CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

tts : fix n_ubatch + make WavTokenizer cache-less

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-05-22 21:53:13 +03:00
parent 797990c4bc
commit 108d484ab2
2 changed files with 5 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
src/llama-model.cpp
View File

@@ -13189,6 +13189,7 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
case LLM_ARCH_JINA_BERT_V2:
case LLM_ARCH_NOMIC_BERT:
case LLM_ARCH_NOMIC_BERT_MOE:
case LLM_ARCH_WAVTOKENIZER_DEC:
{
res = nullptr;
} break;

6
tools/tts/tts.cpp
View File

@@ -579,6 +579,8 @@ int main(int argc, char ** argv) {
params.model = params.vocoder.model;
params.embedding = true;
params.ctx_shift = false; // silence warning
params.n_ubatch = params.n_batch;
common_init_result llama_init_cts = common_init_from_params(params);
@@ -1020,8 +1022,8 @@ lovely<|t_0.56|><|code_start|><|634|><|596|><|1766|><|1556|><|1306|><|1285|><|14
}
GGML_ASSERT(batch.n_tokens == n_codes);
if (llama_decode(ctx_cts, batch) != 0) {
LOG_ERR("%s: llama_decode() failed\n", __func__);
if (llama_encode(ctx_cts, batch) != 0) {
LOG_ERR("%s: llama_encode() failed\n", __func__);
return 1;
}