Haha I appreciate it. HyperWrite is worth it, especially if you write frequently!
@mattshumer_
-
Jamba Chat Model to Be Open-Sourced Tonight
By
–
Jamba is in the trainer right now. If all goes well, I'll open-source a chat version tonight.
-
Jamba Model Excitement and Testing Plans
By
–
Hell yeah @winglian Jamba looks like an awesome model Excited to push it to the limit
-
Claude 3 Prompt for Automatic Unit Test Generation
By
–
Here is an insanely useful Claude 3 prompt for engineers. Use it to automatically generate unit tests for your code: — You are an expert software tester tasked with thoroughly testing a given piece of code. Your goal is to generate a comprehensive set of
-
Full Fine-Tuning vs Other Methods: Overfitting Prevention
By
–
Not necessarily. Done right, full fine-tuning is usually better… my assumption for why this prevents overfitting is it’s not really affecting the parameters as much as a FT. Total speculation, I may be wrong.
-
LoRA QLoRA Fine-tuning Prevents Overfitting Better
By
–
While LoRA/QLoRA isn’t as powerful as full fine-tuning, it often can prevent overfitting from suboptimal tuning
-
Mistral 7B Extended Context Training and Dataset Improvements
By
–
On the model side, longer context length for the 7B! If you could train over the new Mistral 7B w/ 32K that’d be awesome. As for the dataset, I haven’t found many issues with it 🙂
-
Developer Credits Training Code Repository Contribution
By
–
I pulled the training code from one of my old notebooks from over a year ago, which was based on yours. I’m so sorry, I totally forgot where I got the training code from. Will absolutely add credit to the repo — love your work (esp. LazyMergeKit!!)
-
Claude-Investor Transformed Into Mobile App by Martin
By
–
Crazy… Martin turned Claude-Investor into an app! https://t.co/NXmuZpyUjN
— Matt Shumer (@mattshumer_) 27 mars 2024Crazy… Martin turned Claude-Investor into an app!