TomGarden

My understanding from reading news sources recently is that a vast amount of posttraining and even posttraining harnesses in many cases are LLM-overseen now in the frenzy of the AI race. Less and less human oversight in the part that does the rewards training. What could go wrong...

I'm still excited about this, Fable 5.1 and all the great open source models, but some very religious undertones have been entering the AI sphere

I'm sure this will be a great model.

Personally, I'm far away from screaming 'AGI is here!' from the rooftops, until jaggedness and silly mistakes disappear at the very least . (what is going on with that Mario Kart game...) So many benchmarks are 'best of x tries' or using very specific harnesses. AGI would not need a babysitter.

Honestly, even being able to do simple tasks like summarization or basic knowledge work without the constant paranoia of unforseen failure would be remarkable and useful.

not a chance its gonna be realtime lol. It's fun marketing though, I'll allow it

They're all complimentary. I've had double subs at the top tiers for a while now. Best of both worlds

I don't think we'll be able to meet, and that's ok since the definition isn't universal. I align more with Demis Hasabis' views on this

There's so much that the term includes that isn't even feasible with an LLM

Only Sonnet working over here, but it feels really fast and good. I could get used to this... Really wanted to take Fable 5.1 for a spin though

Any source on the 90% number? Last I saw (recently) one of the big players just announced they were planning to do 50% AI going forward

First time a topical LLM joke made me laugh IRL. Stellar work

Interesting idea! I wonder if, after some iteration, a variation on this could help curb some of the LLM spaghetti mess

Any tips on the best approach at running this at an M4 Max 128GB? Token throughput was a bit slow with the last 27B one (MLX), ended up using the A3B variant but if I could get this one to reasonable speed I'd much prefer it.

Really excited to see what people do with this. 3.7 27B was probably the best compromise between size and intelligence to run on consumer hardware

I run the same, if you want try a simple filter post transcription to remove them, and while you're at it add some simple word replacements like 'cloud MD' to 'CLAUDE.MD'

Anyone else feel like they're going insane sometimes?

It's so simple, everyone on HN knows this, but I want to scream at every non-technical person "these tools are not reliable! You can't trust them!"

They are so useful but the failure modes need to be hammered into everyone at the very start of each onboarding.

It would be totally rational as a layperson to just ignore AI and just check in every few months to ask "is it reliable yet".

The big thing for me is the TTS, custom UI and persistent background mode! ie it switches turns automatically etc, no need to touch screen or keep screen on.

The STT on Gboard is very solid, so if that covers your use case you're good!

The biggest difference is that mine is audio-first - it reads everything out over Android tts by default, and runs a computer-side parakeet + Silero VAD server for transcription (My eyes struggle with small screens, though I use it text only occasionally). It's like a voice assistant but with Claude Code. I also made a custom GUI the with shortcuts and stuff, making saying "end conversation" actually end the conversation etc.

Maybe something similar can be done with tmux still, I'm definitely going to explore it

This sucks. I use Claude -p over tailscale to code over voice when I’m on the go for accessibility reasons, and most of the time I do the same while at the computer. Running through $200 in API pricing takes no time. Oh well, time to switch providers I guess.

Claude voice mode has come a long way! I'd say it's smarter than CGPT AVM last time i tried it.

But personally I've settled on just speaking to the slower models over a custom tts app, I find it being instant was not actually that important, and in the silence I find myself marinating in the discussion more anyway

The kissing noise analogy is spot on! Made me smile

I a European who spent the last decade in America and I'm not sure I'd call Americans privileged compared to Europe. With money being the one means you have to be treated well in society, comparing it to Europe, America feels like the hunger games. Want healthcare (ie surviving)? Healthy food? To own your house? Welcome to the games

I used to think like what you describe, but I've fallen on the side of "art is just more emotionally resonant human communication". And most of the time human communication with more effort and thought behind it. AI art falls short on both being human and, on average, having more effort or thought behind it than your general interaction at the supermarket.

I will say, it can be emotionally resonant though - but it's a borrowed property from the perception of human communication and effort that made the art the models were trained on.

Had better, faster results by changing to medium effort. Weird, but the xhigh default chugged forever just to come back with poorer solutions than 4.6 on medium

Asked Opus 4.7 to extend an existing system today. After thorough exploration and a long back and forth on details it came up with a plan. Then proceeded to build a fully parallel, incompatible system from scratch with the changes I wanted but everything else incompatible and full of placeholders

The performance issues are crazy. Thanks for sharing this

It seems unclear if this covers all Claude -p uses or just the ones they can identify as misuse/third party. Did they speak on this anywhere?

I use Claude -p for a lot if not most of my coding workflows

Nah, there's apparently a few caching bugs, one --resume and some noisy tool use. I have a little app that monitors and resets the context window at 70% usage based on 200k tokens and I'm about to run out of weekly allowance after just a couple days. Never happened before

For many it's not about the children. For many it is.

I haven't made my mind up on this topic, but Jesus, the comments here strawmanning everyone who supports this kind of thing as disingenuous or worse... Wow.

I'm not sure how we make any corner of the internet usable within the next few years without verification given all the misinfo, bots & AI slop anyway.

This makes no sense, their value in the marketplace is in usage and inflated promise, not actual revenues

It will be a begrudged but normalized compromise of modern life, like GMO, Auto-Tune and foreign child labour. Which of those it will be perceived as the closest to is hard to say right now