Topic: 01

The ggml/llama.cpp project fixed a CI break and restored Kronecker product FWHT support, addressing issues to improve the backend operations testing. These changes are crucial for enhancing the library's functionality and reliability in handling specific tensor operations.

A new dual-form ASR approach called Semantics-Aware Inverse Text Normalization has been developed for Chinese speech recognition, aiming to produce both accurate spoken-form transcripts and readable written-form transcripts. This method addresses the need in modern ASR scenarios where both forms are crucial for faithful transcription and readability.

arxiv.org↗

The ggml/llama.cpp project released a fix to ensure the entire source code is not rebuilt on each new commit, addressing an issue that could slow development. This update is crucial for improving workflow efficiency in contributing to the llama model's implementation.

The ggml/llama.cpp project released an update that enhances support for multi-NPU devices, making the backend fully asynchronous by default. This update is significant as it improves performance and flexibility in deploying models across multiple hardware accelerators.