In practice the safest way to get an LLM to do counting or string operations is to use external tools, like Python. ChatGPT has Python built in, but for simple problems you need to ask it to use it. So, if you *must*, this is the best way to count the r's in "strawberry" is:
LLMS
-
LLMs struggle counting tokens without Chain-of-Thought or tools
By
–
What I'm disputing is the specific, but I think common, misconception that of course any LLM could count *tokens* easily; it's just that tokenization makes letters unnecessarily hard. Without CoT or tools, LLMs are just bad at counting.
-
New ‘Autofocus’ feature improves Claude prompt window workflow
By
–
Every click matters 👀
— 🚨 AI News | TestingCatalog (@testingcatalog) 21 septembre 2024
A very minor "autofocus" feature for Claude power users.
Now, regardless of where your cursor is, if you start typing, the text will automatically be added to the prompt window. pic.twitter.com/VWQvNYCWXXEvery click matters A very minor "autofocus" feature for Claude power users. Now, regardless of where your cursor is, if you start typing, the text will automatically be added to the prompt window.
-
Claude Artifacts to VSCode Export Workflow
By
–
An upcoming Claude's Artifact to VSCode export works quite smoothly 👀 https://t.co/9XVBwgm8uw pic.twitter.com/dKxtcxLhfN
— 🚨 AI News | TestingCatalog (@testingcatalog) 21 septembre 2024An upcoming Claude's Artifact to VSCode export works quite smoothly
-
Hacker News for AI Research Papers Launched on GitHub
By
–
Hacker News for AI research papers is up on @github built with o1-mini and sonnet 3.5 in @cursor_ai github: https://
github.com/AK391/dailypap
ersHN
…
app: https://
huggingface.co/spaces/akhaliq
/dailypapershackernews
… -
ChatGPT passes apples-and-oranges reasoning challenge
By
–
ChatGPT o1 passes apples and oranges reasoning challenge. Impressive. Prompt: 8 apples 5 oranges 25 bananas 8 grapes 15 strawberries 23 watermelons 1 apple 18 raspberries 5 lemons 25 kiwis 15 peaches 21 blueberries
-
O1 Model Excels at Multi-File Refactoring and Architecture
By
–
Yeah the multi file refactors/architecture stuff on o1 is incredible.
-
ChatGPT Linguistic Bias Reinforces Dialect Discrimination
By
–
New blog post: Linguistic Bias in ChatGPT: Language Models Reinforce Dialect Discrimination
-
LLM Evaluation Methods and Future AI Development Implications
By
–
🎥 Watch as we dive into the world of #AI & cover:
— SambaNova (@SambaNovaAI) 20 septembre 2024
✅ How does the industry evaluate #LLMs today?
✅ Why use LLMs to evaluate other AI models?
✅ Why is this important?
✅ Future implications for AI development.
Accelerate your dev journey with: ➡️ https://t.co/GgqtX917iq pic.twitter.com/Vr7aE0i1znWatch as we dive into the world of #AI & cover: How does the industry evaluate #LLMs today? Why use LLMs to evaluate other AI models? Why is this important? Future implications for AI development. Accelerate your dev journey with: https://
cloud.sambanova.ai -
OpenAI o1 Models Advance AI Research at Microsoft
By
–
The @OpenAI o1 models represent one of the smartest advances in AI in a long time. Having just joined @Microsoft AI, one of the things I really look forward to is being able to contribute to some of these fruitful ideas to advance OpenAI’s mission. The opportunity to work