One of the coolest things about r1 is how exposing the model’s CoT affects prompting.
Bad responses are more fixable.
Because the thoughts are exposed you can see where the errors were, remove ambiguity, and steer the AI.
Exposed CoT is the prompt debugger we’ve been missing.
Calling all @langchain developers
There are no Langchain communities in Twitter
Since there is no community
I created one for developers
To interact and learn from each other
Plan to keep a small and focused LangChain group
More details below ↓
ChatGPT won't replace jobs
Rather it will create more AI-related jobs
How?
I have prepared a ClickUp resource that will get chatGPT-based jobs
And for 24 hours, it's FREE
1. Follow me (so I can DM you)
2. RT this tweet
3. Reply "Jobs"
I made an early version of an EVERYTHING APP just like @friedberg describes on the @theallinpod 🤯
1. User describes the chatbot app they want📑
2. AI App creates all the variables for the NEW app📟
3. Transforms into a NEW custom chatbot.🤖
Demo and link to app↓
🚀 Only 1 month into coding, and I'm proud to present DeFiDocBot 🤖, a ChatGPT #DeFi research bot built to explore DeFi protocol Docs using natural language.
🛠️ Crafted with @langchain, @pinecone, hosted on @streamlit, and powered by @OpenAI 🌟