update tutorials links for 2.4 release (tf2)

Summary:
updates for tf2 branch (they are slightly different so raising two diffs)
(T51570)
TF2.4 only

Test Plan: NA

Reviewers: #documentation, georgep, #tensorflow, #framework_ip_review_-_any_oss_or_third-party_code_use_has_been_approved

Reviewed By: georgep, #tensorflow, #framework_ip_review_-_any_oss_or_third-party_code_use_has_been_approved

Subscribers: louisev

Maniphest Tasks: T51570

Differential Revision: https://phabricator.sourcevertex.net/D57094

Author: cBog

tensorflow/compiler/plugin/poplar/docs/custom_codelet.rst

@@ -52,7 +52,7 @@ C++ code compiled for, and executed on, the IPU.
 For more information about writing Poplar graph programs and codelets, refer to
 the `Poplar and PopLibs User Guide
 <https://docs.graphcore.ai/projects/poplar-user-guide/>`_ and the `Poplar
-tutorials <https://github.com/graphcore/tutorials/tree/sdk-release-2.3/tutorials/poplar>`_
+tutorials <https://github.com/graphcore/tutorials/tree/sdk-release-2.4/tutorials/poplar>`_
 on the Graphcore GitHub tutorials respository.
 
 Your program must contain a function to build the graph, which will be called
@@ -460,10 +460,10 @@ sections. There are further examples in the Graphcore GitHub tutorials
 repository:
 
 * `Custom op with gradient
-  <https://github.com/graphcore/tutorials/tree/sdk-release-2.3/feature_examples/tensorflow/custom_gradient>`_
+  <https://github.com/graphcore/tutorials/tree/sdk-release-2.4/feature_examples/tensorflow/custom_gradient>`_
 
 * `Custom op with codelet for custom vertex code
-  <https://github.com/graphcore/tutorials/tree/sdk-release-2.3/feature_examples/tensorflow/custom_op>`_
+  <https://github.com/graphcore/tutorials/tree/sdk-release-2.4/feature_examples/tensorflow/custom_op>`_
 
 
 .. _inplace_operations:

tensorflow/compiler/plugin/poplar/docs/keras_tf2.rst

@@ -6,7 +6,7 @@ Keras model creation is no different than what you would use if you were
 training on other devices. To target the Poplar XLA device, Keras model creation
 must be inside the ``strategy.scope`` of an ``IPUStrategy``.
 
-For a more practical walkthrough, see `this tutorial about using Keras on the IPU <https://github.com/graphcore/tutorials/tree/sdk-release-2.3/tutorials/tensorflow2/keras>`_
+For a more practical walkthrough, see `this tutorial about using Keras on the IPU <https://github.com/graphcore/tutorials/tree/sdk-release-2.4/tutorials/tensorflow2/keras>`_
 from the Graphcore tutorials repository.
 
 Single IPU models

tensorflow/compiler/plugin/poplar/docs/perf_training.rst

@@ -98,7 +98,7 @@ array and will contain `x1 - x2` for each iteration in the loop.
 
 See entries in the :ref:`api-section` for more details.
 
-For a more practical example, the Graphcore tutorials repository contains a `detailed tutorial about using infeeds and outfeeds with TensorFlow <https://github.com/graphcore/tutorials/tree/sdk-release-2.3/tutorials/tensorflow2/infeed_outfeed>`_.
+For a more practical example, the Graphcore tutorials repository contains a `detailed tutorial about using infeeds and outfeeds with TensorFlow <https://github.com/graphcore/tutorials/tree/sdk-release-2.4/tutorials/tensorflow2/infeed_outfeed>`_.
 
 
 Accessing outfeed queue results during execution