James Prudente

Co-Founder Spatiality LLC.

[James Prudente @jamesprudente]

@JagersbergKnut gpt3 is 175b right?

[James Prudente @jamesprudente]

@hrishioa whisper.cpp include some command detection examples https://t.co/qGuRVTYp9X

[James Prudente @jamesprudente]

@agihippo chain of deep thoughts (by jack handey)

[James Prudente @jamesprudente]

@browserdotsys no worries, there's lots of time, just make sure you're there while they're receptive.

[James Prudente @jamesprudente]

@MarkovMagnifico isn't this zero shot?

[James Prudente @jamesprudente]

@browserdotsys its called FUD

[James Prudente @jamesprudente]

@multimodalart amusing that shanghai world financial center building is both in new york and shanghai.

[James Prudente @jamesprudente]

oops, that 4080, still very cool.

[James Prudente @jamesprudente]

384 tokens/sec on 3080 with 32 clients!

[Georgi Gerganov @ggerganov]

Another example of parallel decoding + continuous batching

This time we are using CUDA with an RTX 4080
Due to memory limits, we are running a smaller model: LLaMA v2 7B Q8_0

[James Prudente @jamesprudente]

dall-e-3 claims to follow sophisticated instructions better, looks pretty good!

https://t.co/L62fro73Sv

[Teknium 🪽 @Teknium]

Dale 3 is coming and it will be built in to chatgpt

[James Prudente @jamesprudente]

@ZstormVR ah, yes, that boomer headset that's *checks notes* 11 months old.

[James Prudente @jamesprudente]

@chrisprucha gemini needs to scare openai, otherwise they will continue to slow roll gpt-4/5

[Suhail @Suhail]

You can do really fun interior design concepts with AI. Here's a tutorial where we show you how to do it:

[James Prudente @jamesprudente]

@valvesoftware reconnected my cable, and its better, still some sparkles, no not a full on RGB blizzard.

[James Prudente @jamesprudente]

my index display is starting to sparkle and the audio is popping, please @valvesoftware release the deckard before I have to buy another $130 cable

[James Prudente @jamesprudente]

Looking forward to the leadership changes

[Unity @unity]

We have heard you. We apologize for the confusion and angst the runtime fee policy we announced on Tuesday caused. We are listening, talking to our team members, community, customers, and partners, and will be making changes to the policy. We will share an update in a couple of days. Thank you for your honest and critical feedback.

[Radamés Ajna @radamar]

Try the Meta Segment Anything Model (SAM) right in your browser! It performs both embedding and point prompting inferences, all powered by the Rust Candle Framework compiled to Wasm

Space: https://t.co/YxYgpvttxt
Source: https://t.co/1PlzklUjC0

#webml

[James Prudente @jamesprudente]

@CezaryXR I have QuestPro and love it for cordless/MR. But still use Index as daily driver for PCVR due to its seamlessness. If Deckard had that seamless experience with no cords + MR would be nearly perfect device.

[James Prudente @jamesprudente]

@yacineMTB always loved rails, but it's full of monkeypatching. makes it hard to understand what's going on sometimes.

[James Prudente @jamesprudente]

60% of the time it works every time.

[anton @abacaj]

What is happening here?