RavOp
RavOP
Ravop is the Ravenverse op library. Ops are a fundamental unit of Ravenverse. Ravop Library contains the main object data types Tensor
and Scalar
upon which we can perform the various arithmetic and mathematical operations available in the ravop library in a distributed and decentralized manner.
###Ravop in Ravenverse
Ravop is responsible for the creation of ops and retrieving ops from ravcom and ravsock.
Setup and Installation
Create a virtual environment
Activate the virtual environment
Install RavOp
This step will automatically install the dependencies
Set RAVSOCK_SERVER_URL in the ravop/config.py file
Note: Make sure ravsock is up and running
Initializing Tensors , Scalars and Graphs
Tensors
:Raven Tensors are data type objects which are multidimensional arrays
Scalars
:Raven data type object for int
or float
type.
Graph
: For evaluating all the ops after creating it in a computational graph.
These data types can be initialized as follows:
Example
We can use raven ops to operate on these Tensors. Let's consider adding two Tensors . Adding two tensors can be achieved using the R.add() operation
Raven supports Unary and Binary ops :
unary ops
Binary ops
:
##Supported Ops Arithmetic Ops
Op name | parameters |
---|---|
lin | |
add | |
mul | |
sub | |
pos | |
neg | |
exp | |
natlog | |
pow | |
square | |
cube | |
square_root | |
cube_root | |
abs |
Comparision Opa
Op name | parameters |
---|---|
greater | |
greater_equal | |
less | |
less_equal | |
equal | |
not_equal |
Logical Operations
Op name | parameters |
---|---|
logical_and | |
logical_or | |
logical_not | |
logical_xor |
Statistical Ops
Op name | parameters |
---|---|
mean | |
average | |
mode | |
median | |
variance | |
std | |
percentile |
Tensor Ops
Op name | parameters |
---|---|
random | |
bincount | |
where | |
sign | |
foreach | |
one_hot_encoding | |
matmul | |
multiply | |
dot | |
transpose | |
sum | |
sort | |
split | |
reshape | |
concat | |
min | |
max | |
unique | |
argmax | |
argmin | |
expand_dims | |
inv | |
gather | |
reverse | |
stack | |
tile | |
slice | |
find_indices | |
shape |
Federated Learning
Op name | parameters |
---|---|
federated_mean | |
federated_variance | |
federated_standard_deviation |
Last updated