[EvgeniyZh]

Obviously I don't have any proofs that I can provide. Just small talk with people here and there. Some NVidia engineer told me that Google is (was?) very uncooperative to work on some DL stuff together, as opposed to, say Facebook.

About internal version, this is pure speculation, based on idea that TPU are programmed with some framework, thus out should be TF, thus there is closed part and there should (could) be others.

[vrv]

I'm sorry to hear of that experience, and it's certainly not intentional. We should (and I will) always try to be better (I know some will forever view Google as an antagonistic behemoth but usually the engineers on both sides are trying to do the right thing; it just takes time to come to consensus and understanding sometimes).

Regarding internal version: we built TensorFlow to support devices as modular plugins: the CPU and GPU devices are built this way (you can read the source code to see how device registration works), and the same registration mechanism is used for the TPU code, which can't be opensourced due to internal dependencies. Internal customers just link in an additional library to get TPU support, but it still uses the same core codebase that is available in the opensource world. I know this because I wrote a lot of the device modularity and TPU binding code :)

[EvgeniyZh]

I see, maybe I was a bit exaggerating.

I'd really like to talk more about birth Tensorflow and TPU, but unfortunately I won't come to OSCON. May be some other time :)