Shubham Subhnil's repositories
RacingCARLA
Learning Model Predictive Control (LMPC) for autonomous racing in CARLA 3D environment.
BAC-DAC-gym
Bayesian Actor-Critic with Neural Networks. Developing an OpenAI Gym toolkit for Bayesian AC reinforcement learning.
CoGen_Benchmarking
Benchmarking existing RL algorithms including model-free and model-based approaches on confounded versions of popular environments. Tests generalization and sample efficiency.
Vehicle-Dynamics-Toolkit
Some advanced tools for race car design - Steady state and transient dynamics, Tyre Data synthesis
Causal-Gridworld
Testing the causal implications of the wind in the gridworld environment. The wind is the confounder.
CausalBench
Official data and code for our paper Systematic Evaluation of Causal Discovery in Visual Model Based Reinforcement Learning
CausalCuriosity-test
Causal Curiosity fork for testing in confounded environments.
CDL-bench
Benchmarking CDL in confounded MDP and POMDP settings
D4PG-bench
Benchmarking D4PG in confounded environements.
dreamerv3-benchmod
Modifying DreamerV3 for benchmarking in confounded environments
Lane-Lines-Detection-Python-OpenCV
Lane Lines Detection using Python and OpenCV for self-driving car
mamba-test
Meta-RL Model-Based Algorithm - Confounding tests
dreamer-new
Updated version of DreamerV3 cloned from danijar/dreamerv3
dv3-torch
Benchmarking DreamerV3 with Plan2Explore.
FCD-bench
Fine-Grained Causal Dynamics Learning with Quantization for Improving Robustness in Reinforcement Learning (ICML 2024)
GRADER-bench
Repository for benchmarking GRADER in confounded environments for zero and few-shot generalization.
mocoda-b
Testing MoCoDA in DM Control Suite and confounded environments.
mpo-bench
Baseline tests on MPO with unobserved confounders
MWM-bench
Benchmarking MWM in confounded environments
P2P-bench
Code accompanying paper "Plan To Predict: Learning an Uncertainty-Foreseeing Model for Model-Based Reinforcement Learning".
RIA-bench
Benchmarking RIA in confounded environments for zero and few-shot generalization. Now compatible with TF2.
RIA_base
RIA base version. With new Walker environment similar to DM Control Suite physics and reward function.
rl2-bench
Implementation of 'RL^2: Fast Reinforcement Learning via Slow Reinforcement Learning'
sac-bench
PyTorch implementation of Soft Actor-Critic (SAC) for Unobserved Confounders
slac-bench
Stochastic Latent Actor-Critic: Deep Reinforcement Learning with a Latent Variable Model
TMCL-b
Trajectory-wise Multiple Choice Learning for Dynamics Generalization in Reinforcement Learning (NeurIPS 2020)