mod: chat template add support for deepseek-r1 series

Author: swordow

src/llama-chat.cpp

@@ -48,6 +48,7 @@ static const std::map<std::string, llm_chat_template> LLM_CHAT_TEMPLATES = {
     { "deepseek",          LLM_CHAT_TEMPLATE_DEEPSEEK          },
     { "deepseek2",         LLM_CHAT_TEMPLATE_DEEPSEEK_2        },
     { "deepseek3",         LLM_CHAT_TEMPLATE_DEEPSEEK_3        },
+    { "deepseek-r1",       LLM_CHAT_TEMPLATE_DEEPSEEK_R1       },
     { "command-r",         LLM_CHAT_TEMPLATE_COMMAND_R         },
     { "llama3",            LLM_CHAT_TEMPLATE_LLAMA_3           },
     { "chatglm3",          LLM_CHAT_TEMPLATE_CHATGML_3         },
@@ -154,6 +155,8 @@ llm_chat_template llm_chat_detect_template(const std::string & tmpl) {
         return LLM_CHAT_TEMPLATE_MINICPM;
     } else if (tmpl_contains("'Assistant: ' + message['content'] + eos_token")) {
         return LLM_CHAT_TEMPLATE_DEEPSEEK_2;
+    } else if (tmpl_contains(LU8("<｜Assistant｜>")) && tmpl_contains(LU8("<｜User｜>")) && tmpl_contains(LU8("<｜end▁of▁sentence｜>")) && tmpl_contains(LU8("<｜Assistant｜><think>\\n"))) {
+        return LLM_CHAT_TEMPLATE_DEEPSEEK_R1;
     } else if (tmpl_contains(LU8("<｜Assistant｜>")) && tmpl_contains(LU8("<｜User｜>")) && tmpl_contains(LU8("<｜end▁of▁sentence｜>"))) {
         return LLM_CHAT_TEMPLATE_DEEPSEEK_3;
     } else if (tmpl_contains("[|system|]") && tmpl_contains("[|assistant|]") && tmpl_contains("[|endofturn|]")) {
@@ -492,7 +495,24 @@ int32_t llm_chat_apply_template(
         if (add_ass) {
             ss << LU8("<｜Assistant｜>");
         }
-    } else if (tmpl == LLM_CHAT_TEMPLATE_EXAONE_3) {
+    } else if (tmpl == LLM_CHAT_TEMPLATE_DEEPSEEK_R1) {
+        // DeepSeek-R1
+        ss << "<｜begin▁of▁sentence｜>";
+        for (auto message : chat) {
+            std::string role(message->role);
+            if (role == "system") {
+                ss << message->content;
+            } else if (role == "user") {
+                ss << LU8("<｜User｜>") << message->content;
+            } else if (role == "assistant") {
+                ss << LU8("<｜Assistant｜>") << message->content << LU8("<｜end▁of▁sentence｜>");
+            }
+        }
+        if (add_ass) {
+            ss << LU8("<｜Assistant｜><think>\n");
+        }
+    }
+    else if (tmpl == LLM_CHAT_TEMPLATE_EXAONE_3) {
         // ref: https://huggingface.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct/discussions/8#66bae61b1893d14ee8ed85bb
         // EXAONE-3.0-7.8B-Instruct
         for (auto message : chat) {

src/llama-chat.h

@@ -27,6 +27,7 @@ enum llm_chat_template {
     LLM_CHAT_TEMPLATE_DEEPSEEK,
     LLM_CHAT_TEMPLATE_DEEPSEEK_2,
     LLM_CHAT_TEMPLATE_DEEPSEEK_3,
+    LLM_CHAT_TEMPLATE_DEEPSEEK_R1,
     LLM_CHAT_TEMPLATE_COMMAND_R,
     LLM_CHAT_TEMPLATE_LLAMA_3,
     LLM_CHAT_TEMPLATE_CHATGML_3,

tests/test-chat-template.cpp

@@ -270,6 +270,15 @@ int main(void) {
             /* .bos_token= */ "",
             /* .eos_token= */ "",
         },
+        {
+            /* .name= */ "deepseek-ai/DeepSeek-R1",
+            /* .template_str= */ "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% set ns = namespace(is_first=false, is_tool=false, is_output_first=true, system_prompt='') %}{%- for message in messages %}{%- if message['role'] == 'system' %}{% set ns.system_prompt = message['content'] %}{%- endif %}{%- endfor %}{{bos_token}}{{ns.system_prompt}}{%- for message in messages %}{%- if message['role'] == 'user' %}{%- set ns.is_tool = false -%}{{'<｜User｜>' + message['content']}}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is none %}{%- set ns.is_tool = false -%}{%- for tool in message['tool_calls']%}{%- if not ns.is_first %}{{'<｜Assistant｜><｜tool▁calls▁begin｜><｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '\\n' + '```json' + '\\n' + tool['function']['arguments'] + '\\n' + '```' + '<｜tool▁call▁end｜>'}}{%- set ns.is_first = true -%}{%- else %}{{'\\n' + '<｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '\\n' + '```json' + '\\n' + tool['function']['arguments'] + '\\n' + '```' + '<｜tool▁call▁end｜>'}}{{'<｜tool▁calls▁end｜><｜end▁of▁sentence｜>'}}{%- endif %}{%- endfor %}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is not none %}{%- if ns.is_tool %}{{'<｜tool▁outputs▁end｜>' + message['content'] + '<｜end▁of▁sentence｜>'}}{%- set ns.is_tool = false -%}{%- else %}{% set content = message['content'] %}{% if '</think>' in content %}{% set content = content.split('</think>')[-1] %}{% endif %}{{'<｜Assistant｜>' + content + '<｜end▁of▁sentence｜>'}}{%- endif %}{%- endif %}{%- if message['role'] == 'tool' %}{%- set ns.is_tool = true -%}{%- if ns.is_output_first %}{{'<｜tool▁outputs▁begin｜><｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- set ns.is_output_first = false %}{%- else %}{{'\\n<｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- endif %}{%- endif %}{%- endfor -%}{% if ns.is_tool %}{{'<｜tool▁outputs▁end｜>'}}{% endif %}{% if add_generation_prompt and not ns.is_tool %}{{'<｜Assistant｜><think>\\n'}}{% endif %}",
+            /* .expected_output= */ "<｜begin▁of▁sentence｜>You are a helpful assistant<｜User｜>Hello<｜Assistant｜>Hi there<｜end▁of▁sentence｜><｜User｜>Who are you<｜Assistant｜>   I am an assistant   <｜end▁of▁sentence｜><｜User｜>Another question<｜Assistant｜><think>\n",
+            /* .expected_output_jinja= */ "<｜begin▁of▁sentence｜>You are a helpful assistant<｜User｜>Hello<｜Assistant｜>Hi there<｜end▁of▁sentence｜><｜User｜>Who are you<｜Assistant｜>   I am an assistant   <｜end▁of▁sentence｜><｜User｜>Another question<｜Assistant｜><think>\n",
+            /* .bos_token= */ "<｜begin▁of▁sentence｜>",
+            /* .eos_token= */ "<｜end▁of▁sentence｜>",
+            /* .supported_with_jinja= */ true, // Requires additional_special_tokens as extra context
+        },
     };
     std::vector<char> formatted_chat(1024);
     int32_t res;