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 podcast: https://www.linkedin.com/company/life-with-ai

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-clearly-explained-ed008044e07e#97f4

Amazing YouTube videos explaining diffusion: https://www.youtube.com/watch?v=1CIpzeNxIhU

Explaining stable diffusion with code: https://www.youtube.com/watch?v=-lz30by8-sU

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 podcast: https://www.linkedin.com/company/life-with-ai

Blog post explaining diffusion: https://www.assemblyai.com/blog/diffusion-models-for-machine-learning-introduction/

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: https://github.com/filipelauar/projects/blob/main/similarities_and_contrastive_loss.ipynb

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 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.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 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.linkedin.com/company/life-with-ai

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 copilot assistants using RAG.

In the episode we talk both about technical and product aspects, like similarity metrics, how many documents to use, how to show the answer to the user, how to metrify the quality of the answers...

Linkedin do Joao: https://www.linkedin.com/in/joaobatista-cordeironeto/

Linkedin da Stackspot AI: https://www.linkedin.com/company/stackspot/

Instagram do podcast: https://www.instagram.com/podcast.lifewithai

Linkedin do podcast: https://www.linkedin.com/company/life-with-ai

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/podcast.lifewithai

Linkedin of the podcast: https://www.linkedin.com/company/life-with-ai

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 which layer of the LLM.

Paper MoD: ⁠https://arxiv.org/pdf/2404.02258.pdf⁠

Paper layer pruning: ⁠https://arxiv.org/pdf/2403.17887v1.pdf⁠

Instagram of the podcast: https://www.instagram.com/podcast.lifewithai

Linkedin of the podcast: https://www.linkedin.com/company/life-with-ai