norci

It's a good idea. There's an implementation called GA3C: ### code https://github.com/NVlabs/GA3C ### paper - https://arxiv.org/abs/1611.06256 - https://on-demand.gputechconf.com/gtc/2017/presentation/s7169-Iuki-Frosio-a3C-for-deep-reinforcement-learning.pdf

I think a parameter server is not necessary for A3C. there's a simple implementation: https://github.com/MorvanZhou/pytorch-A3C It uses multiple processes to run the agents. (multi-thread is not supported in Python?) The...

I tried to implement a cuda wrapper like this: ``` using CUDA struct CudaEnv{E}

what about update the env? that the state and actions are matrix, so as to use BLAS to accelerate it. This might be faster than a `MultiThreadEnv`, for the create...

It seems `cule` ported the whole atari to gpu. see - https://github.com/NVlabs/cule/blob/master/cule/atari/wrapper.hpp#L140 - https://github.com/NVlabs/cule/blob/master/cule/atari/cuda/kernels.hpp#L220 - https://github.com/NVlabs/cule/blob/master/cule/atari/m6502.hpp So the env can be run in batches. I think the internal state should...

My reason for a weighted features: My application is using a decision tree, it's output is a function. This function will be called in the subsequent code, so there are...

is this code useful? https://github.com/optuna/optuna/blob/master/optuna/multi_objective/samplers/_motpe.py > Multi-objective sampler using the MOTPE algorithm. > This sampler is a multiobjective version of :class:`~optuna.samplers.TPESampler`.

btw, groupby function in DataFrames.jl is much faster than JuliaDB.jl's. Why?

I have figured out: ```julia julia> open("/tmp/a.bson.zstd", "w") do fd stream = ZstdCompressorStream(fd) BSON.@save(stream, x = "abcd"^4096) close(stream) end julia> open("/tmp/a.bson.zstd", "r") do fd stream = ZstdDecompressorStream(fd) display(BSON.load(stream)) close(stream) end...

Sorry I made a mistake. It seems the problem is in https://github.com/JuliaReinforcementLearning/ReinforcementLearningCore.jl . I'm trying to fix it.