-
Notifications
You must be signed in to change notification settings - Fork 28
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
LogGraph #32
LogGraph #32
Conversation
My feeling is we should support some intermediate format. Then we should make that work, and merge that PR that works with that. It might be nice to look at https://github.com/JuliaPlots/GraphRecipes.jl |
@oxinabox As you proposed in slack a function
In short I want to do what you suggested to do with
|
Exposing some intermediate representation is a good idea, and we can take inspiration from the relevant protobuffers, namely For the structure, I would roll a custom |
@shashikdm Provided the code is type-stable, I think that if you inspect the If the code is type-unstable, I believe you can't avoid actually running the type-unstable part of the code, and calling |
I have implemented barebones of |
Co-Authored-By: Mathieu Besançon <mathieu.besancon@gmail.com>
What about the subgraph (I think it's called like this) of tensor board? I'll give it a look this weekend, and probably merge it if you think it's ready. |
that depends on nodelabels. subgraph nodes should be named in a hierarchy using
will form a collapsable subgraph with name |
it's not ready yet. The values are not showing in the attributes. I'll look into that this weekend |
@PhilipVinc You can review and merge this.
The next task is to convert a model, Flux or Turing model to |
- change getdatatype -> jltype2tf - change @error -> throw
Started implementation of log_graph. I am aiming to support
Flux
models. As discussed earlier in Slack, I. am usingTrackedArrays
to build the graph as follows.2. Take
dummy
input along with themodel
(Chain
type)3. Apply the
model
todummy
input layer by layer with each layer creating a subgraph4. Within each layer,
node
s are created from the output to input by finding all theTracked
variables recursively5.
size
of output of eachnode
is also determined in the previous step