Como é estudar na 1ª graduação de IA no Brasil, com nota de corte maior que Medicina. guiadoestudante.abril.com.br…

Esse padawan foi o mais original até agora, adoramos 😂😂 Valeu, Caique!! Estamos felizes demais de ajudar na sua jornada!! 🔥🔥 Agora só falta trocar pra caiqueonpython! 😂

Demanda por FastAPI bombando e tem curso DE GRAÇA do mestre Duno! Tira umas horinhas de Netflix que vale muito a pena, padawan. Dunossauro é atestado de excelência

Things you can do with AI #1: Slow motion waves Video prompt: show the wave motion flow in slow motion

Calma, gente, daqui a pouco chove trampo de Vibe Debugging pra todos os projetos de Vibe Coding 😂

Very practical guide on forecasting by the analytics team at @Meta medium.com/@AnalyticsAtMeta/…

Hoje no @devsfronteiras falei com o @bernardolago, que saiu do negócio de família para se tornar Cientista de Dados em Portugal, de onde também toca o podcast @letsdataAI, um dos mais conhecidos no Brasil dessa área. 🇵🇹 Vai lá escutar! 😎 devsemfronteiras.tech/cienti…

Nessa aventura o @fabriciocarraro falou com o @bernardolago, que saiu do negócio de família para se tornar Cientista de Dados em Portugal, de onde também toca o podcast @letsdataAI, um dos mais conhecidos no Brasil dessa área. 🇵🇹 Vai lá escutar! 😎 devsemfronteiras.tech/cienti…

Artigo publicado no nosso Medium sobre como iniciar com Git e Github! medium.com/lets-data/desvend…

Artigo publicado no nosso Medium sobre expressões regulares (Regex), uma ferramenta poderosa para encontrar padrões em texto medium.com/lets-data/desmist…

Conformal prediction takes Brazil 🇧🇷 and Portugal 🇵🇹 by storm. ”Conformal Prediction seminar” in Portuguese. youtube.com/watch?v=YJDCwyCa… #conformalprediction

.@OctoML is now serving the lowest-cost Mixtral MoE inference I've seen. Input: $0.20/million tokens Output: $0.50/million tokens octoml.ai/blog/mixtral-8x7b-…

OpenAI just came out with their Prompt Engineering guide: platform.openai.com/docs/gui…

🤯🤯

Removed, enjoy !

Let's begin #NeurIPS2023

Thrilled to share #Lyria, the world's most sophisticated AI music generation system. From just a text prompt Lyria produces compelling music & vocals. Also: building new Music AI tools for artists to amplify creativity in partnership w/YT & music industry deepmind.google/discover/blo…

Um papo com muitas dicas de como começaríamos hoje numa carreira de dados. Não só com nossa experiência como profissionais de dados mas também o que temos vivenciado com nossos alunos e alunas no Let's Data. Espero que gostem!! open.spotify.com/episode/2ww…

LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70B paper page: huggingface.co/papers/2310.2… AI developers often apply safety alignment procedures to prevent the misuse of their AI systems. For example, before Meta released Llama 2-Chat, a collection of instruction fine-tuned large language models, they invested heavily in safety training, incorporating extensive red-teaming and reinforcement learning from human feedback. However, it remains unclear how well safety training guards against model misuse when attackers have access to model weights. We explore the robustness of safety training in language models by subversively fine-tuning the public weights of Llama 2-Chat. We employ low-rank adaptation (LoRA) as an efficient fine-tuning method. With a budget of less than $200 per model and using only one GPU, we successfully undo the safety training of Llama 2-Chat models of sizes 7B, 13B, and 70B. Specifically, our fine-tuning technique significantly reduces the rate at which the model refuses to follow harmful instructions. We achieve a refusal rate below 1% for our 70B Llama 2-Chat model on two refusal benchmarks. Our fine-tuning method retains general performance, which we validate by comparing our fine-tuned models against Llama 2-Chat across two benchmarks. Additionally, we present a selection of harmful outputs produced by our models. While there is considerable uncertainty about the scope of risks from current models, it is likely that future models will have significantly more dangerous capabilities, including the ability to hack into critical infrastructure, create dangerous bio-weapons, or autonomously replicate and adapt to new environments. We show that subversive fine-tuning is practical and effective, and hence argue that evaluating risks from fine-tuning should be a core part of risk assessments for releasing model weights.

Apparently it is ten years today since the first release of seaborn on PyPI 🎂