pytorch

/

Qwen3-4B-INT8-INT4

like 1

Follow

pytorch 254

• "torch._utils._rebuild_tensor_v2",
• "torch._utils._rebuild_wrapper_subclass",
• "torchao.quantization.linear_activation_quantized_tensor.LinearActivationQuantizedTensor",
• "torchao.quantization.quant_api._int8_asymm_per_token_quant",
• "torch.serialization._get_layout",
• "torch.device",
• "collections.OrderedDict",
• "torchao.dtypes.uintx.q_dq_layout.QDQLayout",
• "torch.BFloat16Storage",
• "torchao.dtypes.uintx.q_dq_layout.QDQTensorImpl",
• "torch.int8",
• "torch.CharStorage",
• "torch._tensor._rebuild_from_type_v2",
• "torchao.dtypes.affine_quantized_tensor.AffineQuantizedTensor",
• "torchao.quantization.quant_primitives.ZeroPointDomain",
• "torch.FloatStorage",
• "torch.float32"

How to fix it?

• "torch.BFloat16Storage",
• "torch.CharStorage",
• "torch._utils._rebuild_wrapper_subclass",
• "collections.OrderedDict",
• "torchao.dtypes.uintx.q_dq_layout.QDQLayout",
• "torch.FloatStorage",
• "torch.float32",
• "torchao.quantization.quant_primitives.ZeroPointDomain",
• "torchao.quantization.linear_activation_quantized_tensor.LinearActivationQuantizedTensor",
• "torchao.dtypes.uintx.q_dq_layout.QDQTensorImpl",
• "torchao.quantization.quant_api._int8_asymm_per_token_quant",
• "torch.int8",
• "torch._utils._rebuild_tensor_v2",
• "torch._tensor._rebuild_from_type_v2",
• "torchao.dtypes.affine_quantized_tensor.AffineQuantizedTensor",
• "torch.serialization._get_layout",
• "torch.device"

How to fix it?