Caleb
@calebsheridan
Followers
2K
Following
2K
Media
90
Statuses
746
Have you tried setting temperature to 0/1?
Joined November 2010
If you've made it this long, also try the @.web command which searches the internet for you if you're having issues using a new library or technique.
0
0
0
Generate videos in just a few seconds. Try Grok Imagine, free for a limited time.
580
2K
8K
For crypto use cases (like memes and agents), the libraries you want to use the most are not well understood by the @cursor_ai models. 🤦♂️. In order to use viem by @wevm_dev you need to add customs docs and reference them with @ tags in your messages. 🙌
1
0
0
Reason 3: Custom rules/context. The best agents today are very focused (v0 + shadcn, replit + flask). Get the magical focus from Cursor by giving some extra rules as context. Prompt Directory --->
1
0
0
Reason 2: Cursor is fast and cheap. It has a great dev experience. It enables you to iterate faster. There is a generous free tier. And even at $20 per month, the paid service is cheaper than using a chatbot or api credits.
1
0
0
Reason 1: Cursor allows you to review changes BEFORE they are applied using diff mode. You don't need to worry about dealing with the aftermath of your coding environment introducing a bunch of unintended changes. 👻
1
0
0
Even with agents like Replit and Bolt available, I find myself turning to @cursor_ai for nearly every dev task. Here are three reasons & a tip for using it to build crypto agents. .
2
2
8
BitNet models are really promising. Looking for more models to try now, but not many are out. Expect we see some good ones released and benchmarked soon. - gh: - fine-tuning instructions (published a month ago):
huggingface.co
How to deploy a 100B model on your CPU devices? 🔥. Excited to introduce bitnet.cpp, our inference framework for BitNet b1.58 🚀🚀.
0
0
2
Cline is open-source alternative to Cursor that runs as a VS Code extension. @sdrzn is putting on an amazing show building this tool with constant shipping.
Introducing Cline (formerly Claude Dev), an AI assistant that can use your CLI aNd Editor. v2.0 brings exciting updates: responses are now streamed into your editor, a cancel button for better control over tasks, a new XML-based tool calling prompt resulting in ~40% fewer
0
0
4
And some tips for using Replit:.- Don't try to manage libraries or frameworks, let Replit do its thing (js not supported).- Do split instructions into smallest possible pieces of work.- Do ask Replit to refactor its code often to help keep files small.
0
0
0
Some things that I wish worked better with @Replit.- Agent often asks for QA before committing code, but sometimes auto commits buggy code 😭.- All environments use the same db service, so if the dev process wipes the db the deployed version is also nuked 🫡.
1
0
0
Replit extremely good for prototyping server work and worth trying. Bootstrapping a python api, db and web ui can just be tasked to Replit at this point. Best feature: automatic checkpoints to back out of any bad code changes.
AI is incredible at writing code. But that's not enough to create software. You need to set up a dev environment, install packages, configure DB, and, if lucky, deploy. It's time to automate all this. Announcing Replit Agent in early access—available today for subscribers:
1
2
6
RT @EdenNetwork: Alright folks, let's have some So Extra fun! 🌈. 👉 Either reply below with, or tag us in a tweet of, a screenshot of your c….
0
6
0
RT @EdenNetwork: 🌈 Introducing: So Extra 🌈. A new proof-of-concept app developed on Flashbots' SUAVE. Live on the new SUAVE Toliman testne….
0
21
0
RT @zeroXbrock: 1. How could you prove that your server is acting in good faith? Even though this isn’t a real concern for NFTs, honesty ca….
0
1
0
So Extra ✨. Sealed bid auction on SUAVE to buy extra data.
9/ So Extra - - Have you ever wanted to put your twitter handle in the extra data of an Ethereum Execution block? Well for the right price now you can! This SUAVE app utilizes Kettle privacy to ensure that bids are sealed.
0
0
1
Big flip from closed to open source models.
Leaked (possibly real?) evals for Llama 3.1. Base models, not instruct. Open-source is about to be SOTA — even the 70B is > gpt-4o, and this is before instruct tuning, which should make it even better. Tomorrow is going to be wild.
0
0
2
Simple reference on Github:
github.com
Contribute to calebsheridan/langfuse-playground development by creating an account on GitHub.
0
0
1
(And today I was worried that LiteLLM might need an upgrade to use the latest models on @GroqInc, but it just worked by specifying the right model names from the docs.).
1
0
1
My first issue using Langfuse is that not every model interface is fully compatible. (It works best with OpenAI by default). Fortunately, the @LiteLLM SDK integration helps keep things consistent. This makes it trivial to grid search across models, prompts and configurations.
1
0
1