Commit History

Autor SHA1 Mensaxe Data
  Michael Yang 74bd09652d ml/backend/ggml: load tensors in 32KiB chunks hai 1 mes
  Jesse Gross 0ff28758b3 ollamarunner: Provide mechanism for backends to report loading progress hai 1 mes
  Michael Yang 63a394068c use 2d pooling hai 1 mes
  Michael Yang 0df1800436 set non-causal attention hai 1 mes
  Michael Yang 4b037a97dc add gemma vision encoder hai 1 mes
  Patrick Devine 5f74d1fd47 gemma2 impl hai 2 meses
  Michael Yang 9926eae015 fix: pad tensor item if ge zero hai 1 mes
  Jesse Gross 4100ed7bdd ml: Add support for quantized KV cache hai 2 meses
  Michael Yang 7bae7fa5ce ml/backend/ggml: create tensor on specific backend hai 2 meses
  Michael Yang 764e199d67 kvcache: create cache ctx per layer hai 2 meses
  Michael Yang 05a01fdecb ml/backend/ggml: consolidate system info logging hai 2 meses
  Jesse Gross 21aa666a1e ml: Enable support for flash attention hai 2 meses
  Jesse Gross ee141cc821 ml: Empty tensor constructor for tensors hai 2 meses
  Jesse Gross 854a9195f3 attention: Remove unnecessary contiguous operations hai 2 meses
  Michael Yang 3e8b8a1933 ml: update Context.Forward interface hai 2 meses
  Michael Yang 53d2990d9b model: add bos token if configured hai 2 meses
  Jesse Gross f53f4198c3 ml: Abstract attention out of model definitions hai 2 meses
  Jesse Gross bd6a7d5e64 ollamarunner: Pass runner performance parameters to backends hai 2 meses
  Daniel Hiltgen df2680b4b9 Wire up system info log for new engine (#9123) hai 2 meses
  Jesse Gross ed443a0393 Runner for Ollama engine hai 4 meses
  Jesse Gross d773b7d671 backend: API to support full precision matmul hai 2 meses
  Jesse Gross 4d4463b2bd backend: Support graph computation that does not return an output hai 2 meses
  Jesse Gross 0e38297f87 backend: Consistently use int (vs. int64) for tensor shapes hai 2 meses
  Jesse Gross 7e13f568dc backend: Don't return an error on Close hai 2 meses
  Michael Yang 58245413f4 next ollama runner (#7913) hai 2 meses