Transformers releases发布v5.20.2补丁更新

Patch release v5.20.2

要点速览

Transformers 发布 v5.20.2 补丁更新，修复了 generate 函数中输入类型丢失的问题，该问题影响多模态模型；同时解决了其他若干问题，包括输入字典类型转换、bfloat16 专家并行梯度为零、仅报告文本的 VLM 输入模态声明以及 eager_mask 中的主机到设备复制。

正文

Patch release: v5.20.2

This patch notably solves an issue where some inputs could loose their original type inside prepare_inputs_for_generation in generate, leading to problems with multimodal models. It also fixes some other issues listed on the below commits.

It contains the following commits:

Fix inputs dict casting in generate (#49731) by @Cyrilvallez
Fix zero expert gradients in bfloat16 with expert parallelism and ep_size < fsdp_size (#49688) by @qgallouedec
Declare the input modalities of the VLMs that report text only (#49677) by @qgallouedec
Avoid a host-to-device copy in eager_mask (#49684) by @hmellor