Update docs for yarn_ext_factor <0.0 as unspecified instead of NaN

common/common.cpp

@@ -792,7 +792,7 @@ void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
     printf("  --rope-freq-base N    RoPE base frequency, used by NTK-aware scaling (default: loaded from model)\n");
     printf("  --rope-freq-scale N   RoPE frequency scaling factor, expands context by a factor of 1/N\n");
     printf("  --yarn-orig-ctx N     YaRN: original context size of model (default: 0 = model training context size)\n");
-    printf("  --yarn-ext-factor N   YaRN: extrapolation mix factor (default: 1.0, 0.0 = full interpolation)\n");
+    printf("  --yarn-ext-factor N   YaRN: extrapolation mix factor (default: 1.0, 0.0 = full interpolation, <0.0 = not specified)\n");
     printf("  --yarn-attn-factor N  YaRN: scale sqrt(t) or attention magnitude (default: 1.0)\n");
     printf("  --yarn-beta-slow N    YaRN: high correction dim or alpha (default: %.1f)\n", params.yarn_beta_slow);
     printf("  --yarn-beta-fast N    YaRN: low correction dim or beta (default: %.1f)\n", params.yarn_beta_fast);

llama.h

@@ -185,7 +185,7 @@ extern "C" {
         // ref: https://github.com/ggerganov/llama.cpp/pull/2054
         float    rope_freq_base;   // RoPE base frequency, 0 = from model
         float    rope_freq_scale;  // RoPE frequency scaling factor, 0 = from model
-        float    yarn_ext_factor;  // YaRN extrapolation mix factor, NaN = from model
+        float    yarn_ext_factor;  // YaRN extrapolation mix factor, <0.0 = from model
         float    yarn_attn_factor; // YaRN magnitude scaling factor
         float    yarn_beta_fast;   // YaRN low correction dim
         float    yarn_beta_slow;   // YaRN high correction dim