feb 2024

Author: swyx

Monthly Notes/Feb 2024 notes.md

@@ -67,13 +67,14 @@ Frontier models
 
 ## product launches
 
-
+- vercel ai integrations - aggregation theory on AI
 
 
 ## Misc reads
 
 - discussions
 	- [Umichigan selling student data](https://x.com/suchenzang/status/1758020313689006374?s=20) 
+	- [Interesting discussion on Replicate's poor GPU cold start with CEO](https://news.ycombinator.com/item?id=39411748)
 - learning
 	- **[TPU-Alignment](https://github.com/Locutusque/TPU-Alignment)** - Fully fine-tune large models like Mistral-7B, Llama-2-13B, or Qwen-14B completely for free. on the weekly 20hrs of TPUv3-8 pod from Kaggle 
 	- [undo llama2 safety tuning with $200 LoRA](https://www.lesswrong.com/posts/qmQFHCgCyEEjuy5a7/lora-fine-tuning-efficiently-undoes-safety-training-from?)