Merge pull request #106 from BlackSamorez/sharding_refactor

[WIP] ZeRO-3 refactoring (sharding)

Merge pull request #113 from BlackSamorez/prepare_model_inputs

Forwarding _prepare_model_inputs

Merge pull request #110 from BlackSamorez/pretrained_fix

find_predefined_tensor_parallel_config try-except fix

Merge pull request #103 from BlackSamorez/gpt2-fix

Gpt2 fix

Merge pull request #102 from BlackSamorez/version

Version bump

Merge pull request #92 from BlackSamorez/falcon

Falcon lm_head split hotfix

Merge pull request #91 from BlackSamorez/falcon

Falcon predefined config

Merge pull request #90 from BlackSamorez/dispatch

Fixed dispatch of tp.Sharded models

Merge pull request #87 from BlackSamorez/peft_unimport

Removing PEFT from dependencies. Replacing with runtime checks

Merge pull request #73 from BlackSamorez/state_dict_fixes

State dict fixes for tied weights