bless me father for i have sinned…. its been 7 minutes since i last completely bricked my conda environment
@jxmnop
-
AI Medical Devices: Risks and Concerns in Healthcare
By
–
hopefully you aren't working on any medical devices rn
-
Code Quality Trade-offs: Speed Over Testing in Academia
By
–
my former boss from google would cry a little bit if he knew the quality of code I am churning out daily in academia today i’m moving much faster though – creating and discarding hypotheses more quickly, learning more, etc but my unit tests don’t always pass
-
Finding Pre-Written Helper Function in Codebase
By
–
navigated to a file to write a helper function, but the code was already there, written by me, exactly where i was about to write it
-
Transformers Architecture Emergence in 2017 AI Timeline
By
–
transformers came out in 2017, just as i was coming onto the scene (maybe slightly before)
-
Three Major ML Breakthroughs: What Comes Next?
By
–
since i got into ML in 2017 or so there have been exactly three major advancements: 1. web-scale language model pretraining (OpenAI, 2020)
2. diffusion models (OpenAI, 2020)
3. language model refinement via human feedback (OpenAI, 2022) what’s next? it’s been a while -
AI tool reverse-engineers matplotlib code from graph images
By
–
free startup idea: Reverse Matplotlib, software that inputs figures and outputs code I often find myself in a position where I have a graph like this one, and want to make minor changes, but the raw data is hard to regenerate need a tool that can recreate the code & data for me
-
DMC Improves Transformer Inference Speed Through Context Compression
By
–
when contexts are long, attending to every single token in the past feels wasteful (and not at all how human brains work). feels like a natural setting for compression… DMC seems like a huge improvement in transformer inference speed — congrats to the authors!
-
Moore’s Law for Language Models: Performance and Speed Trade-offs
By
–
Trying to think about what a Moore's Law for Language Models would look like. the problem i'm running into is that there are two separate axes to improve upon: model ability for a given model size (perplexity) and inference speed (tokens per second) Anyway, we can all agree