Hi and thanks for doing this. I went through CBP twice in the span of 20 hours. I was granted entry once and denied the second time. The second time, I had to sign some electronic pad and later a form (I-275) was given…
There are some efforts to do fully open LLMs including their training data. Allen AI released their model (OLMo) and the data used for training the model under permissive licenses. https://allenai.org/
I wonder what would happen if this was applied to alpha code. Instead of generating 1 million codes for each problem during inference, do this kind of iteration during training and then generate less codes for each…
Hi and thanks for doing this. I went through CBP twice in the span of 20 hours. I was granted entry once and denied the second time. The second time, I had to sign some electronic pad and later a form (I-275) was given…
There are some efforts to do fully open LLMs including their training data. Allen AI released their model (OLMo) and the data used for training the model under permissive licenses. https://allenai.org/
I wonder what would happen if this was applied to alpha code. Instead of generating 1 million codes for each problem during inference, do this kind of iteration during training and then generate less codes for each…