x y

[x y @ydnar]

genuinely mind-blowing that we can run these models locally

i'm getting 30-35t/s with the 27b and over 100t/s for the 35b-a3b

works great with opencode using a custom chat template

[x y @ydnar]

@airalosupport is it possible to use your service if i dl the apk and install on grapheneos?

[x y @ydnar]

"the saddest aspect of life right now is that science gathers knowledge faster than society gathers wisdom."

-- isaac asimov

[x y @ydnar]

created an automatic transcription bot. is there a better way to do this?

- wireguard into home intranet from phone
- transfer audio recording to a folder on home server using the termius app
- my bot watches the folder and then processes the audio (whisperx w/ diarization)

[x y @ydnar]

https://t.co/jBCMsYIH6K

i get that it's not a direct comparison since the stargate ai project is mostly private industry money, but that is an insane amount compared to historical us projects!

[x y @ydnar]

gm 🌄

[x y @ydnar]

@AirhubEsim Absolutely terrible service. Inconvenienced me and my family when traveling to Japan and the support staff refused to admit the service wasn't working properly after asking me to change APN 3 times. Will definitely warn others not to use.

[x y @ydnar]

I got YouTube algo'd back to '96 https://t.co/S2rEdhVQ6j

[Msty AI @msty_ai]

@ydnar Thanks! There is! Not just sync but more. Check out this: https://t.co/hgJjGBkRNK

[x y @ydnar]

@msty_app 🤯 amazing. thank you so much!

[x y @ydnar]

@msty_app love the app! is there a way to sync data between devices? i'd love to keep folder structure, queries, api keys synced between desktop/laptop.

[x y @ydnar]

@adrian_horning_ crunchbase

[x y @ydnar]

@0xluffy this is literally me

[x y @ydnar]

@levelsio I'm so sorry you got stuck with those, but loved your episode with Lex.

[x y @ydnar]

It's also conversational. Meaning, you can continue to ask the program questions about what was discussed in the videos, and save the the conversation to a markdown file.

[x y @ydnar]

LLM API Pricing -- sonnet 3.5 for coding, groq llama3.1-70b-versatile for general queries, gpt-4o web (free) to... hold my notes?

[x y @ydnar]

📉 31 miles this week🏃‍♀️💨

[x y @ydnar]

@graycrawford Thanks for letting us "surf" with you. Incredible work!

[x y @ydnar]

Limited to 12gb VRAM on my GPU so my daily driver is dolphin-2.6-mistral-7B-GGUF, but it's actually pretty good! Interested to see what can be run on the same hardware a year from now.

Running this locally + using Mistral-Medium online and I don't feel like I need GPT4 at all.