Building agents with Claude 3.7 Anthropic just released Claude 3.7, a hybrid reasoning model. Here, we share some tips for prompting / usage / configuration of the model. We also show how to build tool-calling agents w/ configurable reasoning. Video: https://
youtu.be/sHcu0YX-1u4
PROMPT ENGINEERING
-
Building Agents with Claude 3.7 Hybrid Reasoning Model
By
–
-
Claude 3.7 Sonnet’s Impressive One-Shot Code Generation Abilities
By
–
One of the things we've been most impressed by internally at Anthropic is Claude 3.7 Sonnet's one-shot code generation ability. Here are a few of my favorite examples I've seen on here over the past day:
-
LLMs Learn from Feedback During Test-Time Inference
By
–
Learning to Reason from Feedback at Test-Time This new paper proposes a new paradigm called FTTT (Feedback-based Test-Time Training) that enables LLMs to learn iteratively from environment feedback during inference. Key highlights include: • Test-time optimization for
-
Claude’s Autonomous Gameplay: Vision, Memory, and Function Calls
By
–
So, what makes this possible? Claude was given a knowledge base to store notes, vision to see the screen, and function calls which allow it to simulate button presses and navigate the game. Together, they allow Claude to sustain gameplay with tens of thousands of interactions.
-
Claude 3.7 Sonnet: Advanced Planning and Adaptive Problem Solving
By
–
Where previous models wandered aimlessly or got stuck in loops, Claude 3.7 Sonnet plans ahead, remembers its objectives, and adapts when initial strategies fail.
— Anthropic (@AnthropicAI) 25 février 2025
Critical skills for battling pixelated gym leaders. And, we posit, in solving real-world problems too. pic.twitter.com/scvISp14XGWhere previous models wandered aimlessly or got stuck in loops, Claude 3.7 Sonnet plans ahead, remembers its objectives, and adapts when initial strategies fail. Critical skills for battling pixelated gym leaders. And, we posit, in solving real-world problems too.
-
AI Model for Natural Writing Generation Beyond Code
By
–
When is someone going to release a model that can write I’d like to see ‘vibe writing’ instead of vibe coding.
-
Playing with AI prompts for creative writing exercises
By
–
Tbh I think this is a little generous! Going to play with the prompts a bit. But very cool exercise for a writer
-
Python Debugging Prompt for AI Capability Testing
By
–
2. Coding & Debugging Test
— God of Prompt (@godofprompt) 25 février 2025
Prompt:
"Below is a buggy Python script. Find the errors, fix them, and explain what was wrong:
python
CopyEdit
def factorial(n): if n = 1 return 1 else return n * factorial(n-1) print(factorial(5))" pic.twitter.com/XG1UA6cbQz2. Coding & Debugging Test Prompt: "Below is a buggy Python script. Find the errors, fix them, and explain what was wrong: python
CopyEdit
def factorial(n): if n = 1 return 1 else return n * factorial(n-1) print(factorial(5))" -
Creative approaches to maximizing AI model capabilities
By
–
I do! Just a good example of finding creative new ways to get more from the models
-
Wait Instruction Improves AI Model Reasoning Accuracy
By
–
A breakthrough could literally be a single word. UW + Stanford found telling their model to “wait” during reasoning improved answer accuracy https://
arxiv.org/pdf/2501.19393