Add parameter to ignore end of text token

main.cpp

@@ -27,6 +27,8 @@
 #define ANSI_COLOR_RESET   "\x1b[0m"
 #define ANSI_BOLD          "\x1b[1m"
 
+static const int EOS_TOKEN_ID = 2;
+
 // determine number of model parts based on the dimension
 static const std::map<int, int> LLAMA_N_PARTS = {
     { 4096, 1 },
@@ -949,6 +951,11 @@ int main(int argc, char ** argv) {
             {
                 const int64_t t_start_sample_us = ggml_time_us();
 
+                if (params.ignore_eos) {
+                    // set the logit of the eos token to zero to avoid sampling it
+                    logits[logits.size() - n_vocab + EOS_TOKEN_ID] = 0;
+                }
+
                 id = llama_sample_top_p_top_k(vocab, logits.data() + (logits.size() - n_vocab), last_n_tokens, repeat_penalty, top_k, top_p, temp, rng);
 
                 last_n_tokens.erase(last_n_tokens.begin());
@@ -1037,7 +1044,7 @@ int main(int argc, char ** argv) {
         }
 
         // end of text token
-        if (embd.back() == 2) {
+        if (embd.back() == EOS_TOKEN_ID) {
             fprintf(stderr, " [end of text]\n");
             break;
         }

utils.cpp

@@ -60,6 +60,8 @@ bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
             params.use_color = true;
         } else if (arg == "-r" || arg == "--reverse-prompt") {
             params.antiprompt = argv[++i];
+        } else if (arg == "--ignore-eos") {
+            params.ignore_eos = true;
         } else if (arg == "-h" || arg == "--help") {
             gpt_print_usage(argc, argv, params);
             exit(0);
@@ -95,6 +97,7 @@ void gpt_print_usage(int argc, char ** argv, const gpt_params & params) {
     fprintf(stderr, "  --repeat_last_n N     last n tokens to consider for penalize (default: %d)\n", params.repeat_last_n);
     fprintf(stderr, "  --repeat_penalty N    penalize repeat sequence of tokens (default: %.1f)\n", params.repeat_penalty);
     fprintf(stderr, "  -c N, --ctx_size N    size of the prompt context (default: %d)\n", params.n_ctx);
+    fprintf(stderr, "  --ignore-eos          ignore end of stream token and continue generating\n");
     fprintf(stderr, "  --temp N              temperature (default: %.1f)\n", params.temp);
     fprintf(stderr, "  -b N, --batch_size N  batch size for prompt processing (default: %d)\n", params.n_batch);
     fprintf(stderr, "  -m FNAME, --model FNAME\n");

utils.h

@@ -35,6 +35,7 @@ struct gpt_params {
     bool interactive = false; // interactive mode
     bool interactive_start = false; // reverse prompt immediately
     std::string antiprompt = ""; // string upon seeing which more user input is prompted
+    bool ignore_eos = false; // do not stop generating after eos
 };
 
 bool gpt_params_parse(int argc, char ** argv, gpt_params & params);