> I'm sure I would have loved to have a real phone as a toddler, but my sister and I got a lot of mileage out of our Fisher Price phone, making lots of pretend calls, which was probably better for our developing brains…
> ~1000 bytes per context token per user Where's that figure comning from? Last time I checked (could be the 3.6 Qwen 27b) single token needed 32kb
I still can't fathom my company application security decision. Found pretty damning requests in our logs. Escalated. Expected it would result in at least reporting the TOS break from the originating place (one of cloud…
It is? I still found it invaluable, when having a lot of externally managed knowledge. Basically the only gate before it goes on a certain failure, when dealing with proprietary libraries, tools and services
Brain fog? This might be a legitimate symptom worth seeing a doctor for
Supposedly having over 125 (130-ish) can affect "normal functioning", it's very counterintuitive, but I'd recommend to just go through common psychology checks for that case.
Have you ever gotten an IQ test?
Care to share any details? I'm about to check the 2.6b lfm on document editing.
It does? I wonder, what it does to the gut biome then
The needle benchmarks show, that models extended context works for the part, that can be explained as: "I can access/adress that part of the input". I have no idea, why in that context, the number of attention heads…
I have a feeling this comment will make history
It's a mcp, so connects quite easily to agents. With mcpo, I also connected it to open-webui (which has better support for OpenAI style tools/functions). Used it in claude code with that mcp plugin set-up too. Only ever…
With 3.6 27b, I just stopped changing local models and started tinkering with things on top (like mem0). Feels genuinely useful and more than a toy
Also explains why we like music with two simultaneous distinct sections (bass + the rest). One without the other doesn't feel as complete
Disclaimer, I only use it to grow the "knowledge hub". It's a single git project at my $USER home, that is referenced in global memory. It contains as much information about work things, as possible, to be productive. I…
That's where the annotation in plannotator helps. I'm asking to scan projects on gitlab, go through some docs to find more grounding material, write a subarticle (in the same style), scan logs on the test env, issue…
Oh, I know exactly what you mean. I use plannotator with claude a lot and have much better time, since I asked for a specific styleguide. I used "CD era MSDN reference and Raymond Chen blogging style" as a starting…
Similarly, I never try to imagine a bright sunshine. It can wake me up
There never are. Those are going to be viewed as two discreet successful interventions. One for lay-offs, because it was the best move at the time with the knowledge they had. Second for quick correction, ability to…
I remember seeing this popping up in discussions the first time, but never noticed any resolution (other than to train both sides). Has SOTA advanced?
> A shift is happening among major AI labs, who are becoming increasingly skeptical of endless parameter count and training data scaling I'm pretty sure it's mostly due to the training data quality. No idea, why this…
Noticed few cliffs. Sometimes it was a spurious stop (had to write "go on" or "continue" to restart), othertimes it was randomly saying: "Oh the user wants [the thing we already resolved]" and goes back in history.…
If I started today, with building a server, I'd jump right into verified set-ups and writeups, like this one: https://github.com/noonghunna/club-3090 You can find info about running a patched version of vllm for 1x24gb,…
> The model is running so hot, that it shoots past the goal and starts looping later: > My latest experiment was setting up vLLM (the gold standard for production and concurrent serving) and even with an NVLink (175GBP)…
I'm still sour they had only one toast in, in a two slot toaster
> I'm sure I would have loved to have a real phone as a toddler, but my sister and I got a lot of mileage out of our Fisher Price phone, making lots of pretend calls, which was probably better for our developing brains…
> ~1000 bytes per context token per user Where's that figure comning from? Last time I checked (could be the 3.6 Qwen 27b) single token needed 32kb
I still can't fathom my company application security decision. Found pretty damning requests in our logs. Escalated. Expected it would result in at least reporting the TOS break from the originating place (one of cloud…
It is? I still found it invaluable, when having a lot of externally managed knowledge. Basically the only gate before it goes on a certain failure, when dealing with proprietary libraries, tools and services
Brain fog? This might be a legitimate symptom worth seeing a doctor for
Supposedly having over 125 (130-ish) can affect "normal functioning", it's very counterintuitive, but I'd recommend to just go through common psychology checks for that case.
Have you ever gotten an IQ test?
Care to share any details? I'm about to check the 2.6b lfm on document editing.
It does? I wonder, what it does to the gut biome then
The needle benchmarks show, that models extended context works for the part, that can be explained as: "I can access/adress that part of the input". I have no idea, why in that context, the number of attention heads…
I have a feeling this comment will make history
It's a mcp, so connects quite easily to agents. With mcpo, I also connected it to open-webui (which has better support for OpenAI style tools/functions). Used it in claude code with that mcp plugin set-up too. Only ever…
With 3.6 27b, I just stopped changing local models and started tinkering with things on top (like mem0). Feels genuinely useful and more than a toy
Also explains why we like music with two simultaneous distinct sections (bass + the rest). One without the other doesn't feel as complete
Disclaimer, I only use it to grow the "knowledge hub". It's a single git project at my $USER home, that is referenced in global memory. It contains as much information about work things, as possible, to be productive. I…
That's where the annotation in plannotator helps. I'm asking to scan projects on gitlab, go through some docs to find more grounding material, write a subarticle (in the same style), scan logs on the test env, issue…
Oh, I know exactly what you mean. I use plannotator with claude a lot and have much better time, since I asked for a specific styleguide. I used "CD era MSDN reference and Raymond Chen blogging style" as a starting…
Similarly, I never try to imagine a bright sunshine. It can wake me up
There never are. Those are going to be viewed as two discreet successful interventions. One for lay-offs, because it was the best move at the time with the knowledge they had. Second for quick correction, ability to…
I remember seeing this popping up in discussions the first time, but never noticed any resolution (other than to train both sides). Has SOTA advanced?
> A shift is happening among major AI labs, who are becoming increasingly skeptical of endless parameter count and training data scaling I'm pretty sure it's mostly due to the training data quality. No idea, why this…
Noticed few cliffs. Sometimes it was a spurious stop (had to write "go on" or "continue" to restart), othertimes it was randomly saying: "Oh the user wants [the thing we already resolved]" and goes back in history.…
If I started today, with building a server, I'd jump right into verified set-ups and writeups, like this one: https://github.com/noonghunna/club-3090 You can find info about running a patched version of vllm for 1x24gb,…
> The model is running so hot, that it shoots past the goal and starts looping later: > My latest experiment was setting up vLLM (the gold standard for production and concurrent serving) and even with an NVLink (175GBP)…
I'm still sour they had only one toast in, in a two slot toaster