Out of memory error when creating tensor data using a function Ort::Value::CreateTensor

[Beluooo-H]

std::vector<std::vector<int64_t>> input_ids(batch_size, std::vector<int64_t>(max_seq_length));
I have specified batch_size and max_seq_length.
for(int i=0; i < batch_size; i++){ Ort::Value input_tensor = Ort::Value::CreateTensor<int64_t>(memory_info, input_ids[i].data(), input_ids[i].size(), input_dims.data(), input_dims.size()); input_tensors.push_back(std::move(input_tensor)); }
I'm using a loop to add data from input_ids to the input_tensors container by creating a tensor that matches the dimensional requirements, but I always get the following error when creating it.
terminate called after throwing an instance of 'Ort::Exception' what(): not enough space: expected 48, got 24 Aborted
When I specify batch_size to be 1, the above error is not generated, once batch_size is not 1, an error message is reported.
Thanks you.

[RyanUnderhill]

(probably stale, but replying just in case)

My guess is your input_dims has the batch_size in it, and your vector of input_ids is only max_seq_length. So this only works for a batch size of 1 where these match.

You need a single vector (not a vector of vectors) to store your input ids as the memory must be contiguous. Then the size needs to match the shape of the tensor you're creating.