-
(Rank 6/1637) Finetune LLMs to solve the famous abstract reasoning challenge with Unsloth AI...
... and finished 6/1400 in ARC Prize 2024, then met intimately with Francois Chollet!
-
(Rank 13/832) Finetune an LLM to play the game of '20 questions'
...and demonstrate LLMs can act and play interactive games in a Kaggle simulation competition.
-
(Rank 2/98) Build an LLM to classify responses from other LLMs
...and placed 2nd in the 'H2O.ai Predict the LLM', met with H2O.ai CEO, and recruited by them!
-
(Rank 9/874) Santa again gave me another GOLD medal!
Another Christmas, another traditional Santa contest, another constraint optimization problem, another gold medal (rank 9/874)!
-
(Rank 1/2605) I finally WIN 1st place in a Kaggle competition!!!
...which is a prestigious Google Brain Ventilator Pressure Prediction!
-
(Rank 8/1140) Imitation learning by U-Net for a multi-player strategic game
I will show how a data-driven approach (semantic segmentation) can mimic rule-based top bots in the Kaggle 4-Agent Halite IV Competition!
-
(Rank 3/2618) Hidden Markov model (alone!) can almost win me a Kaggle competition
...instead I finished 3rd due to an unfortunate data leak that 2 other teams exploited in the Liverpool Ion Switching Competition
-
(Rank 14/3493) Ensemble zoo never disappoints!
... but must be equipped with counter-overfitting measures to win a Kaggle Gold medal in the presitigious Data Science Bowl!
-
(Rank 17/1620) Using Gurobi with custom handles
and compare with CPLEX, to solve an optimization problem in Kaggle Santa 2019!
-
The elegance of Policy Gradient in reinforcement learning
I will the 'magic' of PG of RL in a magical way.
-
(Rank 10/2749) Graph neural networks shine in quantum chemistry modeling
How I won a gold medal in the Kaggle Predicting Molecular Properties challenge!
-
(Rank 8/1874) Solve a very weird traveling salesman problem
How it forced us to write custom algorithms and finally seized a GOLD medal in Kaggle Santa 2018! (plus an introduction to Kotlin!)
-
(Rank 10/656) The power of DBSCAN -- an unsupervised clustering algorithm
How I surprisingly won my 1st Kaggle GOLD medal (10/656) with DBSCAN in a challenging Quantum Physics particle tracking problem