tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-29 12:35:16 +00:00
Code Issues Packages Projects Releases Wiki Activity

batch : remove logits_all flag (#14141)

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-06-12 11:49:26 +03:00
committed by GitHub
parent e2c0b6e46a
commit c3ee46fab4
10 changed files with 17 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/llama-kv-cache-unified-iswa.cpp
View File

@ -95,12 +95,12 @@ llama_pos llama_kv_cache_unified_iswa::seq_pos_max(llama_seq_id seq_id) const {
return kv_swa->seq_pos_max(seq_id);
}
llama_memory_state_ptr llama_kv_cache_unified_iswa::init_batch(const llama_batch & batch, uint32_t n_ubatch, bool embd_pooled, bool logits_all) {
llama_memory_state_ptr llama_kv_cache_unified_iswa::init_batch(const llama_batch & batch, uint32_t n_ubatch, bool embd_pooled) {
GGML_UNUSED(embd_pooled);
// first try simple split
do {
auto sbatch = llama_sbatch(batch, hparams.n_embd, true, logits_all);
auto sbatch = llama_sbatch(batch, hparams.n_embd, true);
std::vector<llama_ubatch> ubatches;
@ -128,7 +128,7 @@ llama_memory_state_ptr llama_kv_cache_unified_iswa::init_batch(const llama_batch
// if it fails, try equal split
do {
auto sbatch = llama_sbatch(batch, hparams.n_embd, false, logits_all);
auto sbatch = llama_sbatch(batch, hparams.n_embd, false);
std::vector<llama_ubatch> ubatches;