Nvidia (3000 series) Ampere Architecture Support - Update TensorFlow to v2.4.0

[larsvansoest]

Support for Ampere architectures was added in TensorFlow v2.4.0. This architecture is used by Nividia's 3000 series.

Currently, this project supports up to TensorFlow version v2.3.0. As a result, attempting to run on a 3000 series GPU results in an error mentioned in https://github.com/tensorflow/tensorflow/issues/41990. Moreover, as stated by https://github.com/tensorflow/tensorflow/issues/41990#issuecomment-670224813, the error is raised due to an unsupported graphics card.

To support 3000 series graphics cards for TensorFlow in Haskell, the supported version should be upgraded to at least v2.4.0.