Exploring AI Technologies and Applications

LangChain and Tenacity API

  • LangChain uses Tenacity API
  • +1 on using Tenacity API
  • LangChain has an API for the idea of kids as superheroes

Andrej Karpathy’s Neural Nets Video

  • Finished watching Andrej Karpathy’s video on Neural Nets
  • Highly recommended for anyone starting or in the field
  • Link to the video playlist: https://youtube.com/playlist?list=PLAqhIrjkxbuWI23v9cThsA9GvCAUhRvKZ

Image Inpainting and Diffusers

  • Tried Img2img models for inpainting on an online image
  • Asked for suggestions on using diffusers for inpainting
  • Link to Stable Diffusion XL beta: https://stability.ai/blog/stable-diffusion-xl-beta-available-for-api-customers-and-dreamstudio-users

ChatGPT for GitHub Repository

  • Saw a Twitter post about ChatGPT for a GitHub repository
  • Link to the demo video of Chat with GitHub: https://js.langchain.com/docs/modules/indexes/document_loaders/examples/web_loaders/github
  • Link to LangChain API for asking questions on a GitHub repository: https://js.langchain.com/docs/modules/indexes/document_loaders/examples/web_loaders/github

Reflect.Ai and Bard

  • Suggested Reflect.Ai for asking questions on a GitHub repository
  • Bard is performing well for asking questions on a GitHub repository
  • Link to Reflect.Ai: https://reflect.ai/
  • Link to Bard: https://github.com/ashishb/Bard

DreamPose AI Model

  • Discussed DreamPose AI model that can generate video from image
  • Link to DreamPose GitHub repository: https://github.com/johannakarras/DreamPose#finetune-on-sample

Generating Embeddings and Storing in VectorDBs

  • Discussed the difference between generating embeddings and storing them in VectorDBs versus providing them directly to the LLM for question-answering
  • If the dataset has >4000 tokens, it needs to be chunked and stored in VectorDBs for QA by taking relevant chunks
  • If the dataset is <4000 tokens, it can be directly asked to the LLM for QA

AktoGPT and GPT LLM in Production

  • Akto.io launched AktoGPT
  • Ankush discussed things to watch out for before deploying GPT LLM in production
  • Link to AktoGPT: https://www.akto.io/blog/aktogpt
  • Link to Ankush’s tweet: https://twitter.com/Ankush12389/status/1646779395833741313

LangChain with Azure Endpoints

  • Asked if anyone has used LangChain with Azure endpoints instead of OpenAI directly
  • Link to LangChain API for using Azure endpoints: https://python.langchain.com/en/latest/modules/models/llms/integrations/azure_openai_example.html

Latency for Commodity Network

  • Discussed the latency for 2000 Bytes over commodity network from 2009 to 2020
  • Link to the research paper: https://colin-scott.github.io/personal_website/research/interactive_latency.html

Stability Diffusion for Robotics

  • CTO of Stability AI talked about Stability diffusion for robotics
  • Link to Stability AI: https://stability.ai/

Industry Directions and Opportunities

  • Discussed the trends in the industry as a way to understand industry directions and opportunities

Google Colab Link

  • Shared a Google Colab link
  • Link to the Google Colab: https://colab.research.google.com/drive/1VezfmvAg4t1okxs7pJ0qp0pWDAaW7mlo?usp=sharing

The description and link can be mismatched because of extraction errors.

  • https://techcrunch.com/2023/04/13/with-bedrock-amazon-enters-the-generative-ai-race/ - An API for a kids as superheroes idea is mentioned, along with a recommendation to watch Andrej Karpathy’s video on Neural Nets.
  • https://youtube.com/playlist?list=PLAqhIrjkxbuWI23v9cThsA9GvCAUhRvKZ: Finished watching Andrej Karpathy’s video on Neural Nets and highly recommend it to anyone starting or in the field. Message in the same link mentions caching LLM calls and saving money.
  • The URL is https://stability.ai/blog/stable-diffusion-xl-beta-available-for-api-customers-and-dreamstudio-users and the message in the same link is related to seeking advice on GPT-4, diffusers, and YouTube videos. The context also includes a mention of a Twitter post about ‘ChatGPT for a github repository’.
  • https://js.langchain.com/docs/modules/indexes/document_loaders/examples/web_loaders/github - Demo video of Chat with GitHub, mentioned in a Twitter post.
  • https://grail.cs.washington.edu/projects/dreampose/ - Building LLM applications for production. Here’s a new AI model called DreamPose that can generate Video from Image.
  • https://github.com/johannakarras/DreamPose#finetune-on-sample - The link is related to a question about finetuning a model on a subject-specific image before creating a video. The message also includes a question about the difference between the process and an edge detection algorithm.
  • Akto.io just launched AktoGPT - https://www.akto.io/blog/aktogpt - a new tool for generating text using GPT language models. @ankush also shares tips for deploying GPT LLM in production in a related tweet.
  • Akto.io just launched AktoGPT - https://www.akto.io/blog/aktogpt. Ankush talks about things to watch out for before deploying GPT LLM in production - https://twitter.com/Ankush12389/status/1646779395833741313.
  • https://arxiv.org/abs/2303.01469 - The message in the same link as the URL is related to avoiding premature optimization in the software world. The author also mentions that Akto is funded by Accel India as a disclaimer.
  • The LangChain website provides an example of integrating with Azure endpoints instead of OpenAI directly, and asks if anyone has tried it: https://python.langchain.com/en/latest/modules/models/llms/integrations/azure_openai_example.html
  • https://colin-scott.github.io/personal_website/research/interactive_latency.html: The message suggests that something has happened before and the link may provide more information.
  • The given URL is a link to a Google Colab notebook that provides insights into industry directions and opportunities through numbers and trends. The message in the same link mentions having the URL in one’s name.