Return true initial batch size information

[jlarson4]

Description

HuggingFace's GPT-2 generates position_ids with shape [1, seq_len] regardless of batch size, relying on PyTorch broadcasting when adding to token embeddings. TransformerBridge's PosEmbedBridge wraps the HF wpe module, so the hook fires on wpe's output before broadcasting — capturing [1, seq, d_model] instead of [batch, seq, d_model].

Type of change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I have not rewritten tests relating to key interfaces which would affect backward compatibility