frontierkodiak

Maybe not 10x; but it's fantastically talented at intuiting intent, reconstructing contexts, and making aligned judgement calls. You could throw Fable utter garbage and get great results. Fable felt like it was modeling me whereas gpt-5.5 is still very much is riding your prompt, your inputs. I have bit of humility here as this is basically how I felt about 5.4->5.5; but 5.4 was very much a scalpel-very spiky weird intelligence. 5.5 sits somewhere in-between, but still spiky and verbose- code-maxxed; not a great orator, not a good proactive "here's the skip-connection you probably should be thinking about but don't seem to be weighing" in the way that Fable is. Fable is crisp.

At 6x the cost of its predecessor!

Apologies; I do not mean to imply any affiliation.

See my other comment in this thread! I've got you covered; launching open models in ~1month. Happy to answer any questions!

Hey, good news! Pollination ecologist + ML guy here; with open models coming soon.

You can keep an eye on (gh) polli-labs/linnaeus (a bit stale; I'll rebase on my private repo later tonight-). There's some cool ideas in here to exploit the structure of taxonomic hierarchies to help the model approach recognition how a professional taxonomist might.. so working from coarse to fine, taxonomy-guided label smoothing (distributing alpha mass by taxonomic distance)..and (forthcoming) RL on expert consensus to teach abstention (if an expert could only identify a specimen to genus for some set of inputs; then our model should abstain from a species classification for the same inputs). Unfortunately I am very, very compute-constrained- but shooting for late April/first week of May for insect + flowering plant models. (Other taxa will come later; probably as unified model). I'm working on camera-based (automated) ecological monitoring systems for ~6yrs at this point; it's a really fun problem space! dropped out of grad school to go all-in on automating my favorite job I ever had (pollination ecology field research..watching flowers for visitations!); since I knew I'd always be a mediocre ecologist- but an engineer that happens to care about ecology could be very very valuable to my field.

a taxa recognition model turns out to be only a small piece of the system you need to extract structured observational data from cameras in the field :-) Working with one of my partners right now to launch a really cool demo of what's possible these days- Texas folks especially; keep an eye out on wildflower.org around May 1!

I'll spill more ink soon but (anyone) please get in touch if you find these things interesting. Or if you'd like to help me out with compute/expenses!

oh man that's awesome. I have been working for quite some time on big taxonomy/classification models for field research, espec for my old research area (pollination stuff). the #1 capability that I want to build is audio input modality, it would just be so useful in the field-- not only for low-resource (audio-only) field sensors, but also just as a supplemental modality for measuring activity out of the FoV of an image sensor.

but as you mention, labeled data is the bottleneck. eventually I'll be able to skirt around this by just capturing more video data myself and learning sound features from the video component, but I have a hard time imagining how I can get the global coverage that I have in visual datasets. I would give anything to trade half of my labeled image data for labeled audio data!

I've owned 17 Seagate ST12000NM001G (12TB SATA) drives over the last 24mos in a big raidz3 pool. My personal stats, grouping by the first 3-4 SN characters: - 5/8 ZLW2s failed - 1/4 ZL2s - 1/2 ZS80 - 0/2 ZTN - 0/1 ZLW0 All drives were refurbs. Two from the Seagate eBay store, all others from ServerPartDeals. 7/15 of the drives I purchases from ServerPartDeals have failed, at least four of those failures have been within 6 weeks of installation.

I originally used the Backblaze when selecting the drive I'd build my storage pool around. Every time the updated stats pop up in my inbox, I check out the table and double-check that my drives are in fact the 001Gs.. the drives that Backblaze reports has having 0.99% AFR.. I guess the lesson is that YMMV.

Sure. The talk about 8bit refers to quantization-aware training. Pretty common in image models these days to reduce the impact of quantization on accuracy.

Typically this might mean that you simulate an 8bit forward pass to ensure that the model is robust to quantization ‘noise’. You still use FP16/32 for backward pass & weight updates for numerical stability.

It’s just a way to optimize the model in anticipation of future quantization. The experience of using an 8-bit Nemo quant should more closely mirror that of using the full-fat bf16 model compared to if they hadn’t used QAT.

YMMV, but I have tried this several times and have found it very difficult without microsoldering equipment.

Some ESP variants have a built-in U.FL connector which makes things easier (e.g. AITHINKER ESP32-Cam and its many clones), but you still have to remove the 0Ω resistor, which I find very difficult with even the smallest tips on my Hakko iron. Could be a skill issue on my end; should be much easier if you have a simple reflow/rework station.

Not to mention that you can trivially set a 300w power cap on the 4090 and still get 75-80% of peak FLOPS.

Tropical Tidbits is excellent.

Various hibernation modes are great, but 300ma@5v sometimes feels too hungry for the amount of computation in there. I bet things would be better if it wasn’t 45nm ‘pitch’. They are great chips, but require more power than you might expect if you’re doing something network-heavy or, god forbid, using the camera. Unfortunately there’s no competitive alternatives for the hobbyist.