nit fixes to workflow blueprints (#585)

* nit: make spacing consistent

* fix num_records for one blueprint

* change hyphen to underscore in task names

Author: nina-xu

config_templates/gretel/tasks/tabular_ft__default.yaml

@@ -3,40 +3,38 @@ name: default
 task:
   name: tabular_ft
   config:
-      train:
-        # Optionally group records by the column(s) set below.
-        # This is useful if you need to maintain correlations
-        # across multiple records. Otherwise, the training
-        # assumes records are independent.
-        group_training_examples_by: null
+    train:
+      # Optionally group records by the column(s) set below.
+      # This is useful if you need to maintain correlations
+      # across multiple records. Otherwise, the training
+      # assumes records are independent.
+      group_training_examples_by: null
 
-        # Optionally order records by the column set below.
-        # This is useful if your records are sequential.
-        # Note that this parameter can only be used when
-        # your records are grouped using the above parameter.
-        order_training_examples_by: null
+      # Optionally order records by the column set below.
+      # This is useful if your records are sequential.
+      # Note that this parameter can only be used when
+      # your records are grouped using the above parameter.
+      order_training_examples_by: null
 
-        params:
-          # The parameter below is a proxy for training time.
-          # If set to 'auto', we will automatically choose an
-          # appropriate value. An integer value will set the
-          # number of records from the input dataset that the
-          # model will see during training. It can be smaller
-          # (we downsample), larger (we resample), or the same
-          # size as your input dataset. A starting value to
-          # experiment with is 25,000.
-          num_input_records_to_sample: auto
-
-          # Scale the base LLM's context length by this factor
-          # using RoPE scaling to handle datasets with more
-          # columns, or datasets containing groups with more
-          # than a few records. You can try increasing the
-          # rope_scaling_factor (you could first try the value 2)
-          # if you hit an error for maximum tokens. It must be
-          # an integer value. The default is 1 and maximum is 6. 
-          rope_scaling_factor: 1
-
-      generate:
-        num_records: 1000
+      params:
+        # The parameter below is a proxy for training time.
+        # If set to 'auto', we will automatically choose an
+        # appropriate value. An integer value will set the
+        # number of records from the input dataset that the
+        # model will see during training. It can be smaller
+        # (we downsample), larger (we resample), or the same
+        # size as your input dataset. A starting value to
+        # experiment with is 25,000.
+        num_input_records_to_sample: auto
 
+        # Scale the base LLM's context length by this factor
+        # using RoPE scaling to handle datasets with more
+        # columns, or datasets containing groups with more
+        # than a few records. You can try increasing the
+        # rope_scaling_factor (you could first try the value 2)
+        # if you hit an error for maximum tokens. It must be
+        # an integer value. The default is 1 and maximum is 6. 
+        rope_scaling_factor: 1
 
+    generate:
+      num_records: 1000

config_templates/gretel/tasks/tabular_ft__differential_privacy.yaml

@@ -3,55 +3,53 @@ name: differential_privacy
 task:
   name: tabular_ft
   config:
-      train:
-        # Optionally group records by the column(s) set below.
-        # This is useful if you need to maintain correlations
-        # across multiple records. Otherwise, the training
-        # assumes records are independent.
-        group_training_examples_by: null
-
-        # Optionally order records by the column set below.
-        # This is useful if your records are sequential.
-        # Note that this parameter can only be used when
-        # your records are grouped using the above parameter.
-        order_training_examples_by: null
-
-        privacy_params:
-            dp: true
-
-            # Defines the privacy budget - the larger the value, the
-            # less privacy we get. A value between 2 and 8 is deemed
-            # reasonable, usually.
-            epsilon: 8
-
-        params:
-          # The parameter below is a proxy for training time.
-          # If set to 'auto', we will automatically choose an
-          # appropriate value. An integer value will set the
-          # number of records from the input dataset that the
-          # model will see during training. It can be smaller
-          # (we downsample), larger (we resample), or the same
-          # size as your input dataset. A starting value to
-          # experiment with is 25,000.
-          num_input_records_to_sample: auto
-
-          # Scale the base LLM's context length by this factor
-          # using RoPE scaling to handle datasets with more
-          # columns, or datasets containing groups with more
-          # than a few records. You can try increasing the
-          # rope_scaling_factor (you could first try the value 2)
-          # if you hit an error for maximum tokens. It must be
-          # an integer value. The default is 1 and maximum is 6. 
-          rope_scaling_factor: 1
-
-          # You can try increasing this until you run out-of-memory.
-          batch_size: 4
-
-      generate:
-        num_records: 1000
-
-        # With DP, enabling structured generation can help with
-        # increasing the percentage of valid records.
-        use_structured_generation: true
-
-
+    train:
+      # Optionally group records by the column(s) set below.
+      # This is useful if you need to maintain correlations
+      # across multiple records. Otherwise, the training
+      # assumes records are independent.
+      group_training_examples_by: null
+
+      # Optionally order records by the column set below.
+      # This is useful if your records are sequential.
+      # Note that this parameter can only be used when
+      # your records are grouped using the above parameter.
+      order_training_examples_by: null
+
+      privacy_params:
+        dp: true
+
+        # Defines the privacy budget - the larger the value, the
+        # less privacy we get. A value between 2 and 8 is deemed
+        # reasonable, usually.
+        epsilon: 8
+
+      params:
+        # The parameter below is a proxy for training time.
+        # If set to 'auto', we will automatically choose an
+        # appropriate value. An integer value will set the
+        # number of records from the input dataset that the
+        # model will see during training. It can be smaller
+        # (we downsample), larger (we resample), or the same
+        # size as your input dataset. A starting value to
+        # experiment with is 25,000.
+        num_input_records_to_sample: auto
+
+        # Scale the base LLM's context length by this factor
+        # using RoPE scaling to handle datasets with more
+        # columns, or datasets containing groups with more
+        # than a few records. You can try increasing the
+        # rope_scaling_factor (you could first try the value 2)
+        # if you hit an error for maximum tokens. It must be
+        # an integer value. The default is 1 and maximum is 6. 
+        rope_scaling_factor: 1
+
+        # You can try increasing this until you run out-of-memory.
+        batch_size: 4
+
+    generate:
+      num_records: 1000
+
+      # With DP, enabling structured generation can help with
+      # increasing the percentage of valid records.
+      use_structured_generation: true