Upload blackbox yahpo-fcnet

yahpo/fcnet/best_params_resnet.json

@@ -0,0 +1 @@
+{"d": 512, "d_hidden_factor": 2.7803277401283846, "hidden_dropout": 0.04856105544361067, "lr": 0.00038735009551193095, "mixup": true, "n_layers": 4, "opt_tfms_batch_size": true, "opt_tfms_dropout_1": true, "opt_tfms_dropout_2": false, "opt_tfms_epoch": false, "opt_tfms_init_lr": true, "opt_tfms_n_params": false, "opt_tfms_n_units_1": true, "opt_tfms_n_units_2": true, "opt_tfms_replication": true, "opt_tfms_runtime": false, "opt_tfms_valid_loss": false, "opt_tfms_valid_mse": false, "tfms_batch_size": "tlog", "tfms_dropout_1": "tnexp", "tfms_init_lr": "tnexp", "tfms_n_units_1": "tlog", "tfms_n_units_2": "tlog", "tfms_replication": "tlog", "use_residual_dropout": false}

yahpo/fcnet/config_space.json

@@ -0,0 +1,114 @@
+{
+  "hyperparameters": [
+    {
+      "name": "task",
+      "type": "categorical",
+      "choices": [
+        "fcnet_naval_propulsion",
+        "fcnet_protein_structure",
+        "fcnet_slice_localization",
+        "fcnet_parkinsons_telemonitoring"
+      ],
+      "default": "fcnet_naval_propulsion",
+      "probabilities": null
+    },
+    {
+      "name": "activation_fn_1",
+      "type": "categorical",
+      "choices": [
+        "relu",
+        "tanh"
+      ],
+      "default": "relu",
+      "probabilities": null
+    },
+    {
+      "name": "activation_fn_2",
+      "type": "categorical",
+      "choices": [
+        "tanh",
+        "relu"
+      ],
+      "default": "relu",
+      "probabilities": null
+    },
+    {
+      "name": "lr_schedule",
+      "type": "categorical",
+      "choices": [
+        "const",
+        "cosine"
+      ],
+      "default": "const",
+      "probabilities": null
+    },
+    {
+      "name": "epoch",
+      "type": "uniform_int",
+      "log": true,
+      "lower": 1,
+      "upper": 100,
+      "default": 100
+    },
+    {
+      "name": "replication",
+      "type": "uniform_int",
+      "log": true,
+      "lower": 1,
+      "upper": 4,
+      "default": 1
+    },
+    {
+      "name": "batch_size",
+      "type": "uniform_int",
+      "log": true,
+      "lower": 8,
+      "upper": 64,
+      "default": 64
+    },
+    {
+      "name": "init_lr",
+      "type": "uniform_float",
+      "log": true,
+      "lower": 5e-4,
+      "upper": 1e-1,
+      "default": 1e-3
+    },
+    {
+      "name": "dropout_1",
+      "type": "uniform_float",
+      "log": false,
+      "lower": 0.0,
+      "upper": 0.6,
+      "default": 0.5
+    },
+    {
+      "name": "dropout_2",
+      "type": "uniform_float",
+      "log": false,
+      "lower": 0.0,
+      "upper": 0.6,
+      "default": 0.5
+    },
+    {
+      "name": "n_units_1",
+      "type": "uniform_int",
+      "log": false,
+      "lower": 16,
+      "upper": 512,
+      "default": 256
+    },
+    {
+      "name": "n_units_2",
+      "type": "uniform_int",
+      "log": false,
+      "lower": 16,
+      "upper": 512,
+      "default": 256
+    }
+  ],
+  "conditions": [],
+  "forbiddens": [],
+  "python_module_version": "0.4.18",
+  "json_format_version": 0.2
+}

yahpo/fcnet/encoding.json

@@ -0,0 +1 @@
+{"activation_fn_1": {"#na#": 0, "relu": 1, "tanh": 2}, "activation_fn_2": {"#na#": 0, "relu": 1, "tanh": 2}, "lr_schedule": {"#na#": 0, "const": 1, "cosine": 2}, "task": {"#na#": 0, "fcnet_naval_propulsion": 1, "fcnet_parkinsons_telemonitoring": 2, "fcnet_protein_structure": 3, "fcnet_slice_localization": 4}}

yahpo/fcnet/metadata.json

@@ -0,0 +1 @@
+{"metric_elapsed_time": "time", "metric_default": "val_accuracy", "resource_attr": "st_worker_iter"}

yahpo/fcnet/model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ba8114fec92c3c01a36b501c98dba160e6dfbaa98b48b7f20a6b255c9978f5f
+size 35160441

yahpo/fcnet/param_set.R

@@ -0,0 +1,36 @@
+search_space = ps(
+  task = p_fct(levels = c("fcnet_naval_propulsion", "fcnet_protein_structure", "fcnet_slice_localization","fcnet_parkinsons_telemonitoring"), tags = "task_id"),
+  activation_fn_1 = p_fct(levels = c("relu", "tanh")),
+  activation_fn_2 = p_fct(levels = c("relu", "tanh")),
+  lr_schedule = p_fct(levels = c("const", "cosine")),
+  epoch = p_dbl(lower = log(1), upper = log(100), tags = c("int", "log"), trafo = function(x) as.integer(round(exp(x))), tags = "budget"),
+  replication = p_int(lower = 1, upper = 4),
+  batch_size = p_dbl(lower = log(8), upper = log(64), tags = c("int", "log"), trafo = function(x) as.integer(round(exp(x)))),
+  init_lr = p_dbl(lower = log(5e-4), upper = log(1e-1), tags = "log", trafo = function(x) exp(x)),
+  dropout_1 = p_dbl(lower = 0, upper = 0.6), 
+  dropout_2 = p_dbl(lower = 0, upper = 0.6), 
+  n_units_1 = p_int(lower = 16L, upper= 512L),
+  n_units_2 = p_int(lower = 16L, upper= 512L)
+)
+                     
+domain = ps(
+  task = p_fct(levels = c("fcnet_naval_propulsion", "fcnet_protein_structure", "fcnet_slice_localization","fcnet_parkinsons_telemonitoring"), tags = "task_id"),
+  activation_fn_1 = p_fct(levels = c("relu", "tanh")),
+  activation_fn_2 = p_fct(levels = c("relu", "tanh")),
+  lr_schedule = p_fct(levels = c("const", "cosine")),
+  epoch = p_int(lower = 1L, upper = 100L, tags = "budget"),
+  replication = p_int(lower = 1, upper = 4),
+  batch_size = p_int(lower = 8, upper = 64),
+  init_lr = p_dbl(lower = 5e-4, upper = 1e-1),
+  dropout_1 = p_dbl(lower = 0, upper = 0.6), 
+  dropout_2 = p_dbl(lower = 0, upper = 0.6), 
+  n_units_1 = p_int(lower = 16L, upper= 512L),
+  n_units_2 = p_int(lower = 16L, upper= 512L)
+)
+                     
+codomain = ps(
+  valid_loss = p_dbl(lower = 0, upper = Inf, tags = "minimize"),
+  valid_mse = p_dbl(lower = 0, upper = Inf, tags = "minimize"),
+  runtime = p_dbl(lower = 0, upper = Inf, tags = "minimize"),
+  n_params = p_int(lower = 1, upper = Inf, tags = "minimize")
+)