turn off ddp_overlap for non-O2 usage and add synthetic data argument (…

…NVIDIA#396)

TRANSFORMERS_OFFLINE=0

Signed-off-by: Alexandros Koumparoulis <[email protected]>

Fixed llama2 13b perf (NVIDIA#293)

Signed-off-by: Selvaraj Anandaraj <[email protected]>
Co-authored-by: Selvaraj Anandaraj <[email protected]>

Added gc_interval for manual garbage collection (NVIDIA#278)

Signed-off-by: Selvaraj Anandaraj <[email protected]>
Co-authored-by: Selvaraj Anandaraj <[email protected]>

Merge pull request NVIDIA#232 from NVIDIA/chcui/remove_fp8_comment

Remove outdated fp8 comment

change container path

Signed-off-by: dimapihtar <[email protected]>

update container path

Signed-off-by: dimapihtar <[email protected]>

Merge pull request NVIDIA#133 from NVIDIA/23.08.02

change container to 23.08.02

Merge pull request NVIDIA#103 from NVIDIA/23.07

23.07 changes

updated readme

Signed-off-by: David Mosallanezhad <[email protected]>