From 32cf02487e19649a2413ec696fb59c3a1b28a3c0 Mon Sep 17 00:00:00 2001
From: Concedo <39025047+LostRuins@users.noreply.github.com>
Date: Sat, 23 Sep 2023 23:32:00 +0800
Subject: [PATCH] colab use mmq, update lite and ver

---
 colab.ipynb  | 4 ++--
 klite.embd   | 6 ++++--
 koboldcpp.py | 5 ++++-
 3 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/colab.ipynb b/colab.ipynb
index 81f6b0dd1..da9bef59a 100644
--- a/colab.ipynb
+++ b/colab.ipynb
@@ -39,7 +39,7 @@
         "#@title <b>v-- Enter your model below and then click this to start Koboldcpp</b>\n",
         "\n",
         "Model = \"https://huggingface.co/TheBloke/Airoboros-L2-13B-2.2-GGUF/resolve/main/airoboros-l2-13b-2.2.Q4_K_M.gguf\" #@param [\"\"]{allow-input: true}\n",
-        "Layers = 40 #@param [40]{allow-input: true}\n",
+        "Layers = 43 #@param [43]{allow-input: true}\n",
         "\n",
         "%cd /content\n",
         "!git clone https://github.com/LostRuins/koboldcpp\n",
@@ -52,7 +52,7 @@
         "!nohup ./cloudflared-linux-amd64 tunnel --url http://localhost:5001 &\n",
         "!sleep 10\n",
         "!cat nohup.out\n",
-        "!python koboldcpp.py model.ggml --stream --usecublas 0 --gpulayers $Layers --hordeconfig concedo\n"
+        "!python koboldcpp.py model.ggml --stream --usecublas 0 mmq --gpulayers $Layers --hordeconfig concedo\n"
       ]
     }
   ]
diff --git a/klite.embd b/klite.embd
index 9a4a31f6a..1294ddf28 100644
--- a/klite.embd
+++ b/klite.embd
@@ -9183,6 +9183,7 @@ Current version: 68
 				newbodystr = newbodystr.replace(new RegExp(`${contextDict[`${role}Open`]}([^]*?)${contextDict.closeTag}`, 'g'), (match, p) => {
 					let replacedText = match.replace(/<[^>]*>/g, (htmlPart) => { internalHTMLparts.push(htmlPart); return `<internal_html_${internalHTMLparts.length - 1}>`; });
 					replacedText = replacedText.replace(/\*(\S[^*]+\S)\*/g, wrapperSpan(styleRole, 'action')); 		// Apply the actions style to *actions*.
+					replacedText = replacedText.replace(/“(.*?)”/g, wrapperSpan(styleRole, 'speech')); 	// Apply the speech style to "speech".
 					replacedText = replacedText.replace(/&quot;(.*?)&quot;/g, wrapperSpan(styleRole, 'speech')); 	// Apply the speech style to "speech".
 					return replacedText;
 				});
@@ -9199,8 +9200,9 @@ Current version: 68
 		}
 		function wrapperSpan(role, type) {
 			let fontStyle = type=='action'?'italic':'normal';
-			let injectQuotes = type=='speech'?'&quot;':'';
-			let textCol = as[`${type}_tcolor_${role}`]; return `<span style='color: ${textCol}; font-style: ${fontStyle}; font-weight: normal'>${injectQuotes}$1${injectQuotes}</span>`;
+			let injectQuotes1 = type=='speech'?'“':'';
+			let injectQuotes2 = type=='speech'?'”':'';
+			let textCol = as[`${type}_tcolor_${role}`]; return `<span style='color: ${textCol}; font-style: ${fontStyle}; font-weight: normal'>${injectQuotes1}$1${injectQuotes2}</span>`;
 		}
 		function image(role) {
 			if (!as[`${role}_portrait`] || as.border_style == 'None' || role == 'sys') { return ''; }
diff --git a/koboldcpp.py b/koboldcpp.py
index 67fa95441..4fb6d7410 100755
--- a/koboldcpp.py
+++ b/koboldcpp.py
@@ -354,7 +354,7 @@ maxhordelen = 256
 modelbusy = threading.Lock()
 requestsinqueue = 0
 defaultport = 5001
-KcppVersion = "1.44.2"
+KcppVersion = "1.45"
 showdebug = True
 showsamplerwarning = True
 showmaxctxwarning = True
@@ -556,6 +556,9 @@ class ServerRequestHandler(http.server.SimpleHTTPRequestHandler):
             lastc = handle.get_last_token_count()
             stopreason = handle.get_last_stop_reason()
             response_body = (json.dumps({"last_process":lastp,"last_eval":laste,"last_token_count":lastc, "stop_reason":stopreason, "queue":requestsinqueue, "idle":(0 if modelbusy.locked() else 1)}).encode())
+        elif self.path.endswith(('/api')) or self.path.endswith(('/api/v1')):
+            response_body = (json.dumps({"result":"KoboldCpp partial API reference can be found at https://link.concedo.workers.dev/koboldapi"}).encode())
+
 
         if response_body is None:
             self.send_response(404)