runeblaze

no profile record

comments

19 hari yang laludiscuss

tbh the summarized thinking with encrypted raw thinking is there for many purposes; it is there to:

1. make distillation much harder

2. safety: prevent modifications to the thinking leading to injection attacks.

3. also honestly sometimes the model raw thoughts can be deranged and is not a good user experience (consider the varied audience in the market, etc.)

also often the mass underestimate/the model makers over-estimate how people love distilling models

20 hari yang laludiscuss

links to two papers with at least enough apparent quality and novelty to get into ICLR 2026

> So basically... openrouter

:skull:

i now really wonder how many people of the public understood my thesis defense lol

3 bulan yang laludiscuss

> visual similarity

> SigLIP 2

Maybe visual-semantic similarity is more appropriate? Nonetheless the design is fantastic

3 bulan yang laludiscuss

I mean I used to work on model reliability with my little PhD degree and the models i manage go down all the time.

Some profs have a team of PhDs and things go to shit all the time. I don’t know why we expect $FRONTIER_LLM to do better

4 bulan yang laludiscuss

1. openrouter is API usage. There is obviously consumer side

2. people often use openrouter for the sole purpose of using a unified chat completions API

3. OpenAI invented chat completions; if you use openrouter for chat completions often you can just switch your endpoint URL to point to the OAI endpoint to avoid the openrouter surcharge!

4. Hence anyone with large enough volume will very likely not use openrouter for OpenAI; there is an active incentive to take the easy route of changing the endpoint URL to OAI’s

6 bulan yang laludiscuss

Schemas can get pretty complex (and LLMs might not be the best at counting). Also schemas are sometimes the first way to guard against the stochasticity of LLMs.

With that said, the model is pretty good at it.

6 bulan yang laludiscuss

Resouce-affording, if you are chasing the frontier of some more niche task you redo your training regime on the new-gen LLMs

6 bulan yang laludiscuss

Is it though? There is a reason gpt has codex variants. RL on a specific task raises the performance on that task

6 bulan yang laludiscuss

Sure never again is totally fair and I am sure a lot of people hate it. I was mostly objecting to the radioactivity of it. Your friends will be more like “I am looking to sell my Tesla in 3 months” if it is truly radioactive.

Let’s be realistic in our portrayal here.

6 bulan yang laludiscuss

I think radioactive is a strong word here… I have talked to a lot of people in tech

6 bulan yang laludiscuss

Reading what you wrote scares me

7 bulan yang laludiscuss

> And if for some ungodly reason you had to do it in Python

I literally invoke sglang and vllm in Python. You are supposed to (if not using them over-the-network) use the two fastest inference engines there is via Python.

8 bulan yang laludiscuss

Agreed, I am surprised he is happy to stay this long. He would have been on paper a far better match at a place like pre-Gemini-era Google

8 bulan yang laludiscuss

I don't know the data distribution, but are you sure that's generated by an Adobe model? I can only see that it is in Stock + it is tagged as AI generated (that is, was that image generated by some other model?)

Disclaimer: I used to work at Adobe GenAI. Opinions are of my own ofc.

8 bulan yang laludiscuss

Emmmm sure, but throw this to a human artist who has not heard of Indiana Jones and see if they draw something alike.

8 bulan yang laludiscuss

I work in this space. In traditional diffusion-based regimes (paired image and text), one can absolutely check the text to remove all occurrences of Indiana Jones. Likewise, Adobe Stock has content moderation that ensures (up to human moderation limit) no dirty content. It is a world without Indiana Jones to the model

8 bulan yang laludiscuss

My personal mantra (that I myself cannot uphold 100%) is that every dev should at least do the exercise of implementing binary search from scratch in a language with arbitrary-precision integers (e.g., Python) once in a while. It is the best exercise in invariant-based thinking, useful for software correctness at large

9 bulan yang laludiscuss

each text token is often subword unit, but in VLMs the visual tokens are in semantic space. Semantic space obviously compresses much more than subword slices.

disclaimer: not expert, on top of my head

9 bulan yang laludiscuss

beats me. I spent so much time learning what a fundamental group is and I still cannot tell ppl what a fundamental group is convincingly.

I can’t even make stuff with fundamental groups.

9 bulan yang laludiscuss

Ummm guys when we talk about memory access in theory can we just be rigorous and talk about the computational model?

The real RAM model “in theory” tells me that memory access is O(1). Of course real RAM is a spherical cow but like we could use a bit more rigor