|  lchu | 1cc9df19e6
							
							remove unused import | 2 年之前 | 
				
					
						|  Geeta Chauhan | 597e63429c
							
							fix some typos. (#93) | 2 年之前 | 
				
					
						|  Chengyu Ma | e741dde675
							
							fix some typos | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | bf4e0a134b
							
							Add BT after peft module load | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | b2a55022cb
							
							clean up | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | 0941a9ecf7
							
							clean up | 2 年之前 | 
				
					
						|  Geeta Chauhan | 1803c35f3c
							
							Fix filename typo (#94) | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | 8601ae14ea
							
							extend word list | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | 44ef280d31
							
							adding flash attention and xformer memory efficient through PT SDPA | 2 年之前 | 
				
					
						|  Philipp Parzer | e6329a1fac
							
							Fix filename typo | 2 年之前 | 
				
					
						|  luoyifan | 79d0d4fc4e
							
							fix some typos. | 2 年之前 | 
				
					
						|  lchu | c453b668fa
							
							add doc example about using low_cpu_fsdp | 2 年之前 | 
				
					
						|  Geeta Chauhan | dd57dc5cf9
							
							Update spellcheck.sh (#88) | 2 年之前 | 
				
					
						|  lchu | 80a4c36707
							
							further fix #90 | 2 年之前 | 
				
					
						|  lchu | 0c51b47262
							
							fix #90 | 2 年之前 | 
				
					
						|  lchu | c19c5c69aa
							
							fix fsdp construction on low_cpu_fsdp | 2 年之前 | 
				
					
						|  lchu | e216c6f1f3
							
							address #87 | 2 年之前 | 
				
					
						|  sekyonda | 59b59afe88
							
							Update spellcheck.sh | 2 年之前 | 
				
					
						|  lchu | 895dfcea30
							
							add nightly check for using low_cpu_fsdp mode | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | 9b0eae4056
							
							add max_pad length as an arg | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | 6beadcd8dd
							
							adding padding to inference script with embedding size changing | 2 年之前 | 
				
					
						|  lchu | 1e64fc98d9
							
							switch to simpler param_init_fn and meta device init | 2 年之前 | 
				
					
						|  lchu | 101391f46a
							
							Revert "replace init_empty_weights with torch.device(meta)" | 2 年之前 | 
				
					
						|  lchu | c8d4f38d23
							
							replace init_empty_weights with torch.device(meta) | 2 年之前 | 
				
					
						|  lchu | d8a81bb531
							
							save cpu mem by leveraging FSDP rank0 broadcasting | 2 年之前 | 
				
					
						|  Geeta Chauhan | 1387b76e11
							
							fixing the full state path in checkpoint handler+loss report calculation (#51) | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | 88d3e1febc
							
							fix the save_train_param condition | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | b56028c98d
							
							fixing the word list/spell check | 2 年之前 | 
				
					
						|  Hamid Shojanazeri | 62be60355a
							
							resolving conflicts | 2 年之前 | 
				
					
						|  Geeta Chauhan | 174b856591
							
							update README: python 3.9 rec + fix formatting (#63) | 2 年之前 |