fix token limit count

matrix_pygmalion_bot/ai/runpod_pygmalion.py

@@ -114,13 +114,15 @@ async def get_full_prompt(simple_prompt: str, bot, chat_history):
             break
         if chat_item.message["en"].startswith('!'):
             continue
+        if chat_item.message["en"].startswith('<ERROR>'):
+            continue
         #if chat_item.message["en"] == bot.greeting:
         #    continue
         if chat_item.num_tokens == None:
             chat_item.num_tokens = await num_tokens("{}: {}".format(chat_item.user_name, chat_item.message["en"]))
         # TODO: is it MAX_TOKENS or MAX_TOKENS - max_new_tokens??
         logger.debug(f"History: " + str(chat_item) + " [" + str(chat_item.num_tokens) + "]")
-        if total_num_tokens < (MAX_TOKENS - max_new_tokens):
+        if total_num_tokens + chat_item.num_tokens < MAX_TOKENS - max_new_tokens:
             visible_history.append(chat_item)
             total_num_tokens += chat_item.num_tokens
         else: