Kelly Kang

Full-stack SDE with a passion for AI and lifelong learning. Currently exploring new job opportunities. Let's build something amazing together!

[Kelly Kang @KKCodeBrew]

My latest article on enhancing Retrieval Augmented Generation performance! Special thanks to LangChainAI @hwchase17, @OpenAI, @pinecone, Huggingface @ClementDelangue, Lilian Weng @lilianweng, Mitchell Hashimoto @mitchellh, and Mayo Oshin @Mayowaoshin for their valuable insights.

[Kelly Kang @KKCodeBrew]

As a user of ChatGPT, I feel incredibly lucky to engage with such cutting-edge AI technology. However, as a developer working with the APIs, the rapid pace of changes can be overwhelming. Keeping up requires agility, but it’s a challenge I’m committed to.

[Dylan Patel @dylan522p]

ChatGPT system prompt is 1700 tokens?!?!?
If you were wondering why ChatGPT is so bad versus 6 months ago, its because of the system prompt.
Look at how garbage this is.
Laziness is literally part of the prompt.
Formatted in the paste bin below.
https://t.co/XSA85dys1I

[Kelly Kang @KKCodeBrew]

I’m considering creating a “Meme War” app.（In Chinese: “表情包大战”)

[LlamaIndex 🦙 @llama_index]

Announcing 🎵multi-modal support🎵 in create-llama, our simple to use command-line tool to generate a full stack LlamaIndex app! Get your app going with GPT-4-vision, and you'll be able to upload images to the web interface and get answers back about them. It takes just seconds to set up!

Just run `npx create-llama@latest` or visit https://t.co/4ms3SPi5Mm to learn more.

[Andrej Karpathy @karpathy]

LLM OS. Bear with me I'm still cooking.

Specs:
- LLM: OpenAI GPT-4 Turbo 256 core (batch size) processor @ 20Hz (tok/s)
- RAM: 128Ktok
- Filesystem: Ada002

[Kelly Kang @KKCodeBrew]

Had an incredible time at #AIESummit!Connected with amazing people, met top-notch AI companies, and learned so much. @jerryliu0 ’s talk on RAG and Daniel's @drosenwasser talk on TypeChat particularly provided valuable insights on how I will proceed with my AI project SurveyLlama!

[Kelly Kang @KKCodeBrew]

https://t.co/Olt9lBUeIY

[Kelly Kang @KKCodeBrew]

Excited to announce that I’m on the verge of launching #SurveyLlama, an intelligent conversational survey tool! It’s an open-source project and I’d love to have some community support to make it phenomenal. Let’s bring interactive and intuitive surveys to everyone! 🛠️ #OpenSource

[Kelly Kang @KKCodeBrew]

I attended the HTAP Summit! 🚀 Learned that the future of databases is like a Swiss Army knife, but instead of tools, it’s all about data and AI. Who knew? Transactional and analytical, living together in harmony. It’s like watching cats and dogs getting along. 😂 #HTAPSummit

[Kelly Kang @KKCodeBrew]

Dived headfirst into the #RaySummit23 ！🚀 Made more connections than my phone’s WiFi, absorbed info like a techy sponge, and, best part? Walked away with a free t-shirt. In-person events are basically social networking with wardrobe perks! 😎 #TechInStyle #SummitSwag

[Tom Long @Haymingway]

@KKCodeBrew @hwchase17 @OpenAI @pinecone @ClementDelangue @lilianweng @mitchellh @mayowaoshin https://t.co/RF072IXV0v

[Jerry Liu @jerryjliu0]

I have a recurring take that every AI engineer should learn how to build RAG from scratch 🧑‍🍳 - no packaged retrievers/chains allowed.

We’re excited to launch a new low-level tutorial series showing how to build the following:
✅ data ingestion
✅ retrieval

With low-level components only (LLMs, prompts, embeddings, vector stores, no indexes).

There’s more on the way too:
💡 response synthesis
💡advanced retrieval algorithms (routing, auto-retrieval, etc.)
💡agent loops (ReAct, OpenAI)

Check out our full guide: https://t.co/aBmvX0kzyP

Tutorial 1 (data ingestion): https://t.co/ADWxvImGsq

Tutorial 2 (retrieval): https://t.co/elYPT8ACA5

[Goku Mohandas @GokuMohandas]

Excited to share our production guide for building RAG-based LLM applications where we bridge the gap between OSS and closed-source LLMs.

- 💻 Develop a retrieval augmented generation (RAG) based LLM application from scratch.
- 🚀 Scale the major workloads (load, chunk, embed, index, serve, etc.) across multiple workers.
- ✅ Evaluate different configurations of our application to optimize for both per-component (ex. retrieval_score) and overall performance (quality_score).
- 🔀 Implement LLM hybrid routing approach to bridge the gap b/w OSS and closed LLMs.
- 📦 Serve the application in a highly scalable and available manner.
- 💥 Share the 1st order and 2nd order impacts LLM applications have had on our products.

🔗 Links:
- Blog post (45 min. read): https://t.co/QHgOXPT7S0
- GitHub repo: https://t.co/GMNrsHAhpY
- Interactive notebook: https://t.co/UPXSkwDt6h

@pcmoritz and I had a blast developing and productionizing this with the @anyscalecompute team and we're excited to share Part II soon (more details in the blog post).

[Kelly Kang @KKCodeBrew]

🚀 Excited to share my latest article on enhancing Retrieval Augmented Generation (RAG) performance! Dive in! https://t.co/dHbjbbAQwx

Special thanks to @hwchase17, @OpenAI, @pinecone, @ClementDelangue, @lilianweng, @mitchellh, and @Mayowaoshin for their valuable insights!

#AI

[Kelly Kang @KKCodeBrew]

Found out that https://t.co/SfCZFjXWhc is 6000x pricier than https://t.co/BtA2ZgYx0e. Guess even domain names know I’m destined for AI fame! 😂 #NextLevelDomains

Go Wenzheng👩🏻‍💻！

[Kelly Kang @KKCodeBrew]

I always consider I'm a tech newbie. Well, meet the fresher meat—my sister! GeZheng Kang.She just built a Chatbot that's smarter than both of us combined for a charity organization called Development Alert. https://t.co/2OiK5hnCOU

#SiblingRivalryInTech #ChatbotForGood

[LangChain @LangChain]

🎡 Introducing LangChain Hub 🦜🔗

A place to publish, discover, and try out prompts

We’re particularly excited about a centralized hub’s promise to enable:
-Encoding of expertise
-Discoverability of prompts for a variety of models
-Inspectability
-Cross-team collaboration

🧵

[Anil Chandra Naidu Matcha @matchaman11]

Incredible news from Meta 🤯

Code LLaMA Is out 🔥🔥

A version of Llama2 fine-tuned for code tasks

Same license as Llama2, available for commercial use

3 versions released

Code Llama: Base model.
Code Llama Python: Model specialized for Python.
Code Llama Instruct: Fine-tuned for instruction following.

Model link below ↓

[Caiming Xiong @CaimingXiong]

Code LLaMA is finally here. Congrats to @MetaAI @ylecun @syhw, etc..

And we'd like to introduce the finetuned llama2-70B model -- 🐒Lemur-70B🐒 again😉, a complement to Code LLaMA (7B, 13B, 34B) and maintains strong performance in text tasks.

Code: https://t.co/MburXL8pC4

[James Briggs @jamescalam]

@langchain agents are one of the most important tools offered by the library and a key building block in the current AI boom 🤖

Check out this @pinecone deep dive on agents from @fpingham and myself

https://t.co/vTMhWpkRHW

#NLProc #ArtificialIntelligence