Support Deepspeed finetuning (#94) #48

Job	Run time
call-tests / build	2m 3s
call-finetune / finetune test (EleutherAI/gpt-j-6b)	1h 3m 56s
call-inference / inference test (gpt-j-6b)	12m 0s
call-finetune / finetune test (meta-llama/Llama-2-7b-chat-hf)	36m 40s
call-lint / lint check	34s
call-finetune / finetune test (mistralai/Mistral-7B-v0.1)	45m 7s
call-inference / inference test (mistral-7b-v0.1)	5m 30s
call-inference / inference test (mpt-7b-bigdl)	10m 33s
call-inference / inference test (llama-2-7b-chat-hf-vllm)	3m 0s
call-inference / inference test (mpt-7b, nathan0/mpt-7b-deltatuner-model)	38m 33s
	3h 37m 56s

Provide feedback