Content - 92c2639e3b3f0358db0e30bbc3d43f1a9169f311 - b1cf431/PretrainedModels/README.md

visit type:

Tip revision: a05c3c642648373f4ede0956e4286257c3d59a61 authored by liqfu on 24 August 2018, 17:46:51 UTC
CNTK splice allows broadcast. This case is handled in the change. For noop (identity) ops, its inputs and outputs types shall be set according to upstream ops. ToBatch/ToSequence and Unpack Batch/Sequence ops added during model importing need tp be skipped. Model import need to handle ops with multiple outputs.

Tip revision: a05c3c6

README.md

# CNTK Pre-trained Models

This folder contains pre-trained models that can be used as DNN featurizers or as the basis for transfer learning. Some of the models are trained from scratch by CNTK, and some others are converted from other toolkits such as Caffe. The models are categorized by their application domains. 

To download a model run 

`python download_model.py <model_name>`

To see the list of available models run 

`python download_model.py list`

## [Image models](./Image.md)

Browse the archive

https://github.com/Microsoft/CNTK