Add top n sigma sampler and other webui fix (#512)

firecoperana · web-flow · commit 7b1a3eece7e3 · 2025-06-12T08:19:26.000+03:00
Co-authored-by: firecoperana &lt;firecoperana&gt;
diff --git a/common/sampling.h b/common/sampling.h
@@ -11,6 +11,7 @@
 
 // sampler types
 enum class llama_sampler_type : char {
+    DRY            ='d',
     TOP_K       = 'k',
     TOP_P       = 'p',
     MIN_P       = 'm',
@@ -53,6 +54,7 @@ typedef struct llama_sampling_params {
         llama_sampler_type::TYPICAL_P,
         llama_sampler_type::TOP_P,
         llama_sampler_type::MIN_P,
+        llama_sampler_type::TOP_N_SIGMA,
         llama_sampler_type::TEMPERATURE
     };
 
diff --git a/examples/server/public/index.html.gz b/examples/server/public/index.html.gz
diff --git a/examples/server/server.cpp b/examples/server/server.cpp
@@ -970,6 +970,9 @@ struct server_context {
         slot.sparams.temp              = json_value(data, "temperature",       default_sparams.temp);
         slot.sparams.dynatemp_range    = json_value(data, "dynatemp_range",    default_sparams.dynatemp_range);
         slot.sparams.dynatemp_exponent = json_value(data, "dynatemp_exponent", default_sparams.dynatemp_exponent);
+        slot.sparams.xtc_probability = json_value(data, "xtc_probability", default_sparams.xtc_probability);
+        slot.sparams.xtc_threshold = json_value(data, "xtc_threshold", default_sparams.xtc_threshold);
+        slot.sparams.top_n_sigma = json_value(data, "top_n_sigma", default_sparams.top_n_sigma);
         slot.sparams.penalty_last_n    = json_value(data, "repeat_last_n",     default_sparams.penalty_last_n);
         slot.sparams.penalty_repeat    = json_value(data, "repeat_penalty",    default_sparams.penalty_repeat);
         slot.sparams.penalty_freq      = json_value(data, "frequency_penalty", default_sparams.penalty_freq);
@@ -1135,17 +1138,17 @@ struct server_context {
         }
 
         {
-            const auto & samplers_sequence = data.find("samplers");
-            if (samplers_sequence != data.end() && samplers_sequence->is_array()) {
-                std::vector<std::string> sampler_names;
-                for (const auto & sampler_name : *samplers_sequence) {
-                    if (sampler_name.is_string()) {
-                        sampler_names.emplace_back(sampler_name);
-                    }
+            const auto samplers = data.find("samplers");
+            if (samplers != data.end()) {
+                if (samplers->is_array()) {
+                    slot.sparams.samplers_sequence = llama_sampling_types_from_names(*samplers, false);
+                }
+                else if (samplers->is_string()) {
+                    slot.sparams.samplers_sequence = llama_sampling_types_from_chars(samplers->get<std::string>());
+                }
+                else {
+                    slot.sparams.samplers_sequence = default_sparams.samplers_sequence;
                 }
-                slot.sparams.samplers_sequence = llama_sampling_types_from_names(sampler_names, false);
-            } else {
-                slot.sparams.samplers_sequence = default_sparams.samplers_sequence;
             }
         }
 
diff --git a/examples/server/webui/dist/index.html b/examples/server/webui/dist/index.html
diff --git a/examples/server/webui/src/Config.ts b/examples/server/webui/src/Config.ts
@@ -17,7 +17,7 @@ export const CONFIG_DEFAULT = {
   showThoughtInProgress: false,
   excludeThoughtOnReq: true,
   // make sure these default values are in sync with `common.h`
-  samplers: 'dkypmxt',
+  samplers: 'dkypmxnt',
   temperature: 0.8,
   dynatemp_range: 0.0,
   dynatemp_exponent: 1.0,
@@ -26,6 +26,7 @@ export const CONFIG_DEFAULT = {
   min_p: 0.05,
   xtc_probability: 0.0,
   xtc_threshold: 0.1,
+  top_n_sigma: 0.0,
   typical_p: 1.0,
   repeat_last_n: 64,
   repeat_penalty: 1.0,
@@ -44,7 +45,7 @@ export const CONFIG_INFO: Record<string, string> = {
   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:
-    'The order at which samplers are applied, in simplified way. Default is "dkypmxt": dry->top_k->typ_p->top_p->min_p->xtc->temperature',
+    'The order at which samplers are applied, in simplified way. Default is "dkypmxt": dry->top_k->typ_p->top_p->min_p->xtc->top_sigma->temperature',
   temperature:
     'Controls the randomness of the generated text by affecting the probability distribution of the output tokens. Higher = more random, lower = more focused.',
   dynatemp_range:
@@ -60,6 +61,8 @@ export const CONFIG_INFO: Record<string, string> = {
     'XTC sampler cuts out top tokens; this parameter controls the chance of cutting tokens at all. 0 disables XTC.',
   xtc_threshold:
     'XTC sampler cuts out top tokens; this parameter controls the token probability that is required to cut that token.',
+  top_n_sigma:
+    'Top-n-sigma sampling filters out low-value tokens by discarding tokens that fall more than n standard deviations below the maximum probability',
   typical_p:
     'Sorts and limits tokens based on the difference between log-probability and entropy.',
   repeat_last_n: 'Last n tokens to consider for penalizing repetition',
diff --git a/examples/server/webui/src/components/Header.tsx b/examples/server/webui/src/components/Header.tsx
@@ -70,7 +70,7 @@ export default function Header() {
         </svg>
       </label>
 
-      <div className="grow text-2xl font-bold ml-2">llama.cpp</div>
+      <div className="grow text-2xl font-bold ml-2">ik_llama.cpp</div>
 
       {/* action buttons (top right) */}
       <div className="flex items-center">
diff --git a/examples/server/webui/src/components/SettingDialog.tsx b/examples/server/webui/src/components/SettingDialog.tsx
@@ -29,6 +29,7 @@ const SAMPLER_KEYS: SettKey[] = [
   'typical_p',
   'xtc_probability',
   'xtc_threshold',
+  'top_n_sigma'
 ];
 const PENALTY_KEYS: SettKey[] = [
   'repeat_last_n',
@@ -196,7 +197,7 @@ const SETTING_SECTIONS: SettingSection[] = [
         label: (
           <>
             Custom JSON config (For more info, refer to{' '}
-            <OpenInNewTab href="https://github.com/ggerganov/llama.cpp/blob/master/examples/server/README.md">
+            <OpenInNewTab href="https://github.com/ikawrakow/ik_llama.cpp/tree/main/examples/server/README.md">
               server documentation
             </OpenInNewTab>
             )
@@ -224,7 +225,7 @@ const SETTING_SECTIONS: SettingSection[] = [
               <br />
               <br />
               If you encounter any problems, create a{' '}
-              <OpenInNewTab href="https://github.com/ggerganov/llama.cpp/issues/new?template=019-bug-misc.yml">
+              <OpenInNewTab href="https://github.com/ikawrakow/ik_llama.cpp/issues/new?template=019-bug-misc.yml">
                 Bug (misc.)
               </OpenInNewTab>{' '}
               report on Github. Please also specify <b>webui/experimental</b> on
@@ -456,11 +457,11 @@ function SettingsModalLongInput({
       <div className="label inline">{label || configKey}</div>
       <textarea
         className="textarea textarea-bordered h-24"
-        placeholder={`Default: ${CONFIG_DEFAULT[configKey] || 'none'}`}
-        value={value}
-        onChange={(e) => onChange(e.target.value)}
-      />
-    </label>
+          placeholder={`Default: ${CONFIG_DEFAULT[configKey] || 'none'}`}
+          value={value}
+          onChange={(e) => onChange(e.target.value)}
+        />
+      </label>
   );
 }
 
@@ -485,7 +486,7 @@ function SettingsModalShortInput({
         <div className="block md:hidden mb-1">
           <b>{label || configKey}</b>
           <br />
-          <p className="text-xs">{helpMsg}</p>
+          <p className="text-xs whitespace-normal">{helpMsg}</p>
         </div>
       )}
       <label className="input input-bordered join-item grow flex items-center gap-2 mb-2">
@@ -494,7 +495,7 @@ function SettingsModalShortInput({
             {label || configKey}
           </div>
           {helpMsg && (
-            <div className="dropdown-content menu bg-base-100 rounded-box z-10 w-64 p-2 shadow mt-4">
+            <div className="dropdown-content menu bg-base-100 rounded-box z-10 w-64 p-2 shadow mt-4 whitespace-normal break-words">
               {helpMsg}
             </div>
           )}
diff --git a/examples/server/webui/src/utils/app.context.tsx b/examples/server/webui/src/utils/app.context.tsx
@@ -200,6 +200,7 @@ 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,
         repeat_last_n: config.repeat_last_n,
         repeat_penalty: config.repeat_penalty,
         presence_penalty: config.presence_penalty,
