sublimefire

8 ms·

91.
by sublimefire 10mo ago
A service that allows you to post generic agent/task workload and execute and observe it in the background. Sort of for long timelines. I just want to have background agents to monitor various aspects of my digital life and keep those
92.
by sublimefire 10mo ago
The problem is that we are still talking only about software. If that is the only thing agents can do then they are just better IDE’s for programmers. This would be an enabler for building stuff faster. Think about it, there is already a hu
93.
by sublimefire 10mo ago
Good discussion of the paper and the observations and ironies. A thing to note is that we do have software factories already, with a bunch of automation in place and folks being trained to deal with incidents. The pools of agents just eleva
94.
by sublimefire 10mo ago
There should be an existing simpler way to do it. Image contains a bunch of pixels so you could just group the to see main colors
95.
by sublimefire 10mo ago
This is just signing, nothing cutting edge. Verification of signatures is a fairly old tech. What is the exact problem here? Is it that manufacturers do not publish the signed binaries or is it that you want to run something on your car com
96.
by sublimefire 10mo ago
One of the points was that European manufacturers will start making more cars in US purely because it is cheaper to do so due to the lower bar. Why would we want that? Our market is quite big anyway and this agreement is an attempt to shoul
97.
by sublimefire 10mo ago
It makes sense in US but not in Europe where public transport infra is good. There are so many places where robotaxis would just be stuck on narrow European roads. Why would I even use a robotaxi when I can get one with the actual person wh
98.
by sublimefire 11mo ago
The problem is how to even define a task using the English language and make sure there is enough entropy to infer the detailed intent. For it to be later split into zillions of small steps which can be executed over time by an LLM.
99.
by sublimefire 11mo ago
Great point, it might be just pure ignorance. Even OSS pentesting tooling such as metasploitable have great capabilities. I see how LLM could be leveraged to build custom modules on top of those tools or how can you add basic LLM “decision”
100.
by sublimefire 11mo ago
Sorry I was meaning to say "passport terminal" capabilities which would require a cert to be issued by a country whose passport chips you want to read. Well maybe they had this for a while but AFAIK you could not read passport det
101.
by sublimefire 11mo ago
There was a post from trailofbits blog recently about how passport crypto works. Kind of related here. I wonder if this is some zero knowledge proofs here or what? Reading the passport and its chip implies some terminal authentication capab
102.
by sublimefire 11mo ago
Did not know about it much before but apparently was doing something similar (in spirit) when experimenting with simple task execution queues in go. This is quite interesting and given that hot code reloading is a thing in beam it might be
103.
by sublimefire 11mo ago
A pretty thin opinion piece, I was expecting more details. But there are a bunch of comments under that article which is probably juicier than the main text.
104.
by sublimefire 11mo ago
It would be great to use this moment and do something like Cyber Resilience Act CRA to force companies to deal with the cybersecurity issues.
105.
by sublimefire 11mo ago
Yes but at the end of the day you need to trust the cloud provider tools which expands the trust boundary from just hardware root of trust. Who is to guarantee they will not create a malicious tool update and push it then retract it? It is
106.
by sublimefire 11mo ago
Quite similar to what Azure with conf ai inference did [1]. [1] https://techcommunity.microsoft.com/blog/azureconfidentialco...
107.
by sublimefire 11mo ago
If you trust the provider then it does not make it much better to use such architecture. If you do not then at least the execution should be inside a confidential system so that even soldering would not get you to data
108.
by sublimefire 11mo ago
Yes it seems like it is absent in this discussion but maybe it should have been “it” the whole time as a default option. I wonder if it could attain similar throughput numbers; bet the article would feel slightly sarcastic then though
109.
by sublimefire 11mo ago
With the advent of tools like llms in editors, it is now viable to create clients and solve these gaps quite easily. It feels like the next low hanging fruit to do in many places not client friendly enough.
110.
by sublimefire 1y ago
Surely that would need to be both input and output. But even then you could easily get lost in translation as the intent in one language might mean slightly different thing in another. Thus you could get subpar results.
111.
by sublimefire 1y ago
It seems there is some weird grouping of the language data which LLM cannot distinguish well. I wonder if it is the same for other similar languages like scandinavian or western slavic
112.
by sublimefire 1y ago
As a European citizen I think it boils down to access to the capital. EU/EEA is not a country and the market is sort of fragmented. The big players are UK, France, Germany, everyone else does not have the same access to money as say in
113.
by sublimefire 1y ago
It is just one of the theories, there is no clear evidence to suggest that Baltic and Slavic were the same language thousands of years ago.
114.
by sublimefire 1y ago
Another angle to look at it is that cars cost a lot, petrol and electric alike. The only way to reduce depreciation is for the folks to be able to buy used cars for a lot which means financing. > A U.K.-based study found 3-year-old EVs l
115.
by sublimefire 1y ago
Without delving into the specifics Ireland already has a defacto UBI. This is subject to you having an address and a social security number. There is a notion of a minimum income you need to make and the State compensates the rest if you do
116.
by sublimefire 1y ago
The question is if the company can add more value to the models than someone else. I still see a lot of gaps in the ecosystem, eg evaluation/testing systems and integrations beyond chat interface and active control to get good results,
117.
by sublimefire 1y ago
How is this bad? I welcome the decision and the outcome. It is beyond crazy what some groups are doing with the misinformation and outright lies on the platforms. Ad companies just want to rake the money with the least effort. They have sho
118.
by sublimefire 1y ago
I would think this is naive to imagine that targeted ads do not have influence when used in large numbers. There is a reason why this market exists. I am happy this law has an effect and potentially opens the doors to other players.
119.
by sublimefire 1y ago
I find using “Vibe” makes it sound less valuable. It also needs to fit the next frontier which is “refactoring” and tuning the codebase to work with agents to make it easier to change existing code. A suggestion to use “augmented” was great
120.
by sublimefire 1y ago
It’s kind of useful but I suppose they just admit that failure rate increases with large context windows. My guess is that what happened to the presentation of those Meta glasses where the model would not do what was asked for. Another inte

More ›