Life with AI

In this episode I interview Mateus Costa-Ribeiro, Founder and CEO of Enter. This interview was recorded in Sequoia's HQ in San Francisco and tells a little bit of Enter's history and mission! It's an incredible story and I hope you all enjoy this podcast as most as I enjoyed recording!! Mateus' Linkedin: https://www.linkedin.com/in/mateus-costa-ribeiro-456475195/ Filipe's Linke...

Hello everyone, in this episode I talk about GraphRAG. This new RAG technique is very useful to better retrieve global information from your document or set of documents. It's also a subject that is growing a lot! GraphRAG paper: https://arxiv.org/pdf/2404.16130 LightRAG paper: https://arxiv.org/pdf/2410.05779 Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the p...

Hello guys, in this episode of the podcast I talk about on-device AI and the SmolLM blog post published by Hugging Face. I first give my thoughts on the differences between on-device and on-cluster AI with the different needs of applications we may have. Then I go through the SmolLM blog post explaining some of the details. SmolLM blog post: https://huggingface.co/blog/smollm MobileLLM paper: http...

Hey guys, in the Brazilian version of the podcast I interviewed Felipe, CEO and Founder of Clarice AI, the Brazilian Grammarly. In the episode we discussed on how they created both their corrector and their style improvement algorithms. Clarice AI website: https://clarice.ai/ Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast: https://www.linkedin.com/co...

Hey guys, in the brazilian version of the podcast I interviewed Hugo Abonizio, he is LLM engineer at Maritaca AI, the only brazilian LLM company that just released Sabia 3. Their model is specialized in Portuguese and is between gpt4o and Claude 3.5 Sonnet, while being way smaller and cheaper. In the episode I try to give an idea of our conversation in the portuguese version of the podcast. Sabia...

Hey guys, in this episode I try to explain my ideas on why Chain of Thought works. Of course nothing that I say is proved, the idea is really to give my intuitions on it! Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast: https://www.linkedin.com/company/life-with-ai

Hey guys, in this episode I talk about the new model of OpenAI, the OpenAI o1, the model that thinks and reflects before answering. In the episode I share my opinion and my thoughts after the first contact with the model. Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast: https://www.linkedin.com/company/life-with-ai

Hey guys, in this episode I talk about the different types of AI. I don't go deep into the models, I focus more on explaining the different types and which kind of problems they solve. As types of AI I talk about statistics, machine learning, reinforcement learning, deep learning and generative AI. Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast:...

Hey guys, this is the last episode of the Llama3 paper. In this episode, I talk about their discussion on the contamination analysis of the benchmarks, the vision and the speech parts of the model. I hope you enjoyed the series of episodes about the paper, I personally learned a lot! Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast: https://www.linkedi...

Hey guys, in this episode I talk about the Llama 3 paper pre-training and post-training! Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast: https://www.linkedin.com/company/life-with-ai

Hey guys, FINALLY we have Llama 3 paper, the release of the 405B model and the update of the 8B and 70B models. In this episode I give my thoughts about the paper and also an overview about it. In the next episode I will go more deeper in more details of the paper. Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast: https://www.linkedin.com/company/life-...

Hey guys, in this episode I talk about how to choose the best model for your AI application. I discuss the different tradeoffs you should take into considering before choosing which model to test and to deploy. Transformers tutorials github by Niels Rogge: https://github.com/NielsRogge/Transformers-Tutorials Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the pod...

Hey guys, in this episode of the podcast I talk about Stable Diffusion, a famous open source image generation algorithm. Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast: https://www.linkedin.com/company/life-with-ai Stable diffusion paper: ⁠https://arxiv.org/pdf/2112.10752⁠ Good Medium post about SD: https://medium.com/@steinsfu/stable-diffusion-clear...

Hey guys, in this episode I talk about diffusion models, the algorithm behind all the image generation models today, like stable diffusion, Midjourney and dall-e. In the episode I explain the diffusion process along with some other technical concepts that are important, like white noise and markovian process. Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the po...

Hey guys, in this episode I talk about 3 very important models that use contrastive learning, CLIP, SigLIP and JinaCLIP. They are image-text embedding models that allow us to for instance do image-text retrieval. CLIP paper: https://arxiv.org/pdf/2103.00020 SigLIP paper: https://arxiv.org/pdf/2303.15343 JinaCLIP paper: https://arxiv.org/pdf/2405.20204 Github of similarities and contrastive loss: h...

Hey guys, in this episode I talk about two very important technical concepts in Deep Learning, constrastive learning and cosine similarity. They are very useful when training embedding models or doing RAG. Very good blog post about contrastive losses: https://lilianweng.github.io/posts/2021-05-31-contrastive/ SimCLR paper: https://arxiv.org/abs/2002.05709 Instagram of the podcast: https://www.inst...

Hey guys, in this episode I talk about the FineWeb dataset, the best pre-training open source dataset to date. In the episode I explain how they created the dataset and I also share some results. Link to the huggingface blog: https://huggingface.co/spaces/HuggingFaceFW/blogpost-fineweb-v1 Instagram of the podcast: https://www.instagram.com/podcast.lifewithai Linkedin of the podcast: https://www.li...

Fala galera, nesse episódio eu falo com o João Batista, Technical Product Manager da Stackspot AI. No episódio a gente falou bastante sobre o uso de LLMs como um copilot usando os próprios documentos da empresa para dar as respostas usando RAG. Hey guys, in the brazilian version of the podcast I discussed with Joao from Stackspot AI. In the episode I talk about how they are developing enterprise c...

Hey guys, in this episode I talk about two papers, BitNet and 1.58 bit Transformer. These two papers from microsoft tell a new receipe to train 1 bit transformers, improve hugely the memory and energy consumption along with lower inference times. BitNet paper: https://arxiv.org/pdf/2310.11453 1.58 bit paper: https://arxiv.org/pdf/2402.17764 Instagram of the podcast: https://www.instagram.com/podca...

Extra episode about Llama 3.

Hey guys, continuing the series of episodes about PEFT, in this episode I talk about inference optimization techniques for LLMs. I talk about layer pruning, where we prune consecutive layers of the LLM without almost not losing model performance. I also talk about Mixture of Depths, a similar technique to Mixture of Experts, where we have a router that choses which tokens will be processed in whic...

Hey guys, this is the first episode in a series of episodes about PEFT, Parameter Efficient Fine Tuning. In this episode I talk about LoRA and QLoRA, two widely used methods that allowed us to fine tune LLMs way faster and in a single GPU without losing performance. Video sobre QLoRA:  ⁠https://www.youtube.com/watch?v=6l8GZDPbFn8⁠ LoRA paper:  ⁠https://arxiv.org/pdf/2106.09685.pdf⁠ QLoRA paper:  ⁠...

Hello, in this episode I talk a Retrieval Aware Fine Tuning (RAFT), a paper that proposes a new technique to use both domain specific fine-tuning and RAG to improve the retrieval capabilities of LLMs. In the episode I also talk about another paper that is called RAFT, but this time Reward rAnking Fine Tuning, which proposes a new technique to perform RLHF without the convergence problems of Reinfo...

Hello guys, in this episode I explain how we can scale the context window of an LLM to more than 1M tokens using Ring Attention. In the episode, I also discuss if RAG is dead or not based on these advancements in the context window. Paper Lost in the Middle: https://arxiv.org/pdf/2307.03172.pdf Gemini technical report: https://storage.googleapis.com/deepmind-media/gemini/gemini_v1_5_report.pdf Pap...

Hey guys, in the Brazilian version of the Podcast I interviewed Andre, he is an AI expert on IBM and we talked a lot about how to solve problems using AI. Brains website: https://brains.dev/ Andre's Linkedin: https://www.linkedin.com/in/andrefelipelopes/ Brains' Linkedin: https://www.linkedin.com/company/brains-brazilian-ai-networks/ Instagram of the podcast: https://www.instagram.com/podc...