Bruno Henrique

Fascinado por tecnologia e inovação

[Vaibhav (VB) Srivastav @reach_vb]

Microsoft just released VibeVoice - 1.5B SoTA Text to Speech model - MIT Licensed 🔥

> It can generate up 90 minutes of audio
> Supports simultaneous generation of > 4 speakers
> Streaming and larger 7B model in-coming
> Capable of cross-lingual and singing synthesis

Love the expressiveness and the emotion control on the model! Kudos to Microsoft 🤗

[Bruno Henrique @Brunot3ch]

🎙 VibeVoice da Microsoft: TTS open-source revolucionário! 🗣 Até 4 vozes, 90 min de fala, super expressivo com LLM Qwen2.5-1.5B. 🔓 Ideal para podcasts e dublagens. 💡 Como você usaria essa tech? #VibeVoice #TTS #OpenSourc
#VibeVoice #TTS #OpenSource

[Bruno Henrique @Brunot3ch]

Grok 2 open source 🤩
@elonmusk @xai

[Bruno Henrique @Brunot3ch]

🚀 Grok 2.5 is now open-source! A powerful and accessible AI, ready to drive innovation. Finally, a true @OpenAI AI for everyone! 🙌 #Grok25 #OpenSource #AI @xai
Vlw @elonmusk

[Elon Musk @elonmusk]

The @xAI Grok 2.5 model, which was our best model last year, is now open source.

Grok 3 will be made open source in about 6 months.

https://t.co/TXM0wyJKOh

[Bruno Henrique @Brunot3ch]

@elonmusk @xai \o/
Vlw @elonmusk

[Elon Musk @elonmusk]

The @xAI Grok 2.5 model, which was our best model last year, is now open source.

Grok 3 will be made open source in about 6 months.

https://t.co/TXM0wyJKOh

[Bruno Henrique @Brunot3ch]

🚀 Chatbot Developer seeking new opportunities! Skilled in conversational AI, Gen AI, and API integration. Delivering high-quality, innovative projects. 📩 DMs open for gigs! #JobSearch #Chatbots #JobSearch, #Chatbots, #AI, #DevJobs

[Bruno Henrique @Brunot3ch]

🚀 Desenvolvedor de chatbots em busca de novas oportunidades! Expertise em IA conversacional, NLP e integração de APIs. Projetos entregues com qualidade e inovação. 📩 DM aberta para propostas! #VagaDev

[AK @_akhaliq]

state space models exist, they empirically underperform Transformers in pretraining efficiency and downstream task accuracy. We introduce Megalodon, a neural architecture for efficient sequence modeling with unlimited context length. Megalodon inherits the architecture of Mega

[AK @_akhaliq]

Google presents Mixture-of-Depths

Dynamically allocating compute in transformer-based language models

Transformer-based language models spread FLOPs uniformly across input sequences. In this work we demonstrate that transformers can instead learn to dynamically allocate

[Bruno Henrique @Brunot3ch]

AutoWebGLM: inicialize e reforce um agente de navegação na Web baseado em modelo de linguagem grande

https://t.co/CxQi7Xq2Pk
@thukeg

[AI KATANA @ai_katana]

In a first, AI was able to reconstruct images from brain activity with over 75% accuracy.

Japanese researchers have achieved a significant breakthrough in AI-generated imaging, reaching a record 75% accuracy in reconstructing images from brain activity.

This advancement marks a considerable improvement over previous methods which only achieved 50.4% accuracy. The process involves recording brain activity while subjects view images and later recalling these images.

Utilizing a neural signal translator and generative AI, researchers were able to reconstruct these images with high accuracy. This technology opens new possibilities in understanding the human mind and could lead to novel forms of non-verbal communication.

[Bruno Henrique @Brunot3ch]

💯

[AK @_akhaliq]

Here is my selection of papers for today (4 Dec):

paper pages: https://t.co/QzUWybzXRg

FSGS: Real-Time Few-shot View Synthesis using Gaussian Splatting

Scaffold-GS: Structured 3D Gaussians for View-Adaptive Rendering

PyNeRF: Pyramidal Neural Radiance Fields

Mamba: Linear-Time Sequence Modeling with Selective State Spaces

MoMask: Generative Masked Modeling of 3D Human Motions

Text-Guided 3D Face Synthesis -- From Generation to Editing

DREAM: Diffusion Rectification and Estimation-Adaptive Models

X-Dreamer: Creating High-quality 3D Content by Bridging the Domain Gap Between Text-to-2D and Text-to-3D Generation

VideoBooth: Diffusion-based Video Generation with Image Prompts

StyleCrafter: Enhancing Stylized Text-to-Video Generation with Style Adapter

HiFi Tuner: High-Fidelity Subject-Driven Fine-Tuning for Diffusion Models

Beyond ChatBots: ExploreLLM for Structured Thoughts and Personalized Model Responses

Merlin:Empowering Multimodal LLMs with Foresight Minds

Instruction-tuning Aligns LLMs to the Human Brain

Dolphins: Multimodal Language Model for Driving
Towards Accurate Differential Diagnosis with Large Language Models

SeaLLMs -- Large Language Models for Southeast Asia

GraphDreamer: Compositional 3D Scene Synthesis from Scene Graphs

[Bruno Henrique @Brunot3ch]

Cerca de 20.000 dos cristais identificados experimentalmente no banco de dados ICSD são computacionalmente estáveis. Abordagens computacionais extraídas do Materials Project, do Open Quantum Materials Database e do banco de dados WBM aumentaram esse número para 48.000 cristais estáveis. O GNoME expande o número de materiais estáveis ​​conhecidos pela humanidade para 421.000.

[Google DeepMind @GoogleDeepMind]

Introducing GNoME: an AI tool that helped discover 2.2 million new crystals. 💎

Crystals are found in everything from the chips powering our phones to solar cells creating clean energy.

The model also better predicts the stability of new materials. 🧵 https://t.co/O3YdnVcJt1

[Bruno Henrique @Brunot3ch]

Cerca de 20.000 dos cristais identificados experimentalmente no banco de dados ICSD são computacionalmente estáveis. Abordagens computacionais extraídas do Materials Project, do Open Quantum Materials Database e do banco de dados WBM aumentaram esse número para 48.000 cristais estáveis. O GNoME expande o número de materiais estáveis ​​conhecidos pela humanidade para 421.000.

[Google DeepMind @GoogleDeepMind]

Introducing GNoME: an AI tool that helped discover 2.2 million new crystals. 💎

Crystals are found in everything from the chips powering our phones to solar cells creating clean energy.

The model also better predicts the stability of new materials. 🧵 https://t.co/O3YdnVcJt1

[Bruno Henrique @Brunot3ch]

😍

[Google DeepMind @GoogleDeepMind]

Introducing GNoME: an AI tool that helped discover 2.2 million new crystals. 💎

Crystals are found in everything from the chips powering our phones to solar cells creating clean energy.

The model also better predicts the stability of new materials. 🧵 https://t.co/O3YdnVcJt1

[Bruno Henrique @Brunot3ch]

@ylecun @NandoDF Congratulations on the excellent work! Now, to enhance it even further, it would be amazing to acquire more specific data for Brazilian Portuguese. Let's enrich this experience even more

[Bruno Henrique @Brunot3ch]

@yoachlacombe @metaai Congratulations on the excellent work! Now, to enhance it even further, it would be amazing to acquire more specific data for Brazilian Portuguese. Let's enrich this experience even more

[Bruno Henrique @Brunot3ch]

Notus 7B ready for action and fluent in Brazilian Portuguese! 🚀🇧🇷 Configured and responding with precision. #LLMs #opensource #Notus7B #BrazilianPortuguese

[Bruno Henrique @Brunot3ch]

Model summary
Developed by: Argilla (based on HuggingFace H4 and MistralAI previous efforts and amazing work)
Shared by: Argilla
Model type: GPT-like 7B model DPO fine-tuned
Language(s) (NLP): Mainly English
License: MIT (same as Zephyr 7B-beta)
Finetuned from model: alignment-handbook/zephyr-7b-sft-full
Repository: https://t.co/Imrm6ZjJvr
Paper: N/A
Play with Notus on HuggingChat: https://t.co/pXhgcXCTRk
@argilla_io @huggingface

[Bruno Henrique @Brunot3ch]

Model summary
Developed by: Argilla (based on HuggingFace H4 and MistralAI previous efforts and amazing work)
Shared by: Argilla
Model type: GPT-like 7B model DPO fine-tuned
Language(s) (NLP): Mainly English
License: MIT (same as Zephyr 7B-beta)
Finetuned from model: alignment-handbook/zephyr-7b-sft-full
Repository: https://t.co/Imrm6ZjJvr
Paper: N/A
Play with Notus on HuggingChat: https://t.co/pXhgcXCTRk
@argilla_io @huggingface