* DR1 RNN requirements
Add requirements related to RNN networks
remove redundant requirements
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
- Requirement Name
- Description
- * - RF-?
+ * - RF-6
- HE: profiling
- Runtime should be able to gather or import information about execution time of specific operations
- * - RF-?
+ * - RF-7
- HE: scheduling
- Runtime should be able to use profile information to choose optimal assignment of operations to execution devices
- * - RF-?
+ * - RF-8
- HE: mixing backends
- Runtime should be able to mix different types of execution devices in one compilation to achieve best performance
RNN
---
-Detailed requirements will be described on DR1 milestone (end of July)
-
+No special requirements
SW Non-Functional Requirements
==============================
- Description
* - RNF-2
- Reliability: input
- - If the incorrect input data are provided inference should not be performed, but user should be informed about all errors which were met
+ - If the incorrect input data is provided inference should not be performed, but user should be informed about all errors which were met
Security
--------
- Neural network model (Instance of NN built with ML framework)
* - NN compiler
- The compiler for neural network
+ * - NN Package
+ - Serialized representation of NN model generated by NN Compiler
+ * - NN runtime
+ - Infrastructure to perform neural network inference
* - ML framework
- The machine learning framework
* - TF/TF Lite
- Computational Graph
* - RNN
- Recurrent Neural Network
- * - NN Package
- - Serialized representation of NN model generated by NN Compiler