vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 17:44:38 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 3

some tweaks and cleanup

Browse source
This commit is contained in:
Concedo 2025-01-13 23:50:54 +08:00
parent 636beac6d2
commit e77d566268
3 changed files with 8 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
otherarch/tts_adapter.cpp
View file

@ -697,7 +697,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
if(!inputs.quiet)
{
printf("\nTTS Generating (%d input tokens)...", prompt_inp.size());
printf("\nTTS Processing (%d input tokens)...\n", prompt_inp.size());
}
prompt_add(prompt_inp, model_ttc, "<|text_end|>\n<|audio_start|>\n", false, true);
@ -771,6 +771,10 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
output.status = 0;
return output;
}
if(!inputs.quiet)
{
printf("\rTTS Generating (%d AudioTokens)", n_decode);
}
}
if(!inputs.quiet && ttsdebugmode==1)
@ -818,7 +822,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
audio[i] = 0.0f;
}
//add some silence at the end
for (int i = 0; i < 24000/20; ++i) {
for (int i = 0; i < 24000/10; ++i) {
audio.push_back(0.0f);
}

2
src/llama-context.cpp
View file

@ -87,7 +87,7 @@ void llama_set_inputs(llama_context & lctx, const llama_ubatch & ubatch) {
//GGML_ASSERT(lctx.inp_out_ids && "every model that can must skip unused outputs");
if (!lctx.inp_out_ids) {
LLAMA_LOG_WARN("%s: 'lctx.inp_out_ids' is not created\n", __func__);
//LLAMA_LOG_WARN("%s: 'lctx.inp_out_ids' is not created\n", __func__);
} else {
const int64_t n_tokens = ubatch.n_tokens;

2
src/llama.cpp
View file

@ -11124,7 +11124,7 @@ static int llama_decode_impl(
GGML_ASSERT_CONTINUE(n_tokens_all <= cparams.n_batch);
GGML_ASSERT_CONTINUE((cparams.causal_attn || cparams.n_ubatch >= n_tokens_all) && "non-causal attention requires n_ubatch >= n_tokens");
//GGML_ASSERT_CONTINUE((cparams.causal_attn || cparams.n_ubatch >= n_tokens_all) && "non-causal attention requires n_ubatch >= n_tokens");
if (lctx.t_compute_start_us == 0) {
lctx.t_compute_start_us = ggml_time_us();