Omnicrola's recent activity
-
Comment on The busiest place you've never seen [the island of Tristan da Cunha] in ~travel
-
The busiest place you've never seen [the island of Tristan da Cunha]
30 votes -
Comment on Google releases Gemma 4 in ~comp
Omnicrola Link ParentThat's a really interesting project! What kind of errors is whisperx introducing that you're seeing? I haven't gotten to the cleanup phase of my project yet, but I'm sure I'm going to have to do...That's a really interesting project! What kind of errors is whisperx introducing that you're seeing? I haven't gotten to the cleanup phase of my project yet, but I'm sure I'm going to have to do some, since a lot of the words aren't "real" and basic phenoms detection can only go so far.
re: your prompt - have you tried providing examples? I got very mediocre results from my model until I gave it a multishot example of "this transcript should result in this output" along with the base prompt.
-
Comment on Google releases Gemma 4 in ~comp
Omnicrola Link ParentAre you willing to share more? This sounds similar to a pet project I'm currently working on; extracting specific data (not summarized) from podcast transcripts. Currently I'm splitting the...(I'm trying to provide a blow-by-blow of TTRPG sessions)
Are you willing to share more? This sounds similar to a pet project I'm currently working on; extracting specific data (not summarized) from podcast transcripts. Currently I'm splitting the transcript into chunks and running it through a Qwen3 model, and getting pretty good results. On par with running the same transcript through Claude Opus.
-
Comment on Here’s what the world had to say about the AI economy in ~tech
Omnicrola Link ParentI interpreted this from a slightly different angle. Think of "faith leader" not as your minister or pastor, but as "the ministers in my community, who are ostensibly trying to make thier...This one is honestly surprising to me. I'm not religious or involved with community groups, but why would you be involved in those organizations if you didn't trust their leaders?
I interpreted this from a slightly different angle. Think of "faith leader" not as your minister or pastor, but as "the ministers in my community, who are ostensibly trying to make thier communities better, in addition to preaching thier faith".
Even if the majority of people in a community "go to church", they don't all go to the same church, or even belong to the same faith. So it would be incorrect to assume that if someone goes to church and does trust their own minister, that doesn't necessarily mean they trust all of them.
-
Comment on Claude Code's source code leaked in ~tech
Omnicrola LinkWhile it didn't fully give away the farm, it pretty much gives away all the land around the farm. So to speak. There's a lot of interesting things to learn from this chunk of code that I'm sure...Anthropic appears to have accidentally revealed the inner workings of one of its most popular and lucrative AI products, the agentic AI harness Claude Code, to the public.
A 59.8 MB JavaScript source map file (.map), intended for internal debugging, was inadvertently included in version 2.1.88 of the @anthropic-ai/claude-code package on the public npm registry pushed live earlier this morning.
The most significant takeaway for competitors lies in how Anthropic solved "context entropy"—the tendency for AI agents to become confused or hallucinatory as long-running sessions grow in complexity.
The leaked source reveals a sophisticated, three-layer memory architecture that moves away from traditional "store-everything" retrieval.
Perhaps the most discussed technical detail is the "Undercover Mode." This feature reveals that Anthropic uses Claude Code for "stealth" contributions to public open-source repositories.
The system prompt discovered in the leak explicitly warns the model: "You are operating UNDERCOVER... Your commit messages... MUST NOT contain ANY Anthropic-internal information. Do not blow your cover."
While Anthropic may use this for internal "dog-fooding," it provides a technical framework for any organization wishing to use AI agents for public-facing work without disclosure.
While it didn't fully give away the farm, it pretty much gives away all the land around the farm. So to speak. There's a lot of interesting things to learn from this chunk of code that I'm sure all the other major AI players are currently analyzing in earnest to see how they can apply it to their own products.
-
Claude Code's source code leaked
50 votes -
Comment on No Kings protests live updates: hundreds of thousands rally in cities around the world against US President Donald Trump and his administration in ~society
Omnicrola LinkMore photos from Saturday : Photos: 'No Kings' protests across the countryMore photos from Saturday :
Photos: 'No Kings' protests across the country -
Comment on This eerily accurate ‘LinkedIn Speak’ translation tool will help you sound like an instant thinkfluencer in ~tech
Omnicrola LinkThis is eerily good (and I mean bad). Next I need a plugin that scans work email and rates how "thinkfluencer" it is on a scale of 1-10 and blocks anything that rates too highly. A lot of our...This is eerily good (and I mean bad). Next I need a plugin that scans work email and rates how "thinkfluencer" it is on a scale of 1-10 and blocks anything that rates too highly. A lot of our internal communications would get auto-trashed.
-
Comment on A.T.L.A.S: outperform Claude Sonnet with a 14B local model and RTX 5060 Ti in ~tech
Omnicrola Link ParentI've considered doing this (loading a large model into a cloud host) but hadn't found a project that warranted it yet, so haven't looked at provider pricing or features. Who did you end up using?I've considered doing this (loading a large model into a cloud host) but hadn't found a project that warranted it yet, so haven't looked at provider pricing or features.
Who did you end up using?
-
Comment on Google’s TurboQuant AI-compression algorithm can reduce LLM memory usage by 6x in ~tech
Omnicrola Link ParentWell done xDApplies a slight penalty to repeated tokens to hopefully discourage idiot doom spirals. Applies a slight penalty to repeated tokens to hopefully disc^C
Well done xD
-
Comment on A.T.L.A.S: outperform Claude Sonnet with a 14B local model and RTX 5060 Ti in ~tech
Omnicrola Link ParentIf I follow along with what they're describing, then I'm unclear on how easily this is generalizable. If the smaller Cost Field network needs to be trained on what a good solution looks like,...If I follow along with what they're describing, then I'm unclear on how easily this is generalizable. If the smaller Cost Field network needs to be trained on what a good solution looks like, doesn't that make it limited to a fairly narrow slice of problems that it was trained on? Or am I getting tripped up by the simplified explanation?
-
Comment on Everyone but US President Donald Trump understands what he’s done - allied leaders know that any positive gesture they make will count for nothing in ~society
Omnicrola Link ParentI think both can be true. Meaning that they are both doing what they're doing on purpose, and think the consequences they've been warned about are either exaggerated, are worth it to achieve their...I think both can be true. Meaning that they are both doing what they're doing on purpose, and think the consequences they've been warned about are either exaggerated, are worth it to achieve their goal, or just wrong. And they lack the humility and self awareness to truly consider if they might be wrong (eg, dumb).
-
Comment on Everyone but US President Donald Trump understands what he’s done - allied leaders know that any positive gesture they make will count for nothing in ~society
Omnicrola Link ParentHad to dig a bit to find that one, all I could remember was the much more generic "no villain believes themselves to be the villain".It may be confidently asserted that no man chooses evil, because it is evil; he only mistakes it for happiness, the good he seeks. And the desire of rectifying these mistakes, is the noble ambition of an enlightened understanding, the impulse of feelings that Philosophy invigorates.
-- Mary WollstonecraftHad to dig a bit to find that one, all I could remember was the much more generic "no villain believes themselves to be the villain".
-
Comment on Weekly US politics news and updates thread - week of March 23 in ~society
Omnicrola Link ParentI keep forgetting how much this narcissistic child likes seeing his name on things.I keep forgetting how much this narcissistic child likes seeing his name on things.
-
Comment on Meta and YouTube found liable in landmark social media addiction trial in ~health.mental
Omnicrola Link ParentIt does feel like a double-edge sword. On the one hand great, social media companies should be regulated and legislated into behaving in a less exploitative way. (For everyone, not just minors,...It does feel like a double-edge sword. On the one hand great, social media companies should be regulated and legislated into behaving in a less exploitative way. (For everyone, not just minors, ideally)
And on the other hand yea, this is another weapon for the groups lobbying for universal age verification.
-
Comment on Bernie vs. Claude in ~society
Omnicrola Link ParentI appreciate that, I wanted to do the same but couldn't this morning. It's gotten me thinking though, for the wider audience, how do you convince them that such a conversation is "real"? When the...I appreciate that, I wanted to do the same but couldn't this morning.
It's gotten me thinking though, for the wider audience, how do you convince them that such a conversation is "real"? When the AI in the video itself is telling you about use cases where AI can be used to spread disinformation and propaganda, this seems particularly difficult.
-
Comment on Bernie vs. Claude in ~society
Omnicrola LinkThis is an interesting piece, and I love Bernie for doing things to bring attention to AI's many faceted problems. However there are a few things I don't like about it (as a format, not with any...This is an interesting piece, and I love Bernie for doing things to bring attention to AI's many faceted problems. However there are a few things I don't like about it (as a format, not with any of the content per se).
Firstly - even with no malice or ill intent on anyone's part, AI in default mode is going to agree with the person they're talking too. So that isn't really "Claude's opinion", Claude isn't a person with specific beliefs.
Secondly - even though we see the text on screen, drawing attention to the fact that this is a conversation happened as it is being shown, it is trivially easy to have put in a setup prompt at the very beginning to make sure the conversation goes exactly as they want it to.
-
Comment on When you were first getting your driver's license, what were you afraid of? in ~transport
Omnicrola LinkI have never been afraid of driving overall, have been scared sometimes when something unexpected happens. It'll happen, life is sometimes scary. Number one piece of advice I give every new driver...I have never been afraid of driving overall, have been scared sometimes when something unexpected happens. It'll happen, life is sometimes scary.
Number one piece of advice I give every new driver I have a chance to talk to :
Be predictable
Like with walking down a semicrowded sidewalk, there are unwritten rules and conventions about how people navigate around each other without talking. Sometimes it's subtle, sometimes it's eye contact, or a wave or a nod. Most of us learn them just by doing.Driving is the same way, however the bonus is there *are explicit rules that everyone follows (mostly) so that you can predict what other people are going to do, and they can predict what you're going to do. When in doubt, look at what everyone else is doing, and copy it. If you have an older friend or family member who you consider a good driver (not everyone is!!) ask them to talk to you out loud while they're driving around about what they're looking at and deciding.
Other good rules of thumb:
- Whatever the speed limit is, if you are passing more people than are passing you, you are going too fast. If the reverse is true, you are going too slow. Going slow is ok if you're being cautious, but make that decision on purpose.
- If you are overwhelmed, pull over and park. The road will still be there when you're ready to continue. Panicked brains make rash decisions.
- Don't tailgate. It's one of the chief causes of accidents. Your car is physically incapable of stopping fast enough if the distance between you and the person in front of you isn't long enough. Don't be arrogant and believe that you can bend the laws of physics.
-
Comment on I made a word association game - Noun Sense in ~games
Omnicrola LinkI have a daily standup at work where we often play word games as a group to try and get our brains working in the morning. Things like NYT Connections, Bracket City, Costcodle, etc. So this is...I have a daily standup at work where we often play word games as a group to try and get our brains working in the morning. Things like NYT Connections, Bracket City, Costcodle, etc. So this is right up our alley.
We played today's version and really liked it, we scored about 700/1000 which we were pretty proud of! The feedback from everyone was that they would enjoy adding it to our rotation of morning games. Nice job!
I had no idea this island existed, and this was a really interesting media-rich look at life on the island. In particular the story of how in 1961 the entire population of 264 people had to evacuate due to the volcano they live on erupting. [1] Was also interesting to read how life on the island has changed and evolved as contact with the outside world has increased.
[1] - Wikipedia - Tristan da Canha - Eruption of Queen Mary's Peak