vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 17:14:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Alone in the darkness

Browse source 
They're coming for you
I know they will try to catch me too
Alone in the darkness
They're calling for you
There's nowhere to run for cover
This commit is contained in:
Concedo 2024-10-24 22:29:20 +08:00
commit 94a5a27b85
44 changed files with 6803 additions and 2143 deletions
Download patch file Download diff file Expand all files Collapse all files

8
common/common.cpp
View file

@ -957,7 +957,7 @@ struct common_init_result common_init_from_params(common_params & params) {
}
if (llama_model_has_encoder(model)) {
llama_encode(lctx, llama_batch_get_one(tmp.data(), tmp.size(), 0, 0));
llama_encode(lctx, llama_batch_get_one(tmp.data(), tmp.size()));
llama_token decoder_start_token_id = llama_model_decoder_start_token(model);
if (decoder_start_token_id == -1) {
decoder_start_token_id = bos;
@ -966,7 +966,7 @@ struct common_init_result common_init_from_params(common_params & params) {
tmp.push_back(decoder_start_token_id);
}
if (llama_model_has_decoder(model)) {
llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch), 0, 0));
llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch)));
}
llama_kv_cache_clear(lctx);
llama_synchronize(lctx);
@ -1037,7 +1037,7 @@ static ggml_type kv_cache_type_from_str(const std::string & s) {
return GGML_TYPE_Q5_1;
}
throw std::runtime_error("Invalid cache type: " + s);
throw std::runtime_error("Unsupported cache type: " + s);
}
struct llama_context_params common_context_params_to_llama(const common_params & params) {
@ -1049,7 +1049,7 @@ struct llama_context_params common_context_params_to_llama(const common_params &
cparams.n_ubatch = params.n_ubatch;
cparams.n_threads = params.cpuparams.n_threads;
cparams.n_threads_batch = params.cpuparams_batch.n_threads == -1 ?
params.cpuparams.n_threads : params.cpuparams_batch.n_threads;
params.cpuparams.n_threads : params.cpuparams_batch.n_threads;
cparams.logits_all = params.logits_all;
cparams.embeddings = params.embedding;
cparams.rope_scaling_type = params.rope_scaling_type;