More robust unprintable character check

main.cpp

@@ -10,6 +10,7 @@
 #include <map>
 #include <string>
 #include <vector>
+#include <unordered_set>
 
 // determine number of model parts based on the dimension
 static const std::map<int, int> LLAMA_N_PARTS = {
@@ -123,6 +124,9 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
     }
 
     // load vocab
+    
+    std::unordered_set<std::string> unprintable_characters = {"", "<EFBFBD>", "<EFBFBD><EFBFBD>"};
+
     {
         const int32_t n_vocab = model.hparams.n_vocab;
 
@@ -132,6 +136,7 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
             return false;
         }
 
+        
         std::string word;
         for (int i = 0; i < n_vocab; i++) {
             uint32_t len;
@@ -139,10 +144,8 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
 
             word.resize(len);
             fin.read((char *) word.data(), len);
-            
-            if(i >= 131 && i <= 258) {
-                // seems to be unprintable characters list in this range
-                // TODO maybe they are supposed to be byte reversed or some magic
+
+            if(unprintable_characters.find(word) != unprintable_characters.end()) {
                 continue;
             }
 
@@ -798,7 +801,7 @@ int main(int argc, char ** argv) {
         printf("%6d -> '%s'\n", embd_inp[i], vocab.id_to_token.at(embd_inp[i]).c_str());
     }
     printf("\n");
-    printf("sampling parameters: temp = %f, top_k = %d, top_p = %f\n", params.temp, params.top_k, params.top_p);
+    printf("sampling parameters: temp = %f, top_k = %d, top_p = %f, repeat_last_n = %i, repeat_penalty = %f\n", params.temp, params.top_k, params.top_p, params.repeat_last_n, params.repeat_penalty);
     printf("\n\n");
 
     std::vector<gpt_vocab::id> embd;
@@ -807,6 +810,10 @@ int main(int argc, char ** argv) {
     size_t mem_per_token = 0;
     llama_eval(model, params.n_threads, 0, { 0, 1, 2, 3 }, logits, mem_per_token);
 
+    int last_n_size = params.repeat_last_n;
+    std::vector<gpt_vocab::id> last_n_tokens(last_n_size);
+    std::fill(last_n_tokens.begin(), last_n_tokens.end(), 0);
+
     for (int i = embd.size(); i < embd_inp.size() + params.n_predict; i++) {
         // predict
         if (embd.size() > 0) {
@@ -827,6 +834,7 @@ int main(int argc, char ** argv) {
             // sample next token
             const float top_p = params.top_p;
             const float temp  = params.temp;
+            const float repeat_penalty = params.repeat_penalty;
 
             const int n_vocab = model.hparams.n_vocab;
 
@@ -835,7 +843,10 @@ int main(int argc, char ** argv) {
             {
                 const int64_t t_start_sample_us = ggml_time_us();
 
-                id = llama_sample_top_p(vocab, logits.data() + (logits.size() - n_vocab), top_p, temp, rng);
+                id = llama_sample_top_p(vocab, logits.data() + (logits.size() - n_vocab), last_n_tokens, repeat_penalty, top_p, temp, rng);
+
+                last_n_tokens.erase(last_n_tokens.begin());
+                last_n_tokens.push_back(id);
 
                 t_sample_us += ggml_time_us() - t_start_sample_us;
             }
@@ -846,6 +857,8 @@ int main(int argc, char ** argv) {
             // if here, it means we are still processing the input prompt
             for (int k = i; k < embd_inp.size(); k++) {
                 embd.push_back(embd_inp[k]);
+                last_n_tokens.erase(last_n_tokens.begin());
+                last_n_tokens.push_back(embd_inp[k]);
                 if (embd.size() > params.n_batch) {
                     break;
                 }