up

Author: metascroy

examples/models/llama/export_llama_lib.py

@@ -763,6 +763,7 @@ def _to_edge_and_lower_llama_xnnpack(
         raise NotImplementedError(
             "export_llama does not support XNNPack and generating ETRecord at the moment."
         )
+
     builder = builder_exported.pt2e_quantize(quantizers).to_edge_transform_and_lower(
         partitioners
     )

examples/models/llama/model.py

@@ -16,6 +16,7 @@
     get_default_model_resource_dir,
 )
 from executorch.examples.models.llama.llama_transformer import Transformer
+
 from executorch.examples.models.llama.model_args import ModelArgs
 from torchao.utils import TorchAOBaseTensor
 
@@ -101,7 +102,6 @@ def __init__(self, **kwargs):
         if fairseq2_checkpoint:
             print("Using fairseq2 checkpoint")
             checkpoint = convert_to_llama_checkpoint(checkpoint=checkpoint)
-            print("checkpoint", checkpoint)
         if "model" in checkpoint:
             # NB: some checkpoint contains a "model" field, which is the actual weights dict
             checkpoint = checkpoint["model"]

extension/llm/export/builder.py

@@ -31,6 +31,7 @@
 from executorch.exir.passes import MemoryPlanningPass
 from executorch.exir.passes.quant_fusion_pass import QuantFusionPass
 from executorch.exir.passes.sym_shape_eval_pass import ConstraintBasedSymShapeEvalPass
+
 from executorch.extension.export_util.utils import export_to_edge, save_pte_program
 
 from executorch.extension.llm.export.export_passes import RemoveRedundantTransposes
@@ -231,7 +232,6 @@ def _export(self, module: Optional[torch.nn.Module] = None) -> ExportedProgram:
                     logging.info("Re-exporting with:")
                 else:
                     logging.info("Exporting with:")
-
                 logging.info(f"inputs: {self.example_inputs}")
                 logging.info(f"kwargs: {self.example_kwarg_inputs}")
                 logging.info(f"dynamic shapes: {dynamic_shape}")