Browse Source

runner: Flush pending responses before returning

If there are any pending reponses (such as from potential stop
tokens) then we should send them back before ending the sequence.
Otherwise, we can be missing tokens at the end of a response.

Fixes #6707
Jesse Gross 8 months ago
parent
commit
93ac3760cb
1 changed files with 33 additions and 27 deletions
  1. 33 27
      llm/ext_server/server.cpp

+ 33 - 27
llm/ext_server/server.cpp

@@ -913,7 +913,9 @@ struct llama_server_context
         slot.sampled = result.tok;
 
         // search stop word and delete it
-        slot.generated_text += token_str;
+        if (!llama_token_is_eog(model, result.tok))
+            slot.generated_text += token_str;
+
         slot.has_next_token = true;
 
         if (slot.ctx_sampling->params.use_penalty_prompt_tokens && result.tok != -1)
@@ -954,30 +956,36 @@ struct llama_server_context
         if (!incomplete)
         {
             size_t pos = std::min(slot.n_sent_text, slot.generated_text.size());
-            const std::string str_test = slot.generated_text.substr(pos);
-            bool is_stop_full = false;
-            size_t stop_pos = find_stopping_strings(str_test, token_str.size(), STOP_FULL, slot);
-            if (stop_pos != std::string::npos)
-            {
-                is_stop_full = true;
-                slot.generated_text.erase(
-                    slot.generated_text.begin() + pos + stop_pos,
-                    slot.generated_text.end());
-                pos = std::min(slot.n_sent_text, slot.generated_text.size());
-            }
-            else
-            {
-                is_stop_full = false;
-                stop_pos = find_stopping_strings(str_test, token_str.size(), STOP_PARTIAL, slot);
-            }
 
-            // check if there is any token to predict
-            if (stop_pos == std::string::npos || (!slot.has_next_token && !is_stop_full && stop_pos > 0))
-            {
-                // no send the stop word in the response
-                result.text_to_send = slot.generated_text.substr(pos, std::string::npos);
-                slot.n_sent_text += result.text_to_send.size();
-                // add the token to slot queue and cache
+            if (!llama_token_is_eog(model, result.tok)) {
+                const std::string str_test = slot.generated_text.substr(pos);
+                bool is_stop_full = false;
+                size_t stop_pos = find_stopping_strings(str_test, token_str.size(), STOP_FULL, slot);
+                if (stop_pos != std::string::npos)
+                {
+                    is_stop_full = true;
+                    slot.generated_text.erase(
+                        slot.generated_text.begin() + pos + stop_pos,
+                        slot.generated_text.end());
+                    pos = std::min(slot.n_sent_text, slot.generated_text.size());
+                }
+                else
+                {
+                    is_stop_full = false;
+                    stop_pos = find_stopping_strings(str_test, token_str.size(), STOP_PARTIAL, slot);
+                }
+
+                // check if there is any token to predict
+                if (stop_pos == std::string::npos || (!slot.has_next_token && !is_stop_full && stop_pos > 0))
+                {
+                    // no send the stop word in the response
+                    result.text_to_send = slot.generated_text.substr(pos, std::string::npos);
+                    slot.n_sent_text += result.text_to_send.size();
+                    // add the token to slot queue and cache
+                }
+            } else {
+                    result.text_to_send = slot.generated_text.substr(pos, std::string::npos);
+                    slot.n_sent_text += result.text_to_send.size();
             }
 
             if (slot.params.stream)
@@ -1117,9 +1125,7 @@ struct llama_server_context
             {"multimodal", multimodal}
         };
 
-        if (!llama_token_is_eog(model, tkn.tok)) {
-            res.result_json["content"] = tkn.text_to_send;
-        }
+        res.result_json["content"] = tkn.text_to_send;
 
         if (slot.sparams.n_probs > 0)
         {