coder68

This seems pretty useful for AI inference if it can pass Apple approval. I've wanted to use my Nvidia GPUs with a Mac Mini, this would enable it to run CUDA directly. Very cool!

What would be an example of a positive signal?

In fact it is appreciated that Qwen is comparing to a peer. I myself and several eng I know are trying GLM. It's legit. Definitely not the same as Codex or Opus, but cheaper and "good enough". I basically ask GLM to solve a program, walk away 10-15 minutes, and the problem is solved.

AI is much better at front-end than me, it has really enabled me to build visual apps as a normally backend/ML guy.

A bit more context would be helpful, as someone happy to donate -- what is the current situation, why the urgency? Just some more info would be good.

Hmm you might be able to tweak the settings further. Under llama.cpp on one RTX 6000 Pro I get ~215 tok/s generation speed. The key for me was setting min_p greater than 0. My settings:

``` #!/bin/bash

llama-server \ -hf ggml-org/gpt-oss-120b-GGUF \ -c 0 \ -np 1 \ --jinja \ --no-mmap \ --temp 1.0 \ --top-p 1.0 \ --min-p 0.001 \ --chat-template-kwargs '{"reasoning_effort": "high"}' \ --host 0.0.0.0 ```

I have not delved into the theory yet but it seems that the smaller open-source models do this already to an extent. They have less parameters, but spend much more time/tokens reasoning, as a way to close the performance gap. If you look at "tokens per problem" on https://swe-rebench.com/ it seems to be the case at least.

20 years is quite an optimistic timeline. Of course, we will use agents to solve the problems of agents!

I gave it a whirl but was unenthused. I'll try it again, but so far have not really enjoyed any of the nvidia models, though they are best in class for execution speed.

Are there plans to release a QAT model? Similar to what was done for Gemma 3. That would be nice to see!

120B would be great to have if you have it stashed away somewhere. GPT-OSS-120B still stands as one of the best (and fastest) open-weights models out there. A direct competitor in the same size range would be awesome. The closest recent release was Qwen3.5-122B-A10B.

The good news is local models have significantly improved. If it all goes down today, you can still run e.g. Qwen 3.5 at home, and it's "good enough" for most workloads.

With a gaming GPU you can run Qwen3.5-35B-A3B. I use 122B-A10B on my local rig (1x6000 Pro), and 397B-A17B on my 2x6000 Pro server (some spillover into CPU/RAM). It's pricey now but probably within a few years it'll become very affordable.

We all probably need to touch grass a bit. Our industry is really out of touch with reality right now, although the looming impact of AI is probably quite real.

Thanks! Super interested in LLMs for translation :D glad to see you folks doing this work.

It does? There is a fast drop followed by a long decay, exponential in fact. The cooling rate is proportional to the temperature difference, so the drop is sharpest at the very beginning when the object is hottest.

Is there interest in benchmarking the proprietary LLMs for translation? Curious as I often use Gemini 3 Flash, but I have no idea how good it is for my language family. I prefer open models (in fact the smaller the better for offline), but it'd be useful to know how well the Big Three do.

Even working in "tech" but not FAANG this is so true, 10 days is still the norm at many white collar businesses for your first year of employment, sometimes 15 days if they're generous.

The tradeoff with many EU countries would be that they enjoy their leisure time a lot more and sooner than Americans. Americans make more and save more statistically, but they spend it on cars, houses, and medical care, and generally have way less free time. So I think it's a wash.

As someone studying Polish, and making excellent progress, I mostly agree with your take. If you want to explore other languages, something like Spanish will get you much more mileage. Polish is difficult and the community of speakers isn't exactly warm to foreigners or people acquiring the language. On the other hand, if you truly enjoy languages and are passionate about them, I have found Polish to be really interesting and beautiful in its own way. Definitely not recommended, but still enjoyable to read/write/speak.

To chime in about where I'm at -- one problem was solved with a statistical classifier, but to bootstrap another, I ended up using keywords. It took a few hours to get a reasonable solution, and it leans more towards precision than recall, but it worked quickly!

To some degree manual labeling has to be done anyway, just to validate that any approach works at all, you'll always need ground truth from somewhere. What I suggested is that zero/few-shotting might not be good enough, depending on the problem. Labeling ~1000 samples isn't too bad, I've done it by hand a few times now. If you can source a high quality positive signal from somewhere (e.g. user-behavioral data), even better.

oh this seems like an interesting idea, what tactics do you use for augmentation? For my own use-case, I think I could reorder semantic chunks, or maybe randomly delete pieces, but curious what tactics you use!

I have also considered training a small language model for synthetic data generation.

The outputs are working correctly in terms of formatting, but the answers themselves may be inconsistent. I have experimented with varying the prompt and the answers can change dramatically. I could experiment with lowering temperature, but I just don't think generative models were a good fit for the problem. The appeal is the speed of prototyping and no need for training data, but it honestly didn't take much for my problem: one afternoon and ~1000 samples labeled got me to a good baseline.

I can confirm that Distillbert has worked well when I have used it for classification, especially on shortish sequences. I'm really interested in trying out ModernBert, or a smaller variant due to the larger context window (8192 tokens).

I think the idea of backoff by ratcheting up complexity here is a very good idea, thanks for your suggestions.

Ah yes this does make sense. We are definitely in agreement on the point of "wildly inefficient and subpar". I'll try out decoder model embeddings soon, e.g. Qwen/Qwen3-Embedding-8B. I'm working with largish amounts of data (200M records), so I tried to pick a good balance between size:perf:cost, using BAAI/bge-base-en-v1.5 to start (384 dim).

By LLMs I meant decoder only, e.g. Gemini, Claude, etc. Can you go into more detail on how you're using the encoder models? I'm curious. Typically I have used them for embedding text or for fine-tuning after attaching a classifier head. What are you pre-training on, and for what task?

Do you have any sources/links that talk about this? I'm very interested in synthetic data generation, so curious what you've tried or what works / doesn't work, especially with regards to LTR.

I have been working on text classification tasks at work, and I have found that for my particular use-case, LLMs are not performing well at all. I have spent a few thousand dollars trying, and I have tried everything from few-shot to asking simple binary yes/no questions, and I have had mixed success.

I have stopped trying to use LLMs for this project and switched to discriminative models (Logistic Regression with TFIDF or Embeddings), which are both more computationally efficient and more debuggable. I'm not entirely sure why, but for anything with many possible answers, or to which there is some subjectivity, I have not had success with LLMs simply due to inconsistency of responses.

For VERY obvious tasks like: "is this store a restaurant or not?" I have definitely had success, so YMMV.