fix: correct torch.torch.int32 typo to torch.int32

[Mr-Neutr0n]

Summary

• Fix typo torch.torch.int32 to torch.int32 in internvl_chat/internvl/model/internlm2/modeling_internlm2.py (line 87)

Details

In the _get_unpad_data function, the dtype argument to torch.cumsum is specified as torch.torch.int32. This works at runtime by accident because torch.torch resolves circularly back to the torch module itself (i.e., torch.torch is torch evaluates to True), so torch.torch.int32 ends up being the same object as torch.int32.

However, this is clearly a typo and will cause failures under static type checkers (e.g., mypy, pyright) since torch.torch is not part of the public API. This PR corrects it to torch.int32.

Note that line 84 in the same function already uses the correct torch.int32:

seqlens_in_batch = attention_mask.sum(dim=-1, dtype=torch.int32)

[Mr-Neutr0n]

Closing this one to tidy up my open pull requests.

It's been open around five months with no review activity, and I'd rather withdraw it than leave it sitting in your queue. Nothing needed from you, and no hard feelings at all.

If the fix is still wanted, this can be reopened, or I'm happy to redo it properly against current main. Apologies for the noise.