All pull requests

Pull requests

• Talker LoRA at run time (PEFT adapter beside the quantised weights)

  #43·timoncool opened Oct 19, 2026·

  ·

• tts: accept incremental text during synthesis

  #42·mudler-agent opened Oct 19, 2026·

  ·

• build: add a Metal + BLAS recipe for Apple GPUs

  #39·lvscar opened Oct 12, 2026·

  ·

• bpe: pre-tokenize exactly like the HF Qwen2 tokenizer (Unicode classes, NFC)

  #38·nonml opened Oct 11, 2026·

  ·

• feat: 补齐线程配置与实际后端探测 ABI

  #37·youngshunf opened Oct 11, 2026·

  ·

• Replace fprintf with qt_log across all library sources

  #25·Xydane opened Aug 12, 2026·

  ·
  1

• Add qwentts-serve HTTP server, WebUI, device selection and configurable talker KV context

  #24·soymh opened Aug 7, 2026·

  ·

• talker: reserve prefill compute buffer at load time (max_prefill_tokens)

  #16·DerGary opened Jul 23, 2026Contributor·

  ·
  2

• Implement reference speaker embedding feature for tts-server

  #6·zcmmwbd opened Jun 28, 2026·

  ·