Histórico de Commits

Autor SHA1 Mensagem Data
  Jesse Gross a8e83a7654 Disable causal attention based on batch index há 2 meses atrás
  Michael Yang e95278932b use non-causal mask only for image positions há 2 meses atrás
  Jesse Gross a1cda80bcb model: Update encoder cache to use multimodal input processing handler há 2 meses atrás
  Jesse Gross f52b2615ef kvcache: Set context for shift offsets há 2 meses atrás
  Jesse Gross 6da8b6a879 kvcache: Support non-causal attention há 2 meses atrás
  Michael Yang 7bae7fa5ce ml/backend/ggml: create tensor on specific backend há 2 meses atrás
  Michael Yang 764e199d67 kvcache: create cache ctx per layer há 2 meses atrás
  Jesse Gross 21aa666a1e ml: Enable support for flash attention há 2 meses atrás
  Jesse Gross 854a9195f3 attention: Remove unnecessary contiguous operations há 2 meses atrás
  Michael Yang 3e8b8a1933 ml: update Context.Forward interface há 2 meses atrás
  Jesse Gross ed443a0393 Runner for Ollama engine há 4 meses atrás