Fix typos (#45574)
* squash all changes into one
* not needed
* fix model name in doc
* Hy v3 validated runtime 0415 (#6)
* fix(hy_v3): update test model_id + regenerate with transformers 5.6
- Update integration test model_id to tiny-random-HYV3ForCausalLM
- Regenerate modeling_hy_v3.py with transformers 5.6 converter
* fix(hy_v3): simplify HYV3ExpertMLP — remove unused forward override
HYV3ExpertMLP is only used for shared_experts, which never passes
routing_weights. Remove the dead-code forward override.
Parent HunYuanMoEV1MLP.forward() handles it correctly.
* internal integration test + clean mlp
* add shared experts to tp plan
* feat(hy_v3): add MTP weight ignore and update documentation
- Add _keys_to_ignore_on_load_unexpected for MTP layer 80 weights
- Fix model name capitalization in docs
- Update example input in documentation
* fix repo
* fix
* fix
---------
Co-authored-by: russellfeng <russellfeng@tencent.com>
Co-authored-by: manayang <jackmanayang@gmail.com>
