Baihe

@rockysun

102 Following

66 Followers

Baihe pfp

0 reply

0 recast

0 reaction

agoston nagy pfp

it's where deep learning models become truly meaningful: AlphaFold3 is now open source (version 2 was honoured with a Nobel prize) https://www.nature.com/articles/d41586-024-03708-4

4 replies

7 recasts

41 reactions

Baihe pfp

0 reply

0 recast

0 reaction

Cassie Heart pfp

This week in building /quilibrium (Publishing early, thunderstorm inbound and will likely lose power tonight) (On protocol) - Incorporated domain separation of prover outputs (allows optionality in the event we have a graceful recursion proof that can be applied atop the MPC-in-the-head proofs) - Dynamic fee market testing has held strong from last week, so we published an overview (and small ELI5): https://paragraph.xyz/@quilibrium.com/dynamic-fee-markets (Off protocol) - We're reorganizing the documentation section of the site to be an open source repo detached from the main site and on a separate subdomain, will be less flashy, but also removes the heavy branded assets and bridge interface stuff so people can contribute to it instead of there needing to be so many unofficial guides.

7 replies

87 recasts

244 reactions

Baihe pfp

0 reply

0 recast

0 reaction

Baihe pfp

0 reply

0 recast

0 reaction

Baihe pfp

0 reply

0 recast

0 reaction

Gabriel Haines pfp

PRESENTING THE GABRIEL HAINES PAID GROUP Support my content via @hypersubxyz! Join the gated chat room which includes behind the scenes footage, sneak previews and personalized videos LFG!!!! https://www.hypersub.xyz/s/gabriel-haines-kftf6nnjlo1s

1 reply

24 recasts

113 reactions

Shira Stember pfp

How I feel about getting boosted today 🤩😬👩🏻‍💻💡🥴😎

2 replies

7 recasts

69 reactions

Rocky pfp

Minted Owly Bebop

0 reply

1 recast

0 reaction

Baihe pfp

farcaster is missing news and the memes are too niche

0 reply

0 recast

0 reaction

antimo 🎩 pfp

https://warpcast.com/rish/0x8a60a23d

0 reply

1 recast

55 reactions

Salvino Armati ↑ pfp

Salvino Armati ↑

on the one hand i’m glad ppl are using my app on the other my job is just responding to support issues now

19 replies

83 recasts

323 reactions

kepano pfp

I'm reading the new Kubrick biography "An Odyssey" and this quote made me laugh

6 replies

4 recasts

98 reactions

bibalos pfp

@better-then-you

Collect Farcaster: Horse

0 reply

1 recast

1 reaction

Ryan Sean Adams (rsa.eth) pfp

Ryan Sean Adams (rsa.eth)

I don't have much to contribute to this recent DOT drama just one observation. Over the years i've had a front-row seat on how crypto leaders engage with their "out-tribe" and i've noticed something. The best leaders engage their critics instead of retreating. @vitalik.eth went on "what bitcoin did" with Samson Mow - full enemy territory - Anatoly of Solana has always been willing to come on Bankless even for debates - these leaders win hearts and minds by engaging with their critics. I don't think the DOT leaders have done enough out-tribe engagement over the years so i'm not surprised at the results. Bullish on crypto projects with leaders who engage their out-tribe, bearish those that retreat.

6 replies

4 recasts

176 reactions

𝚐𝔪𝟾𝚡𝚡𝟾 pfp

𝚐𝔪𝟾𝚡𝚡𝟾

i tried the demo and imo it still has a long way to go, now if they in fact release code, model, and paper i believe the community will improve on it and it will be a much better assistant. … some notes: - 1. 7B Multimodal LM - Moshi already runs on apple laptops! (can run on laptop / on consumer GPUs) big w! - local: no data leaving your computer, no internet access. - open source! technical report and open model releases🤞 - latency + - trained by 8+ people in 4 months - used a heavy amount of synthetic data - didn’t get the emotion - missed simple scheduling prompts - no multilingual - needs fine-tuning try demo ↓ https://moshi.chat/?queue_id=talktomoshi

1 reply

2 recasts

4 reactions

𝚐𝔪𝟾𝚡𝚡𝟾 pfp

𝚐𝔪𝟾𝚡𝚡𝟾

Moshi 🔥 - 1. 7b multimodal LM - will be released as open source!! -achieves 160ms latency🤌✨ - trained on Scaleway cluster of 1000 H100 GPUs - expresses emotions and understands accents, like a “french accent.” - handles audio generation and listening simultaneously. - processes thoughts textually during speech. - uses dual audio streams for simultaneous listening and speaking. - jointly pre-trained on text and audio. - utilizes synthetic text from the 7b LLM Helium and fine-tuned on 100k TTS-converted “oral-style” conversations. - voice learned from TTS-generated data. - achieves 200ms end-to-end latency. - includes a smaller version for macbooks or consumer GPUs. - implements watermarking to identify AI-generated audio (in progress).

1 reply

1 recast

21 reactions

Giuliano Giacaglia pfp

Giuliano Giacaglia

The destruction caused by the fiat system is very clear. You just have to look around

1 reply

17 recasts

96 reactions

Michael Pfister pfp

Michael Pfister

surf report: how was your last session?

7 replies

1 recast

32 reactions