sisleyli opened a new issue, #13592: URL: https://github.com/apache/tvm/issues/13592
[ChangeBatch](https://github.com/apache/tvm/blob/65dbee7f0c3a601779718acb3785451f1089ee79/python/tvm/relay/transform/transform.py#L1090) It seems that `ChangeBatch` pass only change the types of vars. ```python def visit_var(self, var): if var in change_batch.data: ty = var.type_annotation new_shape = list(ty.shape) new_shape[change_batch.data[var]] = change_batch.batch_size return relay.Var(var.name_hint, relay.TensorType(new_shape, ty.dtype)) return var ``` But what if there are some ops in model that are related to tensor shape? For example, `reshape`, or something else? I think it's a little difficult to handle these ops, and I don't have a elegant solution now. Maybe we assume batch dimension is always shape[0] will be more easier? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
