Commit History

Author SHA1 Message Date
  Jesse Gross 0fbfcf3c9c model: Pass input tensor instead of raw data to models 1 month ago
  Jesse Gross 0c220935bd input: Rename Options to Batch 1 month ago
  Jeffrey Morgan da0e345200 ml: use input context for extracting outputs (#9875) 1 month ago
  Jesse Gross 282bfaaa95 ollamarunner: Use a separate context per multimodal input 1 month ago
  Jesse Gross 9679f40146 ml: Allow models to constrain inputs to a single batch 1 month ago
  Michael Yang 5e2e0b46b1 fix: error if image requested without vision model 1 month ago
  Michael Yang 63a394068c use 2d pooling 1 month ago
  jmorganca 11bfa62796 add trailing \n\n after <end_of_image> to match reference implementation 1 month ago
  jmorganca f63e62e546 reduce kernel size, add TODO for loading from config 1 month ago
  jmorganca 65b0f329d1 Revert "Allow models to force a new batch" 1 month ago
  Jesse Gross 06007c0a18 Allow models to force a new batch 1 month ago
  Jesse Gross 2c40c4d35e Fix follow up images and images split across batches 1 month ago
  Michael Yang 6b32a2d549 compat with upstream gguf 1 month ago
  Michael Yang 46bb0169c4 update model 1 month ago
  Michael Yang 8934324b72 use fast attention 1 month ago
  Michael Yang 0df1800436 set non-causal attention 1 month ago
  Michael Yang 4b037a97dc add gemma vision encoder 1 month ago
  Patrick Devine 5f74d1fd47 gemma2 impl 2 months ago