huh, interesting, I had no idea. Haha, but don't Windows users agree that WSL2 is a good thing? 😛
@rasbt
-
Installing TensorFlow via pip on Windows Python sessions
By
–
Wait, really? I am not a windows user, but can't you pip install TensorFlow in whatever Python session / terminal your are using?
-
Vector Database Scalability and Relevant Document Ratio Challenges
By
–
Yes but no . This problem also exists with external vector databases as the ratio of relevant-to-irrelevant docs decreases with larger databases. I think the Focused Transformer paper looked into this: https://
arxiv.org/abs/2307.03170 -
Clarification on longer-context models and future evaluation possibilities
By
–
Yes, sorry I see what you mean. I didn’t mean to imply that they studied these longer-context models like LongNet above. I was just referencing that we talked about these models (it would have been impossible for the authors to eval it at that time, but hopefully in the future)
-
LLM learns patterns from training data representation
By
–
Yeah exactly, I think that’s where the LLM gets this from (ie this pattern being represented in the training data)
-
Sebastian Raschka Arrives in Austin for SciPy Conference Workshop
By
–
Just arrived in Austin and looking forward to attending SciPy once more! Ready and excited for my workshop on “Modern Deep Learning with PyTorch” tomorrow! https://
github.com/rasbt/scipy202
3-deeplearning
… -
Google Search Ads Evolution: Subtler Ad Integration Strategy 2024
By
–
For those who complain about Google Search ads: Ads in 2024 will be become a lot more subtle …
-
Focusing on Improving Modeling Performance Over Computational Performance
By
–
*the focus here is on improving modeling performance, not computational performance.
-
Humble approach to long context lengths with minimal finetuning
By
–
This sounds a bit more humble than the recent 1M or 1B context lengths claims. But this requires only minimal (1000 steps) finetuning. And it allows long document summarization using LLaMA 7B and 65B models, for example. Link to the paper here: https://
arxiv.org/abs/2306.15595 3/3