Is it possible to get iPTM for pairs within a multimer?

I'm looking to calculate iPTM for each pair of proteins in a given complex (using the below formula from the Alphafold-Multimer paper). Currently, the Colabfold version of Alphafold-Multimer only gives a single iPTM score for the entire complex...

To do this, I believe I would need the logits, breaks, and asym_id from Alphafold to recalculate it to be pairwise...which it doesn't seem like Colabfold currently gives? Is it possible to access this information?

This is where the function in Alphafold is called
https://github.com/deepmind/alphafold/blob/0be2b30b98f0da7aecb973bde04758fae67eb913/alphafold/model/model.py#L48

confidence_metrics['iptm'] = confidence.predicted_tm_score(
          logits=prediction_result['predicted_aligned_error']['logits'],
          breaks=prediction_result['predicted_aligned_error']['breaks'],
          asym_id=prediction_result['predicted_aligned_error']['asym_id'],
          interface=True)

This is the function it's calling:
https://github.com/deepmind/alphafold/blob/0be2b30b98f0da7aecb973bde04758fae67eb913/alphafold/common/confidence.py#L111

Start with the cited AlphaFold model.py call and alphafold/common/confidence.py predicted_tm_score function, then inspect how ColabFold exposes prediction_result data. Determine whether logits, breaks, and asym_id are available for pairwise calculations; done means the access path or its limitation is clearly established.