Create Financial Agents with Vision 👀 - Powered by Claude 3 Haiku & Opus
Вставка
- Опубліковано 8 чер 2024
- Learn how to create agents without any agentic frameworks like CrewAi or AutoGen. We will create Claude 3 Haiku as sub-agents to read financial documents. Claude 3 Opus will be used as super agent to orchestrate these sub-agents.
🦾 Discord: / discord
☕ Buy me a Coffee: ko-fi.com/promptengineering
|🔴 Patreon: / promptengineering
💼Consulting: calendly.com/engineerprompt/c...
📧 Business Contact: engineerprompt@gmail.com
Become Member: tinyurl.com/y5h28s6h
💻 Pre-configured localGPT VM: bit.ly/localGPT (use Code: PromptEngineering for 50% off).
Signup for Advanced RAG:
tally.so/r/3y9bb0
LINKS:
Github: github.com/PromtEngineer/Yout...
Anthropic cookbook: github.com/anthropics/anthrop...
TIMESTAMPS:
[00:00] Getting started with Agents
[01:09] Setting up the financial agent
[01:55] Set up the environment for agents
[02:45] Financial Documents - quick look
[03:35] Converting Text to images
[05:00] Creating Agents - super prompt
[06:15] sub-agents powered by haiku
[08:32] super-agent
All Interesting Videos:
Everything LangChain: • LangChain
Everything LLM: • Large Language Models
Everything Midjourney: • MidJourney Tutorials
AI Image Generation: • AI Image Generation Tu... - Наука та технологія
Totally agree! Building an agent design using LLM directly is the way to go. LangChain or LlamaIndex abstracting the process is premature. We need to tackle the process ourselves, understand what's happening, and not just encapsulate it. Most "agents" are essentially LLM used multiple times, with different contexts, and in a pipeline. Even tools (functions) are a specific way of calling LLM. Unless we create an autonomous agent that acts in real-time based on environmental events, it all boils down to calling LLM. Let's not miss the chance to learn by simplifying the code instead of jumping to libraries that are still in version zero and constantly changing. (This is my opinion)
I agree, in most cases, the frameworks also adds extra code and details that are not needed at all.
@@engineerprompt Indeed, one for example, the plethora of JavaScript frameworks developed over the years, many arguments between fans, and now can be done by a LLM fine-tuned for coding. Using ES6 directly gives developers full control, bypassing the need for complex frameworks which often create divisive "camps." This to me reflects a form of capitalist consumerism culture in tech, promoting unnecessary layers that complicate frontend development.
Funny is religiously, we forget they're merely encapsulated JavaScript. This fixation shifts educational focus from fundamental JavaScript to framework-specific courses on many online platforms. To me this reinforcing an illusion of knowledge rather than genuine understanding.
While helper libraries are useful, relying on them from the start in learning LLM app development or anything else can obscure essential skills, and build dependency over mastery.
Can you prepare a video on similar way but for to get data from Excel file & SQL database
Python and or PHP, or mysql workbench or phpmyadmin or Google sheets will do this, no need for AI. Unless you are trying to manipulate and understand the data in a way that would take a human longer than 20mins to do... Most AI will use python to do this anyways by generating the code to get the data for it to analyze... I wish people would stop using AI for basic inane tasks, it's a waste of energy! Do you know how much water & power Ai uses, and it's being wasted on these simplistic tasks because people are too lazy to learn nowadays!
Import pandas as pd
DF = pd.read_excel(filepathname_here)
Job done!
sure, let me see what I can do.
Isn't this analysis for a very specific 'prompt' or task ? Don't understand how is this leveraging the 'generic' nature of LLM. The models are only being used for document understanding and information extraction and not for the original problem statement i.e. 'financial agent' which sounds a lot more generic in nature.
I believe the gist of the video is how to build your own agents without having to use agent frameworks (CrewAI, Autogen). Also, the agent using the larger and more capable model (Opus) is providing the final analysis (Financial Agent) from what was gathered by the other agents using the cheaper model (Haiku).
So I'm confused about this platform. Is it open source for individual users and free? I know there's a cost when connecting an API to openai, but outside of that, is there any hidden costs?
the only cost is associated with the api calls.
how to run it with local llm?
I think Mixtral will be able to do it (without the vision part).
Samaj nahi aa raha kuch bhi
It's not free
Yup, you are right