high-level search space, and genetic mixing (map-elites in AE) merges attemps at lower levels.

Losing the mental map is the number one issue for me. I wonder if there could be a way to keep track of it, even at a high level. Keeping the ability to dig in is crucial.

Hello, very interested in the scrollback! I've used mosh for 10+ years and it still runs my 100+ opened terminals to this day ! Would love to try your alternative

Exactly, for real time applications VTO, simulators,...), i.e. 60+FPS, diffusion can't be used efficiently. The gap is still there afaik. One lead has been to distill DPM into GANs, not sure this works for GANs that are small enough for real time.

Is this a bit similar to what tensorrt does, but in a more opened manner ?

Because they'd never hire, but subcontract down to the bone.

I'd be interested in what implementation of D3PM was used (and failed). Diffusion model are more data efficient than their AR LLM counterpart but les compute efficient at training time, so it'd be interesting to know whether with more time.to.converge the diffusion approach does succeed. I guess I'll try :)

True but modern models such as gemma3 pan& scan and other tricks such as training from multiple resolutions do alleviate these issues.

An interesting property of the gemma3 family is that increasing the input image siwmze actually does not increase processing memory requirements, because a second stage encoder actually compresses it into fixed size tokens. Very neat in practice.

Good catch, will add it tomorrow. License is Apache2.

Some colleagues and myself did implemented exactly this six months ago for a French gov agency.

It's open source and available here: https://github.com/jolibrain/colette

It's not our primary business so it's just lying there and we don't advertise much, but it works, somehow and with some tweaks to get it really efficient.

The true genius though is that the whole thing can be made fully differentiable, unlocking the ability to finetune the viz rag on targeted datasets.

The layout model can also be customized for fine grained document understanding.

AlphaEvolve and similar systems based on map-elites + DL/LLM + RL appears to be one of the promising paths.

Setting up the map-elites dimensions may still be problem-specific but this could be learnt unsupervisedly, at least partially.

The way I see LLMs is as a search-spqce within tokens that manipulate broad concepts within a complex and not so smooth manifold. These concepts can be refined within other spaces (pixel -space, physical spaces, ...)

Fix: it's the E2B

This model is fully compatible with anything previously done with gemma3. Just passed it to one of my vlm fine-tuning scripts and it started without issues (hf transformer code). On a single GPU with Lora the E4B model takes 18Gb of VRAM in batch size 1 where gemma-4B was 21Gb. Nice one from deepmind, the gemma3 family tops the open weights VLLMs.

Yes the 2023 reference on island based evolution with LLMs (nature article) https://www.nature.com/articles/s41586-023-06924-6 has more details.

Agreed the dimensions/features are key. These white papers are an insult to science...

An inspiration to Avatar maybe!

Using it for a RAG is smart indeed, especially with a multimodal encoder (vision-rag), as the implementation would be straightforward from what you already have.

Or just we could forget about code and have model act directly :) That's my bet.