drop noise rate

config/loss/sft.yaml

@@ -1,2 +1 @@
-name: sft
-mode_weight: 0.0
+name: sft

preference_datasets.py

@@ -159,7 +159,7 @@ def split_prompt_and_responses(ex):
 
     return data
 
-def get_dataset(name: str, split: str, silent: bool = False, cache_dir: str = None, noise_rate=0.):
+def get_dataset(name: str, split: str, silent: bool = False, cache_dir: str = None):
     """Load the given dataset by name. Supported by default are 'shp', 'hh', and 'se'."""
     if name == 'shp':
         data = get_shp(split, silent=silent, cache_dir=cache_dir)
@@ -288,8 +288,7 @@ def get_batch_iterator(names: List[str],
                        n_examples: Optional[int] = None,
                        seed:int = 0,
                        silent: bool = False,
-                       cache_dir: Optional[str] = None,
-                       noise_rate=0.) -> Iterator[Dict]:
+                       cache_dir: Optional[str] = None) -> Iterator[Dict]:
     """Get an iterator over batches of data. Stops after n_epochs or n_examples, whichever comes first.
 
     Args:
@@ -317,7 +316,7 @@ def get_batch_iterator(names: List[str],
         flat_data = []
         for name in names:
             truncation_mode = 'keep_end' if name == 'hh' else 'keep_start'
-            for prompt, data in get_dataset(name, split, silent=silent, cache_dir=cache_dir, noise_rate=noise_rate).items():
+            for prompt, data in get_dataset(name, split, silent=silent, cache_dir=cache_dir).items():
                 flat_data.append((prompt, data['responses'], data['pairs'], data['sft_target'], truncation_mode))
 
     collate_fn = get_collate_fn(tokenizer)

trainers.py

@@ -193,7 +193,7 @@ def __init__(self, policy: nn.Module, config: DictConfig, seed: int, run_dir: st
             self.loss_mean = torch.zeros(1, device='cuda')
             self.loss_std = torch.zeros(1, device='cuda')
 
-        self.train_iterator = get_batch_iterator(**data_iterator_kwargs, split='train', n_epochs=config.n_epochs, n_examples=config.n_examples, batch_size=config.batch_size, silent=rank != 0, cache_dir=get_local_dir(config.local_dirs), noise_rate=config.noise_rate)
+        self.train_iterator = get_batch_iterator(**data_iterator_kwargs, split='train', n_epochs=config.n_epochs, n_examples=config.n_examples, batch_size=config.batch_size, silent=rank != 0, cache_dir=get_local_dir(config.local_dirs))
         rank0_print(f'Loaded train data iterator')
         self.eval_iterator = get_batch_iterator(**data_iterator_kwargs, split='test', n_examples=config.n_eval_examples, batch_size=config.eval_batch_size, silent=rank != 0, cache_dir=get_local_dir(config.local_dirs))
         self.eval_batches = list(self.eval_iterator)