mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-07-02 13:55:02 +00:00
fix: Fix wrong bool condition for split equal in hybrid cache
Branch: HybridRecurrentCache Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
This commit is contained in:
@ -96,7 +96,7 @@ llama_pos llama_kv_cache_hybrid_recurrent::seq_pos_max(llama_seq_id seq_id) cons
|
||||
llama_memory_state_ptr llama_kv_cache_hybrid_recurrent::init_batch(const llama_batch & batch, uint32_t n_ubatch, bool embd_pooled, bool logits_all) {
|
||||
|
||||
// since this includes a recurrent cache, we cannot use split_simple
|
||||
auto sbatch = llama_sbatch(batch, hparams.n_embd, true, logits_all);
|
||||
auto sbatch = llama_sbatch(batch, hparams.n_embd, false, logits_all);
|
||||
|
||||
// follow the recurrent pattern for creating the ubatch splits
|
||||
std::vector<llama_ubatch> ubatches;
|
||||
|
Reference in New Issue
Block a user