SLM Lab
Search…
SLM Lab
v4.2.3
SLM Lab
🖥Setup
Installation
Quick Start
🚀Using SLM Lab
Lab Command
Lab Organization
Train: REINFORCE CartPole
Resume and Enjoy: REINFORCE CartPole
Agent Spec: DDQN+PER on LunarLander
Env Spec: A2C on Pong
GPU Usage: PPO on Pong
Parallelizing Training: Async SAC on Humanoid
Experiment and Search Spec: PPO on Breakout
Run Benchmark: A2C on Atari Games
Meta Spec: High Level Specifications
Post-Hoc Analysis
TensorBoard: Visualizing Models and Actions
Using SLM Lab In Your Project
📈Analyzing Results
Data Locations
Graphs and Data
Performance Metrics
🥇Benchmark Results
Public Benchmark Data
Discrete Environment Benchmark
Continuous Environment Benchmark
Atari Environment Benchmark
RL GIFs
🔧Development
Modular Design
Algorithm
Memory
Net
Profiling SLM Lab
📖Publications and Talks
Book: Foundations of Deep Reinforcement Learning
Talks and Presentations
🤓Resources
Deep RL Resources
Contributing
Motivation
Help
Contact
Powered By GitBook
Profiling SLM Lab

​
⏱
 Profiling Runtime

When developing a feature in SLM Lab, we may want to profile the program to check its performance and runtime, especially since deep RL software is complicated and involves many components.
We recommend Python's built-in cProfile and snakeviz to profile your program runtime. The example below runs the profiler and visualizes the program runtime broken down hierarchically by components. See an example of the graph: https://jiffyclub.github.io/snakeviz/#interpreting-results​
conda activate lab
pip install snakeviz
​
# say, to profile A2C on Pong
python -m cProfile -o a2c.prof run_lab.py slm_lab/spec/benchmark/a2c_gae_pong.json a2c_gae_pong train
​
# then Ctrl+C to kill the process after some time to collect runtime data
# use snakeviz to render graphs
snakeviz a2c.prof
​
# a browser will open, showing the runtime breakdown
Previous
RNN
Next - 📖Publications and Talks
Book: Foundations of Deep Reinforcement Learning
Last modified 9mo ago
Copy link