From efaee851b409cf4a728722f6a4c54768a6f6d661 Mon Sep 17 00:00:00 2001 From: p27 Date: Thu, 23 Jan 2025 19:37:56 +0800 Subject: [PATCH 1/3] Fix doc format. --- docs/source/conf.py | 2 +- docs/source/introduce/table_data_handle.rst | 14 +++++++------- docs/source/tutorial/gnns.rst | 4 ++-- docs/source/tutorial/rtls.rst | 4 ++-- docs/source/tutorial/tnns.rst | 2 +- 5 files changed, 13 insertions(+), 13 deletions(-) diff --git a/docs/source/conf.py b/docs/source/conf.py index 9fe97391..0c309aca 100644 --- a/docs/source/conf.py +++ b/docs/source/conf.py @@ -21,7 +21,7 @@ # -- Project information ----------------------------------------------------- project = 'rLLM' -copyright = '2024, Rllm Team' +copyright = '2024, rLLM Team' author = 'Zheng Wang, Weichen Li, Xiaotong Huang, Enze Zhang' version = '1.0' # The full version, including alpha/beta/rc tags diff --git a/docs/source/introduce/table_data_handle.rst b/docs/source/introduce/table_data_handle.rst index 852e8a8e..ceff2255 100644 --- a/docs/source/introduce/table_data_handle.rst +++ b/docs/source/introduce/table_data_handle.rst @@ -4,12 +4,12 @@ Table Data Handle Data Handling of Tables ----------------------- -A table contains many different columns with many different types. Each column type in Rllm is described by a certain semantic type, i.e., ColType. Rllm supports two basic column types so far: +A table contains many different columns with many different types. Each column type in rLLM is described by a certain semantic type, i.e., ColType. rLLM supports two basic column types so far: - :obj:`ColType.CATEGORICAL`: represent categorical or discrete data, such as grade levels in a student dataset and diabetes types in a diabetes dataset. - :obj:`ColType.NUMERICAL`: represent numerical or continuous data, such as such as temperature in a weather dataset and income in a salary dataset. -A table in Rllm is described by an instance of :class:`~rllm.data.table_data.TableData` with many default attributes: +A table in rLLM is described by an instance of :class:`~rllm.data.table_data.TableData` with many default attributes: - :obj:`df`: A `pandas.DataFrame`_ stores raw tabular data. - :obj:`col_types`: A dictionary indicating :class:`~rllm.types.ColType` of each column. @@ -73,10 +73,10 @@ A table in Rllm is described by an instance of :class:`~rllm.data.table_data.Tab dataset.y >>> tensor([0, 1, 1, ..., 0, 1, 0]) - dataset.stats_dict[ColType.CATEGORICAL][0] + dataset.stats_dict[ColType.CATEGORICAL][0] >>> {: 3, : 2, : 'Pclass'} - dataset.stats_dict[ColType.NUMERICAL][0] + dataset.stats_dict[ColType.NUMERICAL][0] >>> {: 29.69911766052246, : 80.0, : 0.41999998688697815, : 14.526496887207031, : [0.41999998688697815, 20.125, 28.0, 38.0, 80.0], : 'Age'} Also, an instance of :class:`~rllm.data.table_data.TableData` contains many basic properties: @@ -105,9 +105,9 @@ We support transferring the data in a :class:`~rllm.data.table_data.TableData` t Common Benchmark Datasets (Table Part) --------------------------------------- -Rllm contains a large number of common benchmark datasets. The list of all datasets are available in :mod:`~rllm.datasets`. Our dataset includes graph datasets and tabular datasets. We use tabular data for the demonstration. +rLLM contains a large number of common benchmark datasets. The list of all datasets are available in :mod:`~rllm.datasets`. Our dataset includes graph datasets and tabular datasets. We use tabular data for the demonstration. -Initializing tabular datasets is straightforward in Rllm. An initialization of a dataset will automatically download its raw files and process its columns. +Initializing tabular datasets is straightforward in rLLM. An initialization of a dataset will automatically download its raw files and process its columns. In the below example, we will use one of the pre-loaded datasets, containing the Titanic passengers. @@ -138,7 +138,7 @@ In the below example, we will use one of the pre-loaded datasets, containing the [5 rows x 11 columns] -Rllm also supports a custom dataset, so that you can use Rllm for your own problem. Assume you prepare your `pandas.DataFrame`_ as :obj:`df` with five columns: :obj:`cat1`, :obj:`cat2`, :obj:`num1`, :obj:`num2`, and :obj:`y`. Creating :class:`~rllm.data.table_data.TableData` object is very easy. +rLLM also supports a custom dataset, so that you can use rLLM for your own problem. Assume you prepare your `pandas.DataFrame`_ as :obj:`df` with five columns: :obj:`cat1`, :obj:`cat2`, :obj:`num1`, :obj:`num2`, and :obj:`y`. Creating :class:`~rllm.data.table_data.TableData` object is very easy. .. _pandas.DataFrame: http://pandas.pydata.org/pandas-docs/dev/reference/api/pandas.DataFrame.html#pandas.DataFrame diff --git a/docs/source/tutorial/gnns.rst b/docs/source/tutorial/gnns.rst index f92fa939..50741d8c 100644 --- a/docs/source/tutorial/gnns.rst +++ b/docs/source/tutorial/gnns.rst @@ -3,7 +3,7 @@ Design of GNNs What is a GNN? ---------------- -In machine learning, **Graph Neural Networks (GNNs)** are a class of neural networks specifically designed to process graph-structured data. In a GNN, the input is represented as a graph, where nodes (vertices) correspond to entities and edges represent the relationships or interactions between these entities. A typical GNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in *Understanding Transform* and *Understanding Convolution*. +In machine learning, **Graph Neural Networks (GNNs)** are a class of neural networks specifically designed to process graph-structured data. In a GNN, the input is represented as a graph, where nodes (vertices) correspond to entities and edges represent the relationships or interactions between these entities. A typical GNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in `Understanding Transform ` and :doc:`Understanding Convolution `. Construct a GCN @@ -119,4 +119,4 @@ Finally, we need to implement a :obj:`train()` function and a :obj:`test()` func print(f"Mean time per epoch: {torch.tensor(times).mean():.4f}s") print(f"Total time: {sum(times):.4f}s") - print(f"Best test acc: {best_test_acc:.4f}") \ No newline at end of file + print(f"Best test acc: {best_test_acc:.4f}") diff --git a/docs/source/tutorial/rtls.rst b/docs/source/tutorial/rtls.rst index 35514b6f..2fa3ee68 100644 --- a/docs/source/tutorial/rtls.rst +++ b/docs/source/tutorial/rtls.rst @@ -3,7 +3,7 @@ Design of RTLs What is a RTL? ---------------- -In machine learning, **Relational Table Learnings (RTLs)** typically refers to the learning of relational table data, which consists of multiple interconnected tables with significant heterogeneity. In an RTL, the input comprises multiple table signals that are interrelated. A typical RTL architecture consists of one or more Transforms followed by multiple Convolution layers, as detailed in *Understanding Transform* and *Understanding Convolution*. +In machine learning, **Relational Table Learnings (RTLs)** typically refers to the learning of relational table data, which consists of multiple interconnected tables with significant heterogeneity. In an RTL, the input comprises multiple table signals that are interrelated. A typical RTL architecture consists of one or more Transforms followed by multiple Convolution layers, as detailed in `Understanding Transform ` and :doc:`Understanding Convolution `. Construct a BRIDGE @@ -164,4 +164,4 @@ Finally, we need to implement a :obj:`train()` function and a :obj:`test()` func "BRIDGE result: " f"Best Val acc: {best_val_acc:.4f}, " f"Best Test acc: {best_test_acc:.4f}" - ) \ No newline at end of file + ) diff --git a/docs/source/tutorial/tnns.rst b/docs/source/tutorial/tnns.rst index 583805d7..a8b478d7 100644 --- a/docs/source/tutorial/tnns.rst +++ b/docs/source/tutorial/tnns.rst @@ -2,7 +2,7 @@ Design of TNNs =============== What is a TNN? ---------------- -In machine learning, **Table/Tabular Neural Networks (TNNs)** are recently emerging neural networks specifically designed to process tabular data. In a TNN, the input is structured tabular data, usually organized in rows and columns. A typical TNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in *Understanding Transform* and *Understanding Convolution*. +In machine learning, **Table/Tabular Neural Networks (TNNs)** are recently emerging neural networks specifically designed to process tabular data. In a TNN, the input is structured tabular data, usually organized in rows and columns. A typical TNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in `Understanding Transform ` and :doc:`Understanding Convolution `. Construct a TabTransformer From d3cb699b59850fd33eb82f6ec3c5454a44511269 Mon Sep 17 00:00:00 2001 From: p27 Date: Thu, 23 Jan 2025 19:39:50 +0800 Subject: [PATCH 2/3] Fix doc format. --- docs/source/tutorial/gnns.rst | 2 +- docs/source/tutorial/rtls.rst | 2 +- docs/source/tutorial/tnns.rst | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/source/tutorial/gnns.rst b/docs/source/tutorial/gnns.rst index 50741d8c..df446a9b 100644 --- a/docs/source/tutorial/gnns.rst +++ b/docs/source/tutorial/gnns.rst @@ -3,7 +3,7 @@ Design of GNNs What is a GNN? ---------------- -In machine learning, **Graph Neural Networks (GNNs)** are a class of neural networks specifically designed to process graph-structured data. In a GNN, the input is represented as a graph, where nodes (vertices) correspond to entities and edges represent the relationships or interactions between these entities. A typical GNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in `Understanding Transform ` and :doc:`Understanding Convolution `. +In machine learning, **Graph Neural Networks (GNNs)** are a class of neural networks specifically designed to process graph-structured data. In a GNN, the input is represented as a graph, where nodes (vertices) correspond to entities and edges represent the relationships or interactions between these entities. A typical GNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in :doc:`Understanding Transform ` and :doc:`Understanding Convolution `. Construct a GCN diff --git a/docs/source/tutorial/rtls.rst b/docs/source/tutorial/rtls.rst index 2fa3ee68..54d8b8e4 100644 --- a/docs/source/tutorial/rtls.rst +++ b/docs/source/tutorial/rtls.rst @@ -3,7 +3,7 @@ Design of RTLs What is a RTL? ---------------- -In machine learning, **Relational Table Learnings (RTLs)** typically refers to the learning of relational table data, which consists of multiple interconnected tables with significant heterogeneity. In an RTL, the input comprises multiple table signals that are interrelated. A typical RTL architecture consists of one or more Transforms followed by multiple Convolution layers, as detailed in `Understanding Transform ` and :doc:`Understanding Convolution `. +In machine learning, **Relational Table Learnings (RTLs)** typically refers to the learning of relational table data, which consists of multiple interconnected tables with significant heterogeneity. In an RTL, the input comprises multiple table signals that are interrelated. A typical RTL architecture consists of one or more Transforms followed by multiple Convolution layers, as detailed in :doc:`Understanding Transform ` and :doc:`Understanding Convolution `. Construct a BRIDGE diff --git a/docs/source/tutorial/tnns.rst b/docs/source/tutorial/tnns.rst index a8b478d7..7e873e73 100644 --- a/docs/source/tutorial/tnns.rst +++ b/docs/source/tutorial/tnns.rst @@ -2,7 +2,7 @@ Design of TNNs =============== What is a TNN? ---------------- -In machine learning, **Table/Tabular Neural Networks (TNNs)** are recently emerging neural networks specifically designed to process tabular data. In a TNN, the input is structured tabular data, usually organized in rows and columns. A typical TNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in `Understanding Transform ` and :doc:`Understanding Convolution `. +In machine learning, **Table/Tabular Neural Networks (TNNs)** are recently emerging neural networks specifically designed to process tabular data. In a TNN, the input is structured tabular data, usually organized in rows and columns. A typical TNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in :doc:`Understanding Transform ` and :doc:`Understanding Convolution `. Construct a TabTransformer From 2b443517aac59c5c9b41cad745a5f6ae44a93273 Mon Sep 17 00:00:00 2001 From: p27 Date: Thu, 23 Jan 2025 21:23:32 +0800 Subject: [PATCH 3/3] Fix doc format. --- docs/source/tutorial/gnns.rst | 4 ++-- docs/source/tutorial/rtls.rst | 4 ++-- docs/source/tutorial/tnns.rst | 4 ++-- 3 files changed, 6 insertions(+), 6 deletions(-) diff --git a/docs/source/tutorial/gnns.rst b/docs/source/tutorial/gnns.rst index df446a9b..53f194c7 100644 --- a/docs/source/tutorial/gnns.rst +++ b/docs/source/tutorial/gnns.rst @@ -1,9 +1,9 @@ Design of GNNs =============== -What is a GNN? +What is GNN? ---------------- -In machine learning, **Graph Neural Networks (GNNs)** are a class of neural networks specifically designed to process graph-structured data. In a GNN, the input is represented as a graph, where nodes (vertices) correspond to entities and edges represent the relationships or interactions between these entities. A typical GNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in :doc:`Understanding Transform ` and :doc:`Understanding Convolution `. +In machine learning, **Graph Neural Networks (GNNs)** are a class of neural networks specifically designed to process graph-structured data. In a GNN, the input is represented as a graph, where nodes (vertices) correspond to entities and edges represent the relationships or interactions between these entities. A typical GNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in :doc:`Understanding Transforms ` and :doc:`Understanding Convolutions `. Construct a GCN diff --git a/docs/source/tutorial/rtls.rst b/docs/source/tutorial/rtls.rst index 54d8b8e4..8824a473 100644 --- a/docs/source/tutorial/rtls.rst +++ b/docs/source/tutorial/rtls.rst @@ -1,9 +1,9 @@ Design of RTLs ============== -What is a RTL? +What is RTL? ---------------- -In machine learning, **Relational Table Learnings (RTLs)** typically refers to the learning of relational table data, which consists of multiple interconnected tables with significant heterogeneity. In an RTL, the input comprises multiple table signals that are interrelated. A typical RTL architecture consists of one or more Transforms followed by multiple Convolution layers, as detailed in :doc:`Understanding Transform ` and :doc:`Understanding Convolution `. +In machine learning, **Relational Table Learnings (RTLs)** typically refers to the learning of relational table data, which consists of multiple interconnected tables with significant heterogeneity. In an RTL, the input comprises multiple table signals that are interrelated. A typical RTL architecture consists of one or more Transforms followed by multiple Convolution layers, as detailed in :doc:`Understanding Transforms ` and :doc:`Understanding Convolutions `. Construct a BRIDGE diff --git a/docs/source/tutorial/tnns.rst b/docs/source/tutorial/tnns.rst index 7e873e73..9ea3c452 100644 --- a/docs/source/tutorial/tnns.rst +++ b/docs/source/tutorial/tnns.rst @@ -1,8 +1,8 @@ Design of TNNs =============== -What is a TNN? +What is TNN? ---------------- -In machine learning, **Table/Tabular Neural Networks (TNNs)** are recently emerging neural networks specifically designed to process tabular data. In a TNN, the input is structured tabular data, usually organized in rows and columns. A typical TNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in :doc:`Understanding Transform ` and :doc:`Understanding Convolution `. +In machine learning, **Table/Tabular Neural Networks (TNNs)** are recently emerging neural networks specifically designed to process tabular data. In a TNN, the input is structured tabular data, usually organized in rows and columns. A typical TNN architecture consists of an initial Transform followed by multiple Convolution layers, as detailed in :doc:`Understanding Transforms ` and :doc:`Understanding Convolutions `. Construct a TabTransformer