thek0tyara/llama-cpp-turboquant
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 4

revert : "[Model] Qwen3.5 dense and MoE support (no vision) (#19435)" (#19453)

Browse source 
This reverts commit 39bf692af1.
This commit is contained in:
Georgi Gerganov 2026-02-09 14:57:51 +02:00 • committed by GitHub
parent f5e7734ff2
commit 972f323e73
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
14 changed files with 399 additions and 1532 deletions
Download patch file Download diff file Expand all files Collapse all files

2
src/llama-context.cpp
View file

@ -2013,7 +2013,7 @@ void llama_context::output_reorder() {
//
uint32_t llama_context::graph_max_nodes(uint32_t n_tokens) const {
if (model.arch == LLM_ARCH_QWEN3NEXT || model.arch == LLM_ARCH_QWEN3_5 || model.arch == LLM_ARCH_QWEN3_5_MOE || model.arch == LLM_ARCH_KIMI_LINEAR) {
if (model.arch == LLM_ARCH_QWEN3NEXT || model.arch == LLM_ARCH_KIMI_LINEAR) {
return std::max<uint32_t>(n_tokens * 40, 32u * model.n_tensors());
}
uint32_t res = std::max<uint32_t>(1024u, 8u*model.n_tensors());