Important breakthroughs in AI will come from tinkering around with ideas and models on Jupyter notebooks.
RESEARCH
-
Survival as an Emergent Objective in Hostile Environments
By
–
What if the objective is something that isn’t hand-engineered per se, like an agent’s ability to survive in a hostile (perhaps, simulated) environment?
-
MLX LoRA Fine-tuning Guide for Language Models
By
–
https://
github.com/ml-explore/mlx
-examples/blob/main/llms/mlx_lm/LORA.md
… -
GPU Evolution: From Single Unit PhD Research to Multi-Node Computing
By
–
As a person who only spent one gpu during all phd time on the topic of transfer learning, there is a gap between old days and present day with a bunch of nodes. While I am already yearning for more
-
SymbolicAI: Logic-Based Generative Models and Solvers
By
–
SymbolicAI: A framework for logic-based approaches combining generative models and solvers Dinu et al.: https://
arxiv.org/abs/2402.00854 #ArtificialIntelligence #DeepLearning #MachineLearning -
Observer Perspective Limits in Simulated Universes
By
–
Perspectives of observers are limited by the ways in which observers can be implemented. Eg. you exist in a multi way universe, you don’t know which path you are in. If you exist in a single thread universe, you don’t know what the simulator is up to.
-
Randomness Inevitability From Observer Perspective
By
–
Randomness is inevitable and inevitably from an oberver perspective
-
Database Adaptation for MLX LoRA Implementation
By
–
I took the original database and adjusted it using a Python code so that it could fit the format required by MLX. You can see all the information here. https://
github.com/ml-explore/mlx
-examples/tree/main/lora
… -
IPAdapter Masks with ControlNet Depth Pose Experimentation
By
–
Some experimentation with mixing IPAdapter inputs and masks with fixed controlnet depth and pose (no prompts)
-
NSF NAIRR Pilot Opens Advanced Computing Resources for Safe AI
By
–
ICYMI: The @NSF is accepting applications until March 1 for those interested in requesting access to a set of advanced computing resources as part of the #NAIRR Pilot. The focus in the initial phase is on advancing safe, secure & trustworthy AI. Read more: https://
nairrpilot.org/allocations