[ZeRO] Default disable elastic ckpt in stage 1+2 and reduce CPU memor…

…y overhead during ckpt load (deepspeedai#1525)

Co-authored-by: Olatunji Ruwase <olruwase@microsoft.com>

Various small documentation text improvements (deepspeedai#1665)

Co-authored-by: Jeff Rasley <jerasley@microsoft.com>

Remove unused import of ssl.OP_ENABLE_MIDDLEBOX_COMPAT (deepspeedai#1601

)

Several fixes for our read-the-docs build (deepspeedai#1579)

Fix 1bit extra issue (deepspeedai#1542)

Prevent creation of local temp directory (deepspeedai#1494)

Co-authored-by: Olatunji Ruwase <olruwase@microsoft.com>
Co-authored-by: Jeff Rasley <jerasley@microsoft.com>

Fix the workspace allocation for the transformer kernel (deepspeedai#…

…1397)

* fix the workspace allocation for the transformer kernel

* change layer-id type & rm one unit test due to OOM

[zero_to_fp32] adapt to 4-bytes alignment in z2 (deepspeedai#1372)

Co-authored-by: Olatunji Ruwase <olruwase@microsoft.com>

Update setup.py (deepspeedai#1361)

updated classifiers

Reducing the memory-overhead of creating model for multi-GPU run (dee…

…pspeedai#1244)

Co-authored-by: Jeff Rasley <jerasley@microsoft.com>