[Docs] Fix outdated source install and API reference docs

docs/README.md

@@ -42,7 +42,18 @@ This folder contains the source of TVM's documentation, hosted at https://tvm.ap
 
 ### Native
 
-1. [Build TVM](https://tvm.apache.org/docs/install/from_source.html) first in the repo root folder
+1. [Build TVM](https://tvm.apache.org/docs/install/from_source.html) first in the repo root folder, then make it importable:
+
+   ```bash
+   export TVM_HOME=/path-to-tvm
+   export TVM_LIBRARY_PATH=$TVM_HOME/build
+   pip install --target=$TVM_HOME/python $TVM_HOME/3rdparty/tvm-ffi
+   export PYTHONPATH=$TVM_HOME/python:$PYTHONPATH
+   ```
+
+   `docs/conf.py` unconditionally imports `tvm` at startup, so the build will
+   fail immediately if TVM is not importable.
+
 2. Install dependencies
 
    ```bash
@@ -92,15 +103,25 @@ python tests/scripts/ci.py docs --tutorial-pattern=file_name\.py
 
 ## Helper Scripts
 
-You can run the following script to reproduce the CI sphinx pre-check stage.
-This script skips the tutorial executions and is useful to quickly check the content.
+The following script mirrors the CI docs pipeline: it runs a sphinx pre-check
+(when not running locally) and then performs a full `make htmldepoly` build,
+including tutorial execution. You will need a GPU CI environment.
 
 ```bash
 tests/scripts/task_python_docs.sh
 ```
 
-The following script runs the full build which includes tutorial executions.
-You will need a GPU CI environment.
+To build docs locally without executing tutorials (fastest local iteration):
+
+```bash
+cd docs && TVM_TUTORIAL_EXEC_PATTERN=none make html
+```
+
+Note: the sphinx pre-check (warning validation) only runs in CI (`IS_LOCAL=0`).
+`python tests/scripts/ci.py docs` always sets `IS_LOCAL=1` and skips the
+pre-check regardless of other flags.
+
+To run the full build including tutorial executions:
 
 ```bash
 python tests/scripts/ci.py docs --full

docs/conf.py

@@ -240,20 +240,18 @@ def install_request_hook(gallery_conf, fname):
 pip install apache-tvm=={version}"""
 
 INSTALL_TVM_CUDA_DEV = """\
-%%shell
-# Installs the latest dev build of TVM from PyPI, with CUDA enabled. To use this,
-# you must request a Google Colab instance with a GPU by going to Runtime ->
-# Change runtime type -> Hardware accelerator -> GPU. If you wish to build from
-# source, see https://tvm.apache.org/docs/install/from_source.html
-pip install tlcpack-nightly-cu113 --pre -f https://tlcpack.ai/wheels"""
+%%markdown
+> **Note:** This tutorial requires a CUDA-enabled build of TVM.
+> Pre-built CUDA wheels are not currently available on PyPI.
+> Please build TVM from source with CUDA enabled before running this notebook:
+> https://tvm.apache.org/docs/install/from_source.html"""
 
 INSTALL_TVM_CUDA_FIXED = f"""\
-%%shell
-# Installs TVM version {version} from PyPI, with CUDA enabled. To use this,
-# you must request a Google Colab instance with a GPU by going to Runtime ->
-# Change runtime type -> Hardware accelerator -> GPU. If you wish to build from
-# source, see https://tvm.apache.org/docs/install/from_source.html
-pip install apache-tvm-cu113=={version} --no-index  -f https://tlcpack.ai/wheels"""
+%%markdown
+> **Note:** This tutorial requires a CUDA-enabled build of TVM (version {version}).
+> Pre-built CUDA wheels are not currently available on PyPI.
+> Please build TVM from source with CUDA enabled before running this notebook:
+> https://tvm.apache.org/docs/install/from_source.html"""
 
 
 @monkey_patch("sphinx_gallery.gen_rst", "jupyter_notebook")
@@ -552,6 +550,13 @@ def fixup_tutorials(original_url: str) -> str:
         "v0.14.0/",
         "v0.15.0/",
         "v0.16.0/",
+        "v0.17.0/",
+        "v0.18.0/",
+        "v0.19.0/",
+        "v0.20.0/",
+        "v0.21.0/",
+        "v0.22.0/",
+        "v0.23.0/",
     ],
     "display_github": True,
     "github_user": "apache",

docs/contribute/document.rst

@@ -222,7 +222,8 @@ existing environment to demonstrate the usage.
 
 If you add a new categorization of how-to, you will need to add references to
 `conf.py <https://github.com/apache/tvm/tree/main/docs/conf.py>`_ and the
-`how-to index <https://github.com/apache/tvm/tree/main/docs/how_to/dev/index.rst>`_
+`top-level docs index <https://github.com/apache/tvm/tree/main/docs/index.rst>`_
+(how-to entries are registered there directly, not in a separate how-to index).
 
 Refer to Another Location in the Document
 -----------------------------------------

docs/install/from_source.rst

@@ -72,12 +72,13 @@ one may simply use:
     conda activate tvm-build-venv
 
 .. note::
-   **For Frontend Contributors (TFLite):** If you plan to run or contribute to the frontend tests (e.g., ``test_frontend_tflite.py``), you must strictly use **Python 3.10**.
-
-   The TFLite tests currently require ``tensorflow==2.9.0`` and ``numpy==1.26.4`` to prevent ``_ARRAY_API`` core dumps.
-   Because TensorFlow 2.9.0 does not provide pre-compiled binaries for Python 3.11 or newer,
-   setting up your environment with Python 3.11+ will force incompatible pip upgrades
-   and cause C++ ABI crashes during testing.
+   **For Frontend Contributors (TFLite):** If you plan to run or contribute to the frontend tests (e.g., ``test_frontend_tflite.py``), you must install ``tensorflow==2.19.0``.
+
+   .. code:: bash
+
+       pip install tensorflow==2.19.0
+
+   Python 3.11 is supported.
 
 Step 2. Get Source from GitHub
 ------------------------------
@@ -163,8 +164,8 @@ Leaving the build environment ``tvm-build-venv``, there are two ways to install
 .. code-block:: bash
 
     export TVM_HOME=/path-to-tvm
-    pip install $TVM_HOME/3rdparty/tvm-ffi
-    export PYTHONPATH=$TVM_HOME/python:$TVM_HOME/3rdparty/tvm-ffi/python:$PYTHONPATH
+    pip install --target=$TVM_HOME/python $TVM_HOME/3rdparty/tvm-ffi
+    export PYTHONPATH=$TVM_HOME/python:$PYTHONPATH
 
 - Install via pip local project
 
@@ -173,7 +174,7 @@ Leaving the build environment ``tvm-build-venv``, there are two ways to install
     conda activate your-own-env
     conda install python # make sure python is installed
     export TVM_LIBRARY_PATH=/path-to-tvm/build
-    pip install -e /path-to-tvm/python
+    pip install -e /path-to-tvm
 
 Step 4. Validate Installation
 -----------------------------

docs/reference/api/python/contrib.rst

@@ -37,24 +37,66 @@ tvm.contrib.cc
     :members:
 
 
+tvm.contrib.coreml_runtime
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.coreml_runtime
+    :members:
+
+
 tvm.contrib.cublas
 ~~~~~~~~~~~~~~~~~~
 .. automodule:: tvm.contrib.cublas
     :members:
 
 
+tvm.contrib.cublaslt
+~~~~~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.cublaslt
+    :members:
+
+
+tvm.contrib.cudnn
+~~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.cudnn
+    :members:
+
+
 tvm.contrib.dlpack
 ~~~~~~~~~~~~~~~~~~
 .. automodule:: tvm.contrib.dlpack
     :members:
 
 
+tvm.contrib.dnnl
+~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.dnnl
+    :members:
+
+
+tvm.contrib.download
+~~~~~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.download
+    :members:
+
+
 tvm.contrib.emcc
 ~~~~~~~~~~~~~~~~
 .. automodule:: tvm.contrib.emcc
     :members:
 
 
+tvm.contrib.hipblas
+~~~~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.hipblas
+    :members:
+
+
+tvm.contrib.mkl
+~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.mkl
+    :members:
+
+
 tvm.contrib.ndk
 ~~~~~~~~~~~~~~~
 .. automodule:: tvm.contrib.ndk
@@ -79,6 +121,12 @@ tvm.contrib.pickle_memoize
     :members:
 
 
+tvm.contrib.popen_pool
+~~~~~~~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.popen_pool
+    :members:
+
+
 tvm.contrib.random
 ~~~~~~~~~~~~~~~~~~
 .. automodule:: tvm.contrib.random
@@ -103,6 +151,18 @@ tvm.contrib.tar
     :members:
 
 
+tvm.contrib.thrust
+~~~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.thrust
+    :members:
+
+
+tvm.contrib.tvmjs
+~~~~~~~~~~~~~~~~~
+.. automodule:: tvm.contrib.tvmjs
+    :members:
+
+
 tvm.contrib.utils
 ~~~~~~~~~~~~~~~~~
 .. automodule:: tvm.contrib.utils

docs/reference/api/python/index.rst

@@ -31,6 +31,7 @@ Python API
     driver
     testing
     exec
+    support
 
 .. toctree::
     :maxdepth: 1

docs/reference/api/python/support.rst

@@ -0,0 +1,23 @@
+..  Licensed to the Apache Software Foundation (ASF) under one
+    or more contributor license agreements.  See the NOTICE file
+    distributed with this work for additional information
+    regarding copyright ownership.  The ASF licenses this file
+    to you under the Apache License, Version 2.0 (the
+    "License"); you may not use this file except in compliance
+    with the License.  You may obtain a copy of the License at
+
+..    http://www.apache.org/licenses/LICENSE-2.0
+
+..  Unless required by applicable law or agreed to in writing,
+    software distributed under the License is distributed on an
+    "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+    KIND, either express or implied.  See the License for the
+    specific language governing permissions and limitations
+    under the License.
+
+tvm.support
+-----------
+.. automodule:: tvm.support
+   :members:
+   :imported-members:
+   :autosummary:

python/tvm/contrib/coreml_runtime.py

@@ -23,6 +23,7 @@
 
 def create(symbol, compiled_model_path, device):
     """Create a runtime executor module given a coreml model and context.
+
     Parameters
     ----------
     symbol : str

python/tvm/contrib/cudnn.py

@@ -64,50 +64,20 @@ def algo_to_index(algo_type, algo_name):
 
     Parameters
     ----------
-        algo_type : str
-            ["fwd", "bwd_filter", "bwd_data]
-
-        algo_name : str
-            algorithm name in cudnn definition
-            fwd = [
-                "CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_GEMM",
-                "CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_PRECOMP_GEMM",
-                "CUDNN_CONVOLUTION_FWD_ALGO_GEMM",
-                "CUDNN_CONVOLUTION_FWD_ALGO_DIRECT",
-                "CUDNN_CONVOLUTION_FWD_ALGO_FFT",
-                "CUDNN_CONVOLUTION_FWD_ALGO_FFT_TILING",
-                "CUDNN_CONVOLUTION_FWD_ALGO_WINOGRAD",
-                "CUDNN_CONVOLUTION_FWD_ALGO_WINOGRAD_NONFUSED",
-                "CUDNN_CONVOLUTION_FWD_ALGO_COUNT",
-            ]
-            bwd_filter = [
-                "CUDNN_CONVOLUTION_BWD_FILTER_ALGO_0",
-                # non-deterministic
-                "CUDNN_CONVOLUTION_BWD_FILTER_ALGO_1",
-                "CUDNN_CONVOLUTION_BWD_FILTER_ALGO_FFT",
-                "CUDNN_CONVOLUTION_BWD_FILTER_ALGO_3",
-                # non-deterministic, algo0 with workspaceS
-                "CUDNN_CONVOLUTION_BWD_FILTER_ALGO_WINOGRAD",
-                # not implemented
-                "CUDNN_CONVOLUTION_BWD_FILTER_ALGO_WINOGRAD_NONFUSED",
-                "CUDNN_CONVOLUTION_BWD_FILTER_ALGO_FFT_TILING",
-                "CUDNN_CONVOLUTION_BWD_FILTER_ALGO_COUNT",
-            ]
-            bwd_data = [
-                "CUDNN_CONVOLUTION_BWD_DATA_ALGO_0",
-                # non-deterministic
-                "CUDNN_CONVOLUTION_BWD_DATA_ALGO_1",
-                "CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT",
-                "CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT_TILING",
-                "CUDNN_CONVOLUTION_BWD_DATA_ALGO_WINOGRAD",
-                "CUDNN_CONVOLUTION_BWD_DATA_ALGO_WINOGRAD_NONFUSED",
-                "CUDNN_CONVOLUTION_BWD_DATA_ALGO_COUNT",
-            ]
+    algo_type : str
+        One of ``"fwd"``, ``"bwd_filter"``, or ``"bwd_data"``.
+
+    algo_name : str
+        Algorithm name as defined in cuDNN. For example:
+
+        * fwd: ``CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_GEMM``, etc.
+        * bwd_filter: ``CUDNN_CONVOLUTION_BWD_FILTER_ALGO_0``, etc.
+        * bwd_data: ``CUDNN_CONVOLUTION_BWD_DATA_ALGO_0``, etc.
 
     Returns
     -------
-        algo: int
-            algorithm index
+    algo : int
+        Algorithm index
 
     """
     idx = -1

python/tvm/contrib/popen_pool.py

@@ -213,7 +213,7 @@ def is_alive(self):
         return False
 
     def send(self, fn, args=(), kwargs=None, timeout=None):
-        """Send a new function task fn(*args, **kwargs) to the subprocess.
+        """Send a new function task ``fn(*args, **kwargs)`` to the subprocess.
 
         Parameters
         ----------