Create first version of Dockerfile

[alesolano]

Description

We need a Dockerfile. This way it'll be easier to do CI/CD workflows, train the models with AWS sagemaker and share our work to everybody.

Requirements

Nothing comes to mind right now.

Acceptance Criteria

• Dockerfile is added and can be built.
• python -c "import tensorflow as tf" runs successfully.

[alesolano]

The command

 python -c "import tensorflow as tf; print(tf.reduce_sum(tf.random.normal([1000, 1000])))"

returns

tf.Tensor(582.03674, shape=(), dtype=float32)

Doing python agent.py returns:

Loading environment football failed: No module named 'gfootball'
=== Episode 0 ===
2021-09-28 16:36:14.264100: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)
1/1 [==============================] - 1s 602ms/step - loss: 6.5104e-04
1/1 [==============================] - 0s 33ms/step - loss: 1.9586e-04
1/1 [==============================] - 0s 32ms/step - loss: 6.4046e-04

and on and on...

---

Please @DelgadoPanadero / @chucheria check that works for you.

To build it, do docker build -t clarai:dev-cpu -f Dockerfile.cpu .

To run it, do docker run -v $(pwd):/root -p 8888:8888 -it clarai:dev-cpu bash

[chucheria]

Perfect. The only thing to change in the template is COPY ./requirements.txt . for COPY ./requirements_cpu.txt . in line 32, otherwise it throws an error.

Thanks so much, it works perfect.

[chucheria]

Thinking about this, if we are going to have multiple requirements files, it is better to move to a new folder like:

requirements/
|_ common.txt
|_ gpu.txt
|_ dev.txt

or whatever we have.

[alesolano]

Perfect. The only thing to change in the template is COPY ./requirements.txt . for COPY ./requirements_cpu.txt . in line 32, otherwise it throws an error.

Thanks so much, it works perfect.

True!

Thinking about this, if we are going to have multiple requirements files, it is better to move to a new folder like:

requirements/
|_ common.txt
|_ gpu.txt
|_ dev.txt

or whatever we have.

With this I'm not so sure. I'll check what the best practices are. After all, it would be just two files (requirements_cpu.txt and requirements_gpu.txt).

Though we also have two different Dockerfiles... Yep, let me check what the best practices are.

[chucheria]

Great, thank you!!

[alesolano]

Closing this and moving to #23