changed entry_point naming convention

docs/providers.md

@@ -1,7 +1,7 @@
 ### Model Provider
 
 ```python
-def skyline_model_provider() -> torch.nn.Module:
+def deepview_model_provider() -> torch.nn.Module:
     pass
 ```
 
@@ -10,7 +10,7 @@ The model provider must take no arguments and return an instance of your model (
 ### Input Provider
 
 ```python
-def skyline_input_provider(batch_size: int = 32) -> Tuple:
+def deepview_input_provider(batch_size: int = 32) -> Tuple:
     pass
 ```
 
@@ -20,7 +20,7 @@ The input provider must take a single `batch_size` argument that has a default v
 ### Iteration Provider
 
 ```python
-def skyline_iteration_provider(model: torch.nn.Module) -> Callable:
+def deepview_iteration_provider(model: torch.nn.Module) -> Callable:
     pass
 ```
 
@@ -72,20 +72,20 @@ class ModelWithLoss(nn.Module):
         return self.loss_fn(output, target)
 
 
-def skyline_model_provider():
+def deepview_model_provider():
     # Return a GPU-based instance of our model (that returns a loss)
     return ModelWithLoss().cuda()
 
 
-def skyline_input_provider(batch_size=32):
+def deepview_input_provider(batch_size=32):
     # Return GPU-based inputs for our model
     return (
       torch.randn((batch_size, 3, 256, 256)).cuda(),
       torch.randint(low=0, high=9, size=(batch_size,)).cuda(),
     )
 
 
-def skyline_iteration_provider(model):
+def deepview_iteration_provider(model):
     # Return a function that executes one training iteration
     optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
     def iteration(*inputs):

examples/densenet/entry_point.py

@@ -4,18 +4,18 @@
 import densenet
 
 
-def skyline_model_provider():
+def deepview_model_provider():
     return densenet.densenet121().cuda()
 
 
-def skyline_input_provider(batch_size=16):
+def deepview_input_provider(batch_size=16):
     return (
         torch.randn((batch_size, 3, 224, 224)).cuda(),
         torch.randint(low=0, high=1000, size=(batch_size,)).cuda(),
     )
 
 
-def skyline_iteration_provider(model):
+def deepview_iteration_provider(model):
     optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
     def iteration(*inputs):
         optimizer.zero_grad()

examples/gnmt/entry_point.py

@@ -249,7 +249,7 @@ def forward(self, src, src_len, tgt, tgt_len):
         return loss / B
 
 
-def skyline_model_provider():
+def deepview_model_provider():
     args = get_args()
     vocab_size = 32317
     model_config = {
@@ -267,7 +267,7 @@ def skyline_model_provider():
     return model
 
 
-def skyline_input_provider(batch_size=64):
+def deepview_input_provider(batch_size=64):
     vocab_size = 32000
     src_len = 25
     tgt_len = 25
@@ -297,7 +297,7 @@ def skyline_input_provider(batch_size=64):
     return src, src_len_tensor, tgt, tgt_len_tensor
 
 
-def skyline_iteration_provider(model):
+def deepview_iteration_provider(model):
     args = get_args()
     opt_config = {
         'optimizer': args.optimizer,

examples/resnet/entry_point.py

@@ -4,18 +4,18 @@
 import resnet
 
 
-def skyline_model_provider():
+def deepview_model_provider():
     return resnet.resnet50().cuda()
 
 
-def skyline_input_provider(batch_size=16):
+def deepview_input_provider(batch_size=16):
     return (
         torch.randn((batch_size, 3, 224, 224)).cuda(),
         torch.randint(low=0, high=1000, size=(batch_size,)).cuda(),
     )
 
 
-def skyline_iteration_provider(model):
+def deepview_iteration_provider(model):
     optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
     loss_fn = torch.nn.CrossEntropyLoss()
     def iteration(inputs, targets):

examples/resnet/entry_point_resnext.py

@@ -4,18 +4,18 @@
 import resnet
 
 
-def skyline_model_provider():
+def deepview_model_provider():
     return resnet.resnext50_32x4d().cuda()
 
 
-def skyline_input_provider(batch_size=16):
+def deepview_input_provider(batch_size=16):
     return (
         torch.randn((batch_size, 3, 224, 224)).cuda(),
         torch.randint(low=0, high=1000, size=(batch_size,)).cuda(),
     )
 
 
-def skyline_iteration_provider(model):
+def deepview_iteration_provider(model):
     optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
     def iteration(*inputs):
         optimizer.zero_grad()

examples/testnet/entry_point.py

@@ -13,15 +13,15 @@ def forward(self, input):
         return self.testnet(input).sum()
 
 
-def skyline_model_provider():
+def deepview_model_provider():
     return TestNetWithLoss().cuda()
 
 
-def skyline_input_provider(batch_size=32):
+def deepview_input_provider(batch_size=32):
     return (torch.randn((batch_size, 3, 128, 128)).cuda(),)
 
 
-def skyline_iteration_provider(model):
+def deepview_iteration_provider(model):
     optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
     def iteration(*inputs):
         optimizer.zero_grad()

examples/transformer/entry_point.py

@@ -86,7 +86,7 @@ def model_config():
     return opt
 
 
-def skyline_model_provider():
+def deepview_model_provider():
     opt = model_config()
     return TransformerWithLoss(Transformer(
         opt.src_vocab_size,
@@ -104,7 +104,7 @@ def skyline_model_provider():
         dropout=opt.dropout)).cuda()
 
 
-def skyline_input_provider(batch_size=64):
+def deepview_input_provider(batch_size=64):
     vocab_size = 32000
     src_seq_len = 25
     tgt_seq_len = 25
@@ -141,7 +141,7 @@ def skyline_input_provider(batch_size=64):
     return source, src_pos, target, tgt_pos, gold
 
 
-def skyline_iteration_provider(transformer):
+def deepview_iteration_provider(transformer):
     opt = model_config()
     optimizer = ScheduledOptim(
         optim.Adam(

examples/vgg/entry_point.py

@@ -4,18 +4,18 @@
 import vgg
 
 
-def skyline_model_provider():
+def deepview_model_provider():
     return vgg.vgg11().cuda()
 
 
-def skyline_input_provider(batch_size=16):
+def deepview_input_provider(batch_size=16):
     return (
         torch.randn((batch_size, 3, 224, 224)).cuda(),
         torch.randint(low=0, high=1000, size=(batch_size,)).cuda(),
     )
 
 
-def skyline_iteration_provider(model):
+def deepview_iteration_provider(model):
     optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
     def iteration(*inputs):
         optimizer.zero_grad()

skyline/analysis/session.py

@@ -35,9 +35,9 @@
     habitat_found = False
 
 
-MODEL_PROVIDER_NAME = "skyline_model_provider"
-INPUT_PROVIDER_NAME = "skyline_input_provider"
-ITERATION_PROVIDER_NAME = "skyline_iteration_provider"
+MODEL_PROVIDER_NAME = "deepview_model_provider"
+INPUT_PROVIDER_NAME = "deepview_input_provider"
+ITERATION_PROVIDER_NAME = "deepview_iteration_provider"
 BATCH_SIZE_ARG = "batch_size"
 
 
@@ -355,7 +355,7 @@ def measure_throughput(self):
         )
         if len(samples) == 0 or samples[0].batch_size != self._batch_size:
             raise AnalysisError(
-                "Something went wrong with Skyline when measuring your "
+                "Something went wrong with DeepView.Profile when measuring your "
                 "model's throughput. Please file a bug."
             )