infill : add new example + extend server API (#3296)

* vvhg-code-infill (#1)

* infill in separate example (#2)

* reverted changes to main and added infill example

* cleanup

* naming improvement

* make : add missing blank line

* fix missing semicolon

* brought infill up to current main code

* cleanup

---------

Co-authored-by: Cebtenzzre <cebtenzzre@gmail.com>

1 files changed, 20 insertions, 0 deletions

diff --git a/llama.cpp b/llama.cpp
index bff17135..3a0b2c30 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1076,6 +1076,10 @@ struct llama_vocab {
     id special_pad_id = -1;
 
     id linefeed_id = 13;
+    id special_prefix_id = 32007;
+    id special_middle_id = 32009;
+    id special_suffix_id = 32008;
+    id special_eot_id = 32010;
 
     int find_bpe_rank(std::string token_left, std::string token_right) const {
         replace_all(token_left,  " ",  "\u0120");
@@ -7489,6 +7493,22 @@ llama_token llama_token_eos(const struct llama_context * ctx) {
 llama_token llama_token_nl(const struct llama_context * ctx) {
     return ctx->model.vocab.linefeed_id;
 }
+llama_token llama_token_prefix(const struct llama_context * ctx) {
+    return ctx->model.vocab.special_prefix_id;
+}
+
+llama_token llama_token_middle(const struct llama_context * ctx) {
+    return ctx->model.vocab.special_middle_id;
+}
+
+llama_token llama_token_suffix(const struct llama_context * ctx) {
+    return ctx->model.vocab.special_suffix_id;
+}
+
+llama_token llama_token_eot(const struct llama_context * ctx) {
+    return ctx->model.vocab.special_eot_id;
+}
+
 
 int llama_tokenize(
     const struct llama_model * model,