dannyw

14 ms·

361.
by dannyw 4mo ago
I might be looking at the Kinect with rose-tinted glasses, but bringing depth and camera-based pose tracking to the masses in 2010 is pretty impressive. Sure they were chasing the Wii, but they did try to innovate on the hardware and capabi
362.
by dannyw 4mo ago
I think a lot of people in the gaming community would agree that Microsoft has ruined, or hindered (instead of helping) many studios they bought.
363.
by dannyw 4mo ago
The scale is also very different. Costco earned $70.53 billion in a recent quarter (not 5B here). Their operating margin is 3.93%, which is also healthier.
364.
by dannyw 4mo ago
Eh. Refusals for security related tasks seem to be constantly increasing.
365.
by dannyw 4mo ago
There are lots of ethical concerns around baby formula that’s not prevalent in the case of AI subscription token subsidies. Also, not everyone uses their plan to the usage limits; the idea of a prepaid bundle being better value if you commi
366.
by dannyw 4mo ago
I really love the coil whine of my GPU (a 5090 FE) when it’s doing LLM stuff. I can hear the different stages, like prefill and decode, and the sounds actually make me reminisce about dial up.
367.
by dannyw 4mo ago
Huge kudos for sharing everything including your training code. Awesome project!
368.
by dannyw 4mo ago
Basically like passes@6 or passes@5 if you’re doing a benchmark, except for your real tasks. Pro is quite limited on the web UI I reckon. This approach can be highly effective for reasonably verifiable task, for example, write comprehensive
369.
by dannyw 4mo ago
How are you so sure that frontier API models are always running the same quant/weights/etc? You think OpenAI and Anthropic are running essentially just vLLM endpoints? Of course not. Firstly, we know Anthropic has been doing promp
370.
by dannyw 4mo ago
I wonder if testing during different time/days show patterns? For example, whether the short circuiting happens more often during workday peak hours.
371.
by dannyw 4mo ago
It’s more than power supply. Cooling and ventilation becomes a MUCH bigger deal at rack scale, and that costs electricity too.
372.
by dannyw 4mo ago
Your comment is helpful, useful, and relevant. Please keep going. Just my 2c, but maybe your original post could’ve been written in a less promotional / less excited way. It looked a little like spam at _first_ glance.
373.
by dannyw 4mo ago
It’s a 6bn model. Totally different class. I’m more excited about “frontier small language models” tbh.
374.
by dannyw 4mo ago
It's released and signed by GitHub I believe (although not deterministic builds), but there's at least a little bit of provenance that you're getting the real repository.
375.
by dannyw 4mo ago
The theory is probably Deepseek might be collecting those streams, and sending a portion of it to Anthropic to see what the Anthropic/Opus response would be.
376.
by dannyw 4mo ago
I purchased mine for approximately $4400 AUD before the price hikes. That unit is now ~$5100 AUD. I use my MBP essentially as my workstation, it's almost always plugged in. I have a MBA (M4, 24GB RAM) that I picked up for ~A$1500 or so
377.
by dannyw 4mo ago
I’m running the same model on a 48GB MBP with a q4 quant and it’s pretty decent. You definitely don’t 128GB. That’s the scale for 70B models at q8 or something.
378.
by dannyw 4mo ago
Llama.cpp uses the GPU very effectively because inference of LLMs is very rudimentary and basically as simple as your GPU memory bandwidth. That's essentially the baseline performance ceiling, with model-specific optimisations like MTP
379.
by dannyw 4mo ago
Good luck, getting rid of PDFs is going to be as hard as migrating from JPEG everywhere.
380.
by dannyw 4mo ago
Two parts here. First, well-calibrated systems for detecting API compromise is a good thing (or good intent at least). Credential malware is exploding. Second, the challenge is that significant amount of genuine work — such as evals — seems
381.
by dannyw 4mo ago
RLAIF is a good place to start reading. Claude will also help you with (mostly good advice) if you ask something like “Research and help me make the most effective plan to train a smaller student model to be better from a teacher model”. I
382.
by dannyw 4mo ago
Why use VPNs? Just use a public cloud like AWS, or something like Linode and Vultr and all that. Developers use devboxes on these clouds all the time, it’s totally normal behavior. Most people buying these Chinese resold tokens are probably
383.
by dannyw 4mo ago
Residential IPs don’t even matter. Developers use devboxes, use Claude Code CLI on servers from just about every cloud, etc. There’s probably a decent volume of customers who just buy Claude Max and spend most if not nearly all of their ses
384.
by dannyw 4mo ago
If you’re doing evals, you’re basically doing RLAIF without training a model; just looking at the results. Fundamentally it is very difficult to stop this while still making your AI models useful.
385.
by dannyw 4mo ago
Don’t trust my experiences as fact since it’s a bit opaque, but I believe 20x only offers 4x the 5hr session limits. The weekly limit is still 2x, which is the same as the price increase.
386.
by dannyw 4mo ago
And it’s perfectly normal to be running Claude Code on EC2, a VPS, etc. I do it all the time! You block clouds, you block devboxes and your customers.
387.
by dannyw 4mo ago
I use my Codex and Claude Code subs on like 4-6 different servers, ranging from AWS to Vultr to Linode etc. That’s a major and legitimate use case for developers, Anthropic can’t just block data center/hosting IPs because their actual
388.
by dannyw 4mo ago
To a large extent, all these datacenters are being built because people want tokens. Do you use something like Opus or GPT5.5 for coding? Well, you're part of the demand. If you don't, then yes, as a PC enthusiast I'm sad as
389.
by dannyw 4mo ago
Generally 97.5% token agreement is very positive. Like the article explains, the difference isn’t the model thinking the capital of France isn’t Paris, but rather maybe saying “The capital of France is Paris” instead of “Paris is the capita
390.
by dannyw 4mo ago
Correct. The main bottleneck with LLM inference is, and have always been, memory bandwidth. TPS = active weights in GB / your memory bandwidth. That’s it for decode. That’s all.

More ›