server : add min_p param (ggml-org#3877)

* Update server.cpp with min_p after it was introduced in ggml-org#3841 * Use spaces instead of tabs * Update index.html.hpp after running deps.sh * Fix test - fix line ending
Sanctum-AI · Nov 23, 2023 · d743fe5 · d743fe5
1 parent a775dfc
commit d743fe5
Show file tree

Hide file tree

Showing 4 changed files with 2,191 additions and 2,171 deletions.
diff --git a/examples/server/README.md b/examples/server/README.md
@@ -122,6 +122,8 @@ node index.js
 
     `top_p`: Limit the next token selection to a subset of tokens with a cumulative probability above a threshold P (default: 0.95).
 
+    `min_p`: The minimum probability for a token to be considered, relative to the probability of the most likely token (default: 0.05).
+
     `n_predict`: Set the maximum number of tokens to predict when generating text. **Note:** May exceed the set limit slightly if the last token is a partial multibyte character. When 0, no tokens will be generated but the prompt is evaluated into the cache. (default: -1, -1 = infinity).
 
     `n_keep`: Specify the number of tokens from the prompt to retain when the context size is exceeded and tokens need to be discarded.