alex_petrenko

108 ms·

1. 1.▲
   by alex_petrenko 5y ago

   There is a physics engine underneath, but the physics interactions are simplified for accelerated training. I.e. blocks that the agents carry can pass through other blocks, but the agents can't pass through blocks that are already plac
2. 2.▲
   by alex_petrenko 5y ago

   Yes, this is correct. A single compute node can run enough of these little environments in parallel to achieve 1M FPS throughput.
3. 3.▲
   by alex_petrenko 5y ago

   "Mega" just stands for 10^6 here, the simulation throughput :)
4. 4.▲
   by alex_petrenko 6y ago

   The network in our work is exactly the same as in IMPALA paper. The overall score of our agent is slightly higher, it did better on some envs and it did worse on some other envs. These lasertag levels are exploration problems, and with a bi
5. 5.▲
   by alex_petrenko 6y ago

   Author of the paper here. Ask me anything!