Aurelien Chartier

chore: Handle qwen2audio inputs ids expansion during processing

6

Transformers is expanding input_ids during processing since 4.48: https://github.com/huggingface/transformers/pull/35534 Hence, it does not need to be done in TRT-LLM code anymore.

chore: cleanup py_executor code

3

* Add common loop cleanup function * Remove checks for attention DP if nothing to queue * Remove extra return statements * Remove extra variables * Remove commented debug print

[None][feat] Add environment variable to force spec-dec number of accepted tokens

7

## Summary by CodeRabbit * **New Features** * Introduced environment-driven configuration for speculative decoding token acceptance, allowing operators to override and control the number of accepted tokens for optimization and...