pre-commit

llmfoundry/models/hf/model_wrapper.py

@@ -60,7 +60,8 @@ def __init__(self,
             self.model.forward).args
         # inspect.getfullargspec HuggingFace quantized model could not return args correctly
         if not self.model_forward_args:
-            self.model_forward_args = inspect.signature(self.model.forward).parameters.keys()
+            self.model_forward_args = inspect.signature(
+                self.model.forward).parameters.keys()
 
         # Note: We need to add the FSDP related attributes to the model AFTER the super init,
         # so that the (possible) embedding resizing doesn't destroy them

scripts/eval/eval.py

@@ -122,8 +122,9 @@ def evaluate_model(model_cfg: DictConfig, dist_timeout: Union[float, int],
                                                 model_gauntlet)  # type: ignore
 
     if fsdp_config and model_cfg.model.load_in_8bit:
-        raise ValueError("The FSDP config block is not supported when loading " +
-                         "Hugging Face models in 8bit.")
+        raise ValueError(
+            'The FSDP config block is not supported when loading ' +
+            'Hugging Face models in 8bit.')
 
     if hasattr(model_cfg.model, 'pretrained_lora_id_or_path'):
         composer_model = load_peft_model(model_cfg.model, tokenizer,

scripts/eval/yamls/hf_8bit_eval.yaml

@@ -22,4 +22,4 @@ device_eval_batch_size: 4
 
 # With load_in_8bit, do not specify fsdp_config
 
-icl_tasks: 'eval/yamls/tasks_light.yaml'
+icl_tasks: 'eval/yamls/tasks_light.yaml'

scripts/train/train.py

@@ -90,9 +90,11 @@ def validate_config(cfg: DictConfig):
             'Setting `torch._dynamo.config.suppress_errors = True` and falling back to eager.'
         )
         torch._dynamo.config.suppress_errors = True  # type: ignore
-    
+
     if cfg.model.get('load_in_8bit', False):
-        raise ValueError("`load_in_8bit` is only supported for evaluation rather than training.")
+        raise ValueError(
+            '`load_in_8bit` is only supported for evaluation rather than training.'
+        )
 
 
 def build_composer_model(model_cfg: DictConfig,