config reasoning_content in webui and change default to auto

Author: firecoperana

common/common.h

@@ -253,7 +253,7 @@ struct gpt_params {
     bool use_jinja = false;                                                                                 // NOLINT
     std::string system_prompt = "";
     bool enable_chat_template = true;
-    common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_NONE;
+    common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_AUTO;
     int reasoning_budget = -1;
     bool prefill_assistant = true;
 

examples/server/public/index.html.gz

@@ -16,6 +16,7 @@ export const CONFIG_DEFAULT = {
   showTokensPerSecond: false,
   showThoughtInProgress: false,
   excludeThoughtOnReq: true,
+  reasoning_format: 'auto',
   // make sure these default values are in sync with `common.h`
   samplers: 'dkypmxnt',
   temperature: 0.8,
@@ -42,6 +43,7 @@ export const CONFIG_DEFAULT = {
   pyIntepreterEnabled: false,
 };
 export const CONFIG_INFO: Record<string, string> = {
+  reasoning_format : 'Specify how to parse reasoning content. none: reasoning content in content block. auto: reasoning content in reasoning_content. ',
   apiKey: 'Set the API Key if you are using --api-key option for the server.',
   systemMessage: 'The starting message that defines how model should behave.',
   samplers:

examples/server/webui/dist/index.html

@@ -22,6 +22,7 @@ import toast from 'react-hot-toast'
 type SettKey = keyof typeof CONFIG_DEFAULT;
 
 const BASIC_KEYS: SettKey[] = [
+  'reasoning_format',
   'temperature',
   'top_k',
   'top_p',

examples/server/webui/src/Config.ts

@@ -215,7 +215,7 @@ export const AppContextProvider = ({
         messages,
         stream: true,
         cache_prompt: true,
-        reasoning_format: 'none',
+        reasoning_format: config.reasoning_format===''?'auto':config.reasoning_format,
         samplers: config.samplers,
         temperature: config.temperature,
         dynatemp_range: config.dynatemp_range,
@@ -226,7 +226,7 @@ export const AppContextProvider = ({
         typical_p: config.typical_p,
         xtc_probability: config.xtc_probability,
         xtc_threshold: config.xtc_threshold,
-		top_n_sigma: config.top_n_sigma,
+		    top_n_sigma: config.top_n_sigma,
         repeat_last_n: config.repeat_last_n,
         repeat_penalty: config.repeat_penalty,
         presence_penalty: config.presence_penalty,