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

plamo fixed

Browse source
This commit is contained in:
Concedo 2025-05-02 22:46:17 +08:00
commit 17cbf9fd49
2 changed files with 11 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

10
src/llama-chat.cpp
View file

@ -447,7 +447,7 @@ int32_t llm_chat_apply_template(
if (add_ass) { if (add_ass) {
ss << "<|assistant|>"; ss << "<|assistant|>";
} }
} else if (tmpl == LLM_CHAT_TEMPLATE_CHATGLM_4 || tmpl == LLM_CHAT_TEMPLATE_GLMEDGE) { } else if (tmpl == LLM_CHAT_TEMPLATE_CHATGLM_4) {
ss << "[gMASK]" << "<sop>"; ss << "[gMASK]" << "<sop>";
for (auto message : chat) { for (auto message : chat) {
std::string role(message->role); std::string role(message->role);
@ -456,6 +456,14 @@ int32_t llm_chat_apply_template(
if (add_ass) { if (add_ass) {
ss << "<|assistant|>\n"; ss << "<|assistant|>\n";
} }
} else if (tmpl == LLM_CHAT_TEMPLATE_GLMEDGE) {
for (auto message : chat) {
std::string role(message->role);
ss << "<|" << role << "|>" << "\n" << message->content;
}
if (add_ass) {
ss << "<|assistant|>";
}
} else if (tmpl == LLM_CHAT_TEMPLATE_MINICPM) { } else if (tmpl == LLM_CHAT_TEMPLATE_MINICPM) {
// MiniCPM-3B-OpenHermes-2.5-v2-GGUF // MiniCPM-3B-OpenHermes-2.5-v2-GGUF
for (auto message : chat) { for (auto message : chat) {

4
src/llama-model.cpp
View file

@ -13329,8 +13329,6 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
case LLM_ARCH_DECI: case LLM_ARCH_DECI:
case LLM_ARCH_BAICHUAN: case LLM_ARCH_BAICHUAN:
case LLM_ARCH_STARCODER: case LLM_ARCH_STARCODER:
case LLM_ARCH_PLAMO:
case LLM_ARCH_ORION:
case LLM_ARCH_INTERNLM2: case LLM_ARCH_INTERNLM2:
case LLM_ARCH_MINICPM: case LLM_ARCH_MINICPM:
case LLM_ARCH_XVERSE: case LLM_ARCH_XVERSE:
@ -13368,6 +13366,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
case LLM_ARCH_PHI2: case LLM_ARCH_PHI2:
case LLM_ARCH_PHI3: case LLM_ARCH_PHI3:
case LLM_ARCH_PHIMOE: case LLM_ARCH_PHIMOE:
case LLM_ARCH_PLAMO:
case LLM_ARCH_GEMMA: case LLM_ARCH_GEMMA:
case LLM_ARCH_GEMMA2: case LLM_ARCH_GEMMA2:
case LLM_ARCH_GEMMA3: case LLM_ARCH_GEMMA3:
@ -13375,6 +13374,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
case LLM_ARCH_OPENELM: case LLM_ARCH_OPENELM:
case LLM_ARCH_GPTNEOX: case LLM_ARCH_GPTNEOX:
case LLM_ARCH_CODESHELL: case LLM_ARCH_CODESHELL:
case LLM_ARCH_ORION:
case LLM_ARCH_NEMOTRON: case LLM_ARCH_NEMOTRON:
case LLM_ARCH_EXAONE: case LLM_ARCH_EXAONE:
case LLM_ARCH_MINICPM3: case LLM_ARCH_MINICPM3: