updated readme for new finetune config

Author: wukaixingxp

docs/multi_gpu.md

@@ -138,8 +138,9 @@ It lets us specify the training settings for everything from `model_name` to `da
     mixed_precision: bool=True
     val_batch_size: int=1
     dataset = "samsum_dataset"
-    peft_method: str = "lora" # None,llama_adapter, prefix
+    peft_method: str = "lora" # None, llama_adapter (Caution: llama_adapter is currently not supported with FSDP)
     use_peft: bool=False
+    from_peft_checkpoint: str="" # if not empty and use_peft=True, will load the peft checkpoint and resume the fine-tuning on that checkpoint
     output_dir: str = "PATH/to/save/PEFT/model"
     freeze_layers: bool = False
     num_freeze_layers: int = 1

docs/single_gpu.md

@@ -94,8 +94,9 @@ It let us specify the training settings, everything from `model_name` to `datase
     mixed_precision: bool=True
     val_batch_size: int=1
     dataset = "samsum_dataset"
-    peft_method: str = "lora" # None,llama_adapter, prefix
+    peft_method: str = "lora" # None, llama_adapter (Caution: llama_adapter is currently not supported with FSDP)
     use_peft: bool=False
+    from_peft_checkpoint: str="" # if not empty and use_peft=True, will load the peft checkpoint and resume the fine-tuning on that checkpoint
     output_dir: str = "PATH/to/save/PEFT/model"
     freeze_layers: bool = False
     num_freeze_layers: int = 1
@@ -112,6 +113,7 @@ It let us specify the training settings, everything from `model_name` to `datase
     flop_counter_start: int = 3 # The step to start profiling, default is 3, which means after 3 steps of warmup stage, the profiler will start to count flops.
     use_profiler: bool = False # Enable pytorch profiler, can not be used with flop counter at the same time.
     profiler_dir: str = "PATH/to/save/profiler/results" # will be used if using profiler
+
 ```
 
 * [Datasets config file](../src/llama_recipes/configs/datasets.py) provides the available options for datasets.

recipes/finetuning/README.md

@@ -48,8 +48,9 @@ It lets us specify the training settings for everything from `model_name` to `da
     mixed_precision: bool=True
     val_batch_size: int=1
     dataset = "samsum_dataset"
-    peft_method: str = "lora" # None,llama_adapter, prefix
+    peft_method: str = "lora" # None, llama_adapter (Caution: llama_adapter is currently not supported with FSDP)
     use_peft: bool=False
+    from_peft_checkpoint: str="" # if not empty and use_peft=True, will load the peft checkpoint and resume the fine-tuning on that checkpoint
     output_dir: str = "PATH/to/save/PEFT/model"
     freeze_layers: bool = False
     num_freeze_layers: int = 1
@@ -66,6 +67,7 @@ It lets us specify the training settings for everything from `model_name` to `da
     flop_counter_start: int = 3 # The step to start profiling, default is 3, which means after 3 steps of warmup stage, the profiler will start to count flops.
     use_profiler: bool = False # Enable pytorch profiler, can not be used with flop counter at the same time.
     profiler_dir: str = "PATH/to/save/profiler/results" # will be used if using profiler
+
 ```
 
 * [Datasets config file](../../src/llama_recipes/configs/datasets.py) provides the available options for datasets.