-
Notifications
You must be signed in to change notification settings - Fork 13
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
Add optimizer overrides to explorer #1604
base: main
Are you sure you want to change the base?
Conversation
memory_layout = layout.memory_layout_as_int | ||
if memory_layout is not None: | ||
result.append( | ||
graph_builder.KeyValue( | ||
key="memory_layout", | ||
value=str(ttnn.TensorMemoryLayout(memory_layout)), | ||
make_editable_kv( | ||
graph_builder.KeyValue( | ||
key="Tensor Memory Layout", | ||
value=str(ttnn.TensorMemoryLayout(memory_layout)), | ||
), | ||
editable={ | ||
"input_type": "value_list", | ||
"options": [str(o) for o in ttnn.TensorMemoryLayout], | ||
}, | ||
) | ||
) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I've tested specifically the memory_layout
attribute and it is not returning the editable
property to the UI. 😕
For reference, here is where the editable
attribute is expected in the response:
{
"graphs": [
{
// ...
"nodes": [
{
// ...
"attrs": [
{
"key": "",
"value": "",
"editable": {} // <= Here is the editable property
},
// ...
]
},
// ...
]
}
]
}
Here are the available types for the editable attribute: https://github.com/tenstorrent/model-explorer/blob/main/src/ui/src/components/visualizer/common/input_graph.ts#L156-L178
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Tested on this PR. Editable should be passed on convert command that comes after execute command is called.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The problem that i am hitting after this is that even when i change the value in the editable in the UI it is not sent in settings. Settings is equal to: {'optimizationPolicy': 'DF Sharding'}
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So, the way it is working now, when you click the upload button, it will send the changes to the server and then clean up the changes.
Also, it is not sending the changes to either the execute
command nor the convert
command that comes after.
I'll make a change to send it on both commands
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@odjuricicTT please check this branch: tenstorrent/model-explorer#19
e959b38
to
58bb8aa
Compare
ab70071
to
d9a8568
Compare
d9a8568
to
b394618
Compare
@vprajapati-tt I did make some changes to how location is parsed and used. We might be missing functionality here to cover all the usecases discussed. Let's revisit this after the holidays. |
TT-explorer now supports editing output tensor layout of ops from the UI.
Changes made:
Closes #1178