Transformers 发布补丁 v5.14.1 修复 Inkling 集成问题
原标题：Patch release: v5.14.1
AI 摘要
Hugging Face Transformers 发布了补丁版本 v5.14.1，修复了集成 Inkling 模型时出现的多个问题，包括影响使用 EncoderDecoderCache 的模型在辅助生成（assisted generation）过程中的问题，以及 Inkling 模型在 prefill 阶段使用 StaticCache 和 sdpa 且无 padding 时可能出现的错误。该版本还包含 FP8 内核版本升级和 deepgemm 多设备修复。
正文节选
# Patch release v5.14.1 This patch solves a few issues which appeared when integrating Inkling model, most notably an issue affecting models using EncoderDecoderCache during assisted generation. It also fixes an issue that could appear during prefill with StaticCache and sdpa without padding for Inkling which uses a position_bias. It contains the following commits: - Fix sdpa prefill with position_bias (#47359) by @Cyrilvallez - Fix assisted decoding for models with EncoderDecoder cache