Agents are a huge research focus in the world of #GenAI . Here is a great tutorial explaining what agents are and how to implement an agent based app with Fireworks AI LangChain and #mongodb mongodb.com/developer/prod…

Anthropic Anyscale Amazon Web Services cohere Credal.ai Fireworks AI Google Cloud gravity9 LangChain LlamaIndex 🦙 Microsoft Azure Nomic AI Pureinsights Technology Together AI Welcome launch partner Fireworks AI to the MongoDB AI Applications Program, designed to help organizations build and deploy modern apps enriched with #GenAI technology at enterprise scale.

mongodb.social/6017jHm4J

I put Stable Diffusion 3 through its paces with a handful of prompts using Fireworks AI
It creates compelling images, gets text right most of the time and can sometimes fall into an uncanny valley.

tomsguide.com/ai/ai-image-vi…

Between the two top managed Llama 3 70b providers (Together AI & Fireworks AI), Fireworks is miles ahead for inference. They consistently deliver sub-500ms full completions, while Together has average latencies in the seconds with tail latency >15s.

The OpenAI provider for the Vercel AI SDK v0.0.7 supports object generation and streaming for Fireworks AI

Here are examples that use firefunction-v1:

Looking to explore the capabilities of Stable Diffusion 3? Fireworks AI has added UI playgrounds for Stable Diffusion 3 and Stable Diffusion 3 Turbo, using your Stability AI API key.

You can get started here: fireworks.ai/models/stabili…

Llama 3 8b and 70b now available for fine tuning.

Fireworks lets you deploy 100 fine tuned models for fast, serverless inference at 0 extra cost!

Get started here 👇

readme.fireworks.ai/docs/fine-tuni…

mag e/acc Groq Inc Of course, with total response times (time to first 100 tokens) of just 1 second and Latency (seconds to first chunk) of 0.26 seconds, our platform delivers unrivalled speed without sacrificing an ounce of quality!

Excited to keep pushing the envelope on fast, cost-effective Gen…

Suprising, In my testing Fireworks AI is almost as fast as Groq Inc !
Groq is still faster 1.28s vs 1.42s for fireworks, but I doubt they build their own chip

I updated Siri LLama so it can run LLaVA on any image + support any model on both ollama and Fireworks AI AI. who wants an ai pin now?

Optimizing LLM inference is complicated

- what's your $- & latency budget?
- what is your input / output token ratio?
- are prompts repetitive or unique?
- what hardware do you have?

Lin Qiao explains the many trade-offs Fireworks AI manages for enterprise clients

🔗👇

One key optimization strategy is sharding

In this clip, co-founder Dmytro Ivchenko outlines the 3 main kinds of sharding

Lots more detail in the full episode:

cognitiverevolution.ai/ai-inference-g…

Dialoqbase v1.8.0 has been released with huge improvements and features.

Here is an example using the llama3 model via .Fireworks AI API to chat on a GitHub repo.

Powered by LangChain

hackathon time: MongoDB with support of Fireworks AI, Upstage, and LlamaIndex 🦙 in SF
let‘s see what I can learn 🧵 — it will be a long day: 9:30am breakfast and planned finish at 10pm

GenAI Hackathon by MongoDB
with Karissa Fuller Laurie Voss Fireworks AI Lucy Park Upstage

It was great!

Blazing fast function calling by Fireworks AI ⚡️
This is the fastest function calling ive seen, exceptional work by Lin Qiao and team. This video is not sped up.

code: github.com/phidatahq/phid…

🚀 Today’s the day!!!! In just three hours I’ll be kicking off the 4/20 MongoDB GenAI Hackathon in San Francisco. Will I see you there?!

👋 I hope so! 😃

#mongodbatlas #vectorsearch