vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 09:34:37 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	.devops/llama-cli-intel.Dockerfile
#	.devops/llama-server-intel.Dockerfile
#	README.md
#	ggml/src/CMakeLists.txt
#	tests/test-chat-template.cpp
This commit is contained in:
Concedo 2024-07-24 21:57:50 +08:00
commit cca2fa9a6c
4 changed files with 11 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

2
common/common.cpp
View file

@ -2724,7 +2724,7 @@ std::string llama_chat_format_single(const struct llama_model * model,
const llama_chat_msg & new_msg, const llama_chat_msg & new_msg,
bool add_ass) { bool add_ass) {
std::ostringstream ss; std::ostringstream ss;
auto fmt_past_msg = llama_chat_apply_template(model, tmpl, past_msg, false); auto fmt_past_msg = past_msg.empty() ? "" : llama_chat_apply_template(model, tmpl, past_msg, false);
std::vector<llama_chat_msg> chat_new(past_msg); std::vector<llama_chat_msg> chat_new(past_msg);
// if the past_msg ends with a newline, we must preserve it in the formatted version // if the past_msg ends with a newline, we must preserve it in the formatted version
if (add_ass && !fmt_past_msg.empty() && fmt_past_msg.back() == '\n') { if (add_ass && !fmt_past_msg.empty() && fmt_past_msg.back() == '\n') {

1
examples/main/main.cpp
View file

@ -125,6 +125,7 @@ static std::string chat_add_and_format(struct llama_model * model, std::vector<l
auto formatted = llama_chat_format_single( auto formatted = llama_chat_format_single(
model, g_params->chat_template, chat_msgs, new_msg, role == "user"); model, g_params->chat_template, chat_msgs, new_msg, role == "user");
chat_msgs.push_back({role, content}); chat_msgs.push_back({role, content});
LOG("formatted: %s\n", formatted.c_str());
return formatted; return formatted;
} }

6
include/llama.h
View file

@ -529,12 +529,16 @@ extern "C" {
struct llama_lora_adapter * adapter, struct llama_lora_adapter * adapter,
float scale); float scale);
// Remove a LoRA adapter from given context // Remove a specific LoRA adapter from given context
// Return -1 if the adapter is not present in the context // Return -1 if the adapter is not present in the context
LLAMA_API int32_t llama_lora_adapter_remove( LLAMA_API int32_t llama_lora_adapter_remove(
struct llama_context * ctx, struct llama_context * ctx,
struct llama_lora_adapter * adapter); struct llama_lora_adapter * adapter);
// Remove all LoRA adapters from given context
LLAMA_API void llama_lora_adapter_clear(
struct llama_context * ctx);
// Manually free a LoRA adapter // Manually free a LoRA adapter
// Note: loaded adapters will be free when the associated model is deleted // Note: loaded adapters will be free when the associated model is deleted
LLAMA_API void llama_lora_adapter_free(struct llama_lora_adapter * adapter); LLAMA_API void llama_lora_adapter_free(struct llama_lora_adapter * adapter);

4
src/llama.cpp
View file

@ -16246,6 +16246,10 @@ int32_t llama_lora_adapter_remove(
return -1; return -1;
} }
void llama_lora_adapter_clear(struct llama_context * ctx) {
ctx->lora_adapters.clear();
}
void llama_lora_adapter_free(struct llama_lora_adapter * adapter) { void llama_lora_adapter_free(struct llama_lora_adapter * adapter) {
delete adapter; delete adapter;
} }