Lint and format

Merge branch 'main' into refactor/rename-4b-to-gptq

sane starter defaults and add lora

add example for falcoln support

refactor(example): rename 4bit-lora-7b by gptq-lora-7b

refactor(param): rename load_4bit config param by gptq

Disable formers for QuickStart config

pr comments addressed

fix tokenizer loading, got openllama 3b working

fixes w/ example for super basic lora starter

support for replit lm

fix config for parity with previous change

not everyone has bf16 available

add 4bit lora 7b

push up redpajama 3b example

add support for trust_remote_code for mpt models