model : add LiquidAI/d1-omni-600M decision model (#30114)

* model : add LiquidAI/d1-omni-600M decision model

Assisted-by: Claude Opus 5.5

* mtmd : keep conformer GLU sigmoid on CUDA

Assisted-by: Claude Opus 5.5

* server : take d1omni audio through images and input_audio, scope memory-less lfm2 to non-causal

Assisted-by: Claude Opus 5.5

* common : rename decision type d1omni to lfm2-d1-omni, server : make images an alias of files

Assisted-by: Claude Opus 5.5
This commit is contained in:
Tarek Dakhran authored and GitHub committed 2026-10-08 02:07:49 +02:00
1 parent aa5e0092fd
commit a657f7e981
21 files changed
+833 -35

No files matched your search

+5
View File
@@ -2368,6 +2368,11 @@ ggml_tensor * llama_model::get_rope_factors(const llama_cparams & cparams, int i
llama_memory_i * llama_model::create_memory(const llama_memory_params & params, const llama_cparams & cparams) const {
llama_memory_i * res;
// the non-causal LFM2 decision graph reads the whole prompt in one batch, nothing is kept
if (arch == LLM_ARCH_LFM2 && !hparams.causal_attn && hparams.n_layer_decision > 0) {
return nullptr;
}
switch (arch) {
// Models that need specific instantiation should be handled in the
// switch statement