diff --git a/src/liger_kernel/transformers/monkey_patch.py b/src/liger_kernel/transformers/monkey_patch.py index 0dd8ffdc..ce85508b 100755 --- a/src/liger_kernel/transformers/monkey_patch.py +++ b/src/liger_kernel/transformers/monkey_patch.py @@ -78,8 +78,8 @@ def _patch_rms_norm_module(module, offset=0.0, eps=1e-6, casting_mode="llama", i _bind_method_to_module(module.modules_to_save.default, "extra_repr", LigerRMSNorm.extra_repr) _bind_method_to_module(module.original_module, "forward", LigerRMSNorm.forward) _bind_method_to_module(module.original_module, "extra_repr", LigerRMSNorm.extra_repr) - module.modules_to_save.default.__class__.__name__ = LigerRMSNorm.__name__ - module.original_module.__class__.__name__ = LigerRMSNorm.__name__ + _bind_method_to_module(module.modules_to_save.default, "_get_name", lambda self: LigerRMSNorm.__name__) + _bind_method_to_module(module.original_module, "_get_name", lambda self: LigerRMSNorm.__name__) else: module.offset = offset module.casting_mode = casting_mode @@ -88,7 +88,7 @@ def _patch_rms_norm_module(module, offset=0.0, eps=1e-6, casting_mode="llama", i module.row_mode = row_mode _bind_method_to_module(module, "forward", LigerRMSNorm.forward) _bind_method_to_module(module, "extra_repr", LigerRMSNorm.extra_repr) - module.__class__.__name__ = LigerRMSNorm.__name__ + _bind_method_to_module(module, "_get_name", lambda self: LigerRMSNorm.__name__) def _patch_layer_norm_module(module, eps=1e-6): @@ -110,28 +110,28 @@ def _patch_layer_norm_module(module, eps=1e-6): module.original_module.hidden_size = getattr(module, "hidden_size", None) or getattr( module, "normalized_shape", None ) - _bind_method_to_module(module.modules_to_save.default, "forward", LigerRMSNorm.forward) - _bind_method_to_module(module.modules_to_save.default, "extra_repr", LigerRMSNorm.extra_repr) - _bind_method_to_module(module.original_module, "forward", LigerRMSNorm.forward) - _bind_method_to_module(module.original_module, "extra_repr", LigerRMSNorm.extra_repr) - module.modules_to_save.default.__class__.__name__ = LigerLayerNorm.__name__ - module.original_module.__class__.__name__ = LigerLayerNorm.__name__ + _bind_method_to_module(module.modules_to_save.default, "forward", LigerLayerNorm.forward) + _bind_method_to_module(module.modules_to_save.default, "extra_repr", LigerLayerNorm.extra_repr) + _bind_method_to_module(module.original_module, "forward", LigerLayerNorm.forward) + _bind_method_to_module(module.original_module, "extra_repr", LigerLayerNorm.extra_repr) + _bind_method_to_module(module.modules_to_save.default, "_get_name", lambda self: LigerLayerNorm.__name__) + _bind_method_to_module(module.original_module, "_get_name", lambda self: LigerLayerNorm.__name__) else: module.variance_epsilon = getattr(module, "variance_epsilon", None) or getattr(module, "eps", None) or eps module.hidden_size = getattr(module, "hidden_size", None) or getattr(module, "normalized_shape", None) _bind_method_to_module(module, "forward", LigerLayerNorm.forward) _bind_method_to_module(module, "extra_repr", LigerLayerNorm.extra_repr) - module.__class__.__name__ = LigerLayerNorm.__name__ + _bind_method_to_module(module, "_get_name", lambda self: LigerLayerNorm.__name__) def _patch_swiglu_module(module, liger_module): _bind_method_to_module(module, "forward", liger_module.forward) - module.__class__.__name__ = liger_module.__name__ + _bind_method_to_module(module, "_get_name", lambda self: liger_module.__name__) def _patch_geglu_module(module): _bind_method_to_module(module, "forward", LigerGEGLUMLP.forward) - module.__class__.__name__ = LigerGEGLUMLP.__name__ + _bind_method_to_module(module, "_get_name", lambda self: LigerGEGLUMLP.__name__) def apply_liger_kernel_to_granite(