New Claude 3.5 Haiku released! LFG
It beats the one Sonnet! What a big win for Anthropic!
@kimmonismus
-
Claude 3.5 Haiku Released, Outperforms Sonnet Model
By
–
-
o1 Release Imminent with Preparations Underway
By
–
Confirmed: o1 release imminent, preparation for release were made.
-
Small Text-to-Speech Model Based on Llama Released Openly
By
–
Small text to speech on 350m parameters. Built on Llama, free and open source.
— Chubby♨️ (@kimmonismus) 4 novembre 2024
Love it https://t.co/1L9pIwKSyOSmall text to speech on 350m parameters. Built on Llama, free and open source. Love it
-
URAvatar: Photorealistic Head Avatars from Phone Scans with Real-time Relighting
By
–
URAvatar: a new approach to creating photorealistic and relightable head avatars using a phone scan with unknown illumination. The reconstructed avatars can be animated and relit in real time with the global illumination of diverse environments. pic.twitter.com/ZWv8SreJgy
— Chubby♨️ (@kimmonismus) 4 novembre 2024URAvatar: a new approach to creating photorealistic and relightable head avatars using a phone scan with unknown illumination. The reconstructed avatars can be animated and relit in real time with the global illumination of diverse environments.
-
Grok3 Model Launch: 128K Tokens, Vision Coming Soon
By
–
– new model coming (Grok3?)
– 128k token length – vision next week
– free API for testing ($25) -
xAI Offers $25 Monthly Free Grok API Credits
By
–
$25 / month of free API credits for Grok. That’s pretty nice. Obviously the have to compete with all the other AI company’s but still, xAI is doing good!
-
Polish Radio Station Replaces Journalists with AI Hosts
By
–
So it begins: This Polish radio station fired all its journalists and replaced them with AI hosts — and people are furious. "It is a dangerous precedent that hits us all," said fired journalists.
-
O1 LLM Reasoning Capabilities and Future of Human Labor
By
–
Now that it has been indicated from various sides that o1 is coming tomorrow, I am cautiously optimistic. None of us should underestimate O1. It is proof that LLMs can reason. If this is combined with agents, there is no longer any reason why human labor should be involved.
-
Hertz-dev: Open-Source Full-Duplex Conversational Audio Base Model
By
–
Hertz-dev: an open-source, first-of-its-kind base model for full-duplex conversational audio. It's an 8.5B parameter transformer trained on 20 million unique hours of high-quality audio data. it is a base model, without fine-tuning, RLHF, or instruction-following behavior pic.twitter.com/BtyFRQk0it
— Chubby♨️ (@kimmonismus) 4 novembre 2024Hertz-dev: an open-source, first-of-its-kind base model for full-duplex conversational audio. It's an 8.5B parameter transformer trained on 20 million unique hours of high-quality audio data. it is a base model, without fine-tuning, RLHF, or instruction-following behavior
-
Custom Worlds Feature: Transform Your Starting Image into AI Worlds
By
–
Feature drop: "Custom Worlds" You can now add your own starting image to Oasis and the AI will morph it into your new worldpic.twitter.com/pmKP5rjCJE
— Chubby♨️ (@kimmonismus) 4 novembre 2024Feature drop: "Custom Worlds" You can now add your own starting image to Oasis and the AI will morph it into your new world