TF 2.4.0 Linux GPU (Compute 7.x), Python 3.9, CUDA 11.2, cuDNN 8, AVX2, MKL

SHA1: 9e03c5bd38a007dd1186aa1717b9197522136a42
SHA256: dd7504bbaa15d776012ff91fe48d58508039221209cb54764946f74c660e8c14
MD5: e5f3587a319e069782b9474a6d012aed

TF 2.3.0 GPU (Compute 7.x), CUDA 11.0, cuDNN 8, AVX2, MKL, TensorRT 7

SHA1: b3a600bf14708b3043f187d12406ff4df03c6f13
MD5: dbbfa3dfac47fddc229d575c5057c299

TF 2.4.0 CPU-only, no AVX, no MKL

SHA1: 4bf18e61c42f68eb124992b31f92e8619b9726f0
MD5: e07221fd972c7a18140a4c7d6d609f9d

TF 2.2.0 GPU (Compute 7.x), CUDA 10.2, cuDNN 7, AVX2, MKL, TensorRT 7

SHA1: 8689ba8bcf3d452a48be441f8c7eb230abe7b5b5
MD5: 1833ee60939d62f8c89bfa2578a6d358

TF 2.1.0 GPU (Compute 7.x), CUDA 10.2, cuDNN 7, AVX2, MKL, TensorRT 6

SHA1: 37ab6deac80eee917f6fc1a6164fbf73271ac181
MD5: 0eec55b6cf0668156565f75677df1a6f

TF 1.15.0 GPU (Compute 7.x), CUDA 10.2, cuDNN 7, AVX2, MKL, TensorRT 6

SHA1: 5bc7c355899608fdef60b17688d7542d76ad528b
MD5: 5d5bc9e8ff5bbdcda3b7bdfdb185e2d9

TF 1.14.1 GPU (Compute 7.x), CUDA 10.2, cuDNN 7, AVX2, MKL, TensorRT 6

SHA1: c29c8c07505c2fb4242a7bcbd26faa68e961bafa
MD5: 15238b46d2c1d40894aa436bc979244a