Merge branch 'UI2' of https://github.com/ebolam/KoboldAI into UI2

aiserver.py

@@ -2881,13 +2881,17 @@ def lua_compute_context(submission, entries, folders, kwargs):
         force_use_txt=True,
         scan_story=kwargs["scan_story"] if kwargs["scan_story"] != None else True,
     )
-    txt, _, _ = calcsubmitbudget(
+    if koboldai_vars.alt_gen:
-        len(actions),
+       txt, _, _ = koboldai_vars.calc_ai_text()
-        winfo,
+       print("Using Alt Gen: {}".format(tokenizer.decode(txt)))
-        mem,
+    else:
-        anotetxt,
+        txt, _, _ = calcsubmitbudget(
-        actions,
+            len(actions),
-    )
+            winfo,
+            mem,
+            anotetxt,
+            actions,
+        )
     return utils.decodenewlines(tokenizer.decode(txt))
 
 #==================================================================#
@@ -4458,7 +4462,11 @@ def calcsubmit(txt):
  
     # For all transformers models
     if(koboldai_vars.model != "InferKit"):
-        subtxt, min, max = calcsubmitbudget(actionlen, winfo, mem, anotetxt, koboldai_vars.actions, submission=txt)
+        if koboldai_vars.alt_gen:
+            subtxt, min, max  = koboldai_vars.calc_ai_text(submitted_text=txt)
+            print("Using Alt Gen: {}".format(tokenizer.decode(subtxt)))
+        else:
+            subtxt, min, max = calcsubmitbudget(actionlen, winfo, mem, anotetxt, koboldai_vars.actions, submission=txt)
         if(actionlen == 0):
             if(not koboldai_vars.use_colab_tpu and koboldai_vars.model not in ["Colab", "API", "OAI", "TPUMeshTransformerGPTJ", "TPUMeshTransformerGPTNeoX"]):
                 generate(subtxt, min, max, found_entries=found_entries)
@@ -4601,7 +4609,11 @@ def _generate(txt, minimum, maximum, found_entries):
                 txt = utils.decodenewlines(tokenizer.decode(genout[i, -already_generated:]))
                 winfo, mem, anotetxt, _found_entries = calcsubmitbudgetheader(txt, force_use_txt=True, actions=koboldai_vars._actions)
                 found_entries[i].update(_found_entries)
-                txt, _, _ = calcsubmitbudget(len(koboldai_vars._actions), winfo, mem, anotetxt, koboldai_vars._actions, submission=txt)
+                if koboldai_vars.alt_gen:
+                   txt, _, _ = koboldai_vars.calc_ai_text(submitted_text=txt)
+                   print("Using Alt Gen: {}".format(tokenizer.decode(txt)))
+                else:
+                    txt, _, _ = calcsubmitbudget(len(koboldai_vars._actions), winfo, mem, anotetxt, koboldai_vars._actions, submission=txt)
                 encoded.append(torch.tensor(txt, dtype=torch.long, device=genout.device))
             max_length = len(max(encoded, key=len))
             encoded = torch.stack(tuple(torch.nn.functional.pad(e, (max_length - len(e), 0), value=model.config.pad_token_id or model.config.eos_token_id) for e in encoded))
@@ -4998,7 +5010,11 @@ def tpumtjgenerate(txt, minimum, maximum, found_entries=None):
                     txt = utils.decodenewlines(tokenizer.decode(past[i]))
                     winfo, mem, anotetxt, _found_entries = calcsubmitbudgetheader(txt, force_use_txt=True, actions=koboldai_vars._actions)
                     found_entries[i].update(_found_entries)
-                    txt, _, _ = calcsubmitbudget(len(koboldai_vars._actions), winfo, mem, anotetxt, koboldai_vars._actions, submission=txt)
+                    if koboldai_vars.alt_gen:
+                       txt, _, _ = koboldai_vars.calc_ai_text(submitted_text=txt)
+                       print("Using Alt Gen: {}".format(tokenizer.decode(txt)))
+                    else:
+                        txt, _, _ = calcsubmitbudget(len(koboldai_vars._actions), winfo, mem, anotetxt, koboldai_vars._actions, submission=txt)
                     encoded.append(np.array(txt, dtype=np.uint32))
                 max_length = len(max(encoded, key=len))
                 encoded = np.stack(tuple(np.pad(e, (max_length - len(e), 0), constant_values=tpu_mtj_backend.pad_token_id) for e in encoded))

gensettings.py

@@ -383,6 +383,21 @@ gensettingstf = [
     "sub_path": "UI",
     "classname": "user",
     "name": "show_probs"
+ 	},
+    {
+ 	"uitype": "toggle",
+ 	"unit": "bool",
+ 	"label": "Alt Text Generation",
+ 	"id": "alttextgen",
+ 	"min": 0,
+ 	"max": 1,
+ 	"step": 1,
+ 	"default": 0,
+	"tooltip": "Inserts world info entries behind text that first triggers them for better context with unlimited depth",
+    "menu_path": "Settings",
+    "sub_path": "Other",
+    "classname": "system",
+    "name": "alt_gen"
  	}
 ]
 
@@ -601,6 +616,21 @@ gensettingsik =[{
     "sub_path": "UI",
     "classname": "user",
     "name": "show_probs"
+ 	},
+    {
+ 	"uitype": "toggle",
+ 	"unit": "bool",
+ 	"label": "Alt Text Generation",
+ 	"id": "alttextgen",
+ 	"min": 0,
+ 	"max": 1,
+ 	"step": 1,
+ 	"default": 0,
+	"tooltip": "Inserts world info entries behind text that first triggers them for better context with unlimited depth",
+    "menu_path": "Settings",
+    "sub_path": "Other",
+    "classname": "system",
+    "name": "alt_gen"
  	}
 ]
 

koboldai_settings.py

@@ -36,7 +36,6 @@ def process_variable_changes(socketio, classname, name, value, old_value, debug_
                 socketio.emit("var_changed", {"classname": "actions", "name": "Action Count", "old_value": None, "value":value.action_count}, broadcast=True, room="UI_2")
                 
                 for i in range(len(value.actions)):
-                    print(value.actions[i])
                     socketio.emit("var_changed", {"classname": "story", "name": "actions", "old_value": None, "value":{"id": i, "action": value.actions[i]}}, broadcast=True, room="UI_2")
             elif isinstance(value, KoboldWorldInfo):
                 value.send_to_ui()
@@ -109,7 +108,7 @@ class koboldai_vars(object):
     def reset_model(self):
         self._model_settings.reset_for_model_load()
     
-    def calc_ai_text(self):
+    def calc_ai_text(self, submitted_text=""):
         token_budget = self.max_length
         used_world_info = []
         used_tokens = self.sp_length
@@ -180,7 +179,7 @@ class koboldai_vars(object):
         game_text = ""
         used_all_tokens = False
         for i in range(len(self.actions)-1, -1, -1):
-            if len(self.actions) - i == self.andepth:
+            if len(self.actions) - i == self.andepth and self.authornote != "":
                 game_text = "{}{}".format(self.authornotetemplate.replace("<|>", self.authornote), game_text)
             if self.actions.actions[i]["Selected Text Length"]+used_tokens <= token_budget and not used_all_tokens:
                 used_tokens += self.actions.actions[i]["Selected Text Length"]
@@ -212,7 +211,7 @@ class koboldai_vars(object):
                 used_all_tokens = True
              
         #if we don't have enough actions to get to author's note depth then we just add it right before the game text
-        if len(self.actions) < self.andepth:
+        if len(self.actions) < self.andepth and self.authornote != "":
             game_text = "{}{}".format(self.authornotetemplate.replace("<|>", self.authornote), game_text)
             
         if not self.useprompt:
@@ -244,7 +243,11 @@ class koboldai_vars(object):
             text += self.prompt
         
         text += game_text
-        return text
+        if self.tokenizer is None:
+            tokens = []
+        else:
+            tokens = self.tokenizer.encode(text)
+        return tokens, used_tokens, used_tokens+self.genamt
     
     def __setattr__(self, name, value):
         if name[0] == "_" or name == "tokenizer":
@@ -691,6 +694,7 @@ class system_settings(settings):
         self.full_determinism = False  # Whether or not full determinism is enabled
         self.seed_specified = False  # Whether or not the current RNG seed was specified by the user (in their settings file)
         self.seed        = None   # The current RNG seed (as an int), or None if unknown
+        self.alt_gen = False # Use the calc_ai_text method for generating text to go to the AI
         
         
     def __setattr__(self, name, value):
@@ -1040,6 +1044,8 @@ class KoboldStoryRegister(object):
         if action_id in self.actions:
             old_options = self.actions[action_id]["Options"]
             if option_number < len(self.actions[action_id]["Options"]):
+                if "Probabilities" not in self.actions[action_id]["Options"][option_number]:
+                    self.actions[action_id]["Options"][option_number]["Probabilities"] = []
                 self.actions[action_id]["Options"][option_number]['Probabilities'].append(probabilities)
                 process_variable_changes(self.socketio, "story", 'actions', {"id": action_id, 'action':  self.actions[action_id]}, None)
     
@@ -1329,8 +1335,8 @@ class KoboldWorldInfo(object):
         
     def reset_used_in_game(self):
         for key in self.world_info:
-            if self.world_info[key]["used_in_game"] != constant:
+            if self.world_info[key]["used_in_game"] != self.world_info[key]["constant"]:
-                self.world_info[key]["used_in_game"] = constant
+                self.world_info[key]["used_in_game"] = self.world_info[key]["constant"]
                 self.socketio.emit("world_info_entry", self.world_info[key], broadcast=True, room="UI_2")
         
     def set_world_info_used(self, uid):

static/koboldai.js

@@ -233,7 +233,7 @@ function do_story_text_updates(data) {
 		item.setAttribute("world_info_uids", "");
 		item.classList.remove("pulse")
 		item.scrollIntoView();
-		assign_world_info_to_action(action_item = item);
+		assign_world_info_to_action(item, null);
 	} else {
 		var span = document.createElement("span");
 		span.id = 'Selected Text Chunk '+data.value.id;
@@ -268,7 +268,7 @@ function do_story_text_updates(data) {
 		
 		story_area.append(span);
 		span.scrollIntoView();
-		assign_world_info_to_action(action_item = span);
+		assign_world_info_to_action(span, null);
 	}
 	
 	
@@ -315,6 +315,49 @@ function do_story_text_length_updates(data) {
 	
 }
 
+function do_probabilities(data) {
+	console.log(data);
+	if (document.getElementById('probabilities_'+data.value.id)) {
+		prob_area = document.getElementById('probabilities_'+data.value.id)
+	} else {
+		probabilities = document.getElementById('probabilities');
+		prob_area = document.createElement('span');
+		prob_area.id = 'probabilities_'+data.value.id;
+		probabilities.append(prob_area);
+	}
+	//Clear
+	while (prob_area.firstChild) { 
+		prob_area.removeChild(prob_area.lastChild);
+	}
+	//create table
+	table = document.createElement("table");
+	table.border=1;
+	if ("Probabilities" in data.value.action) {
+		for (token of data.value.action.Probabilities) {
+			actual_text = document.createElement("td");
+			actual_text.setAttribute("rowspan", token.length);
+			actual_text.textContent = "Word Goes Here";
+			for (const [index, word] of token.entries()) {
+				tr = document.createElement("tr");
+				if (index == 0) {
+					tr.append(actual_text);
+				}
+				decoded = document.createElement("td");
+				decoded.textContent = word.decoded;
+				tr.append(decoded);
+				score = document.createElement("td");
+				score.textContent = (word.score*100).toFixed(2)+"%";
+				tr.append(score);
+				table.append(tr);
+			}
+		}
+	}
+	prob_area.append(table);
+	
+	//prob_area.textContent = data.value.action["Probabilities"];
+	
+}
+
 function do_presets(data) {
 	for (select of document.getElementsByClassName('presets')) {
 		//clear out the preset list
@@ -388,38 +431,18 @@ function do_ai_busy(data) {
 }
 
 function var_changed(data) {
-	if ((data.classname =="actions") && (data.name == 'Probabilities')) {
-		console.log(data);
-	}
 	//console.log({"name": data.name, "data": data});
 	//Special Case for Actions
 	if ((data.classname == "story") && (data.name == "actions")) {
-		console.log(data);
 		do_story_text_updates(data);
 		create_options(data);
 		do_story_text_length_updates(data);
+		do_probabilities(data);
 		if (data.value.action['In AI Input']) {
 			document.getElementById('Selected Text Chunk '+data.value.id).classList.add("within_max_length");
 		} else {
 			document.getElementById('Selected Text Chunk '+data.value.id).classList.remove("within_max_length");
 		}
-	//Special Case for Story Text
-	} else if ((data.classname == "actions") && (data.name == "Selected Text")) {
-		//do_story_text_updates(data);
-	//Special Case for Story Options
-	} else if ((data.classname == "actions") && (data.name == "Options")) {
-		//create_options(data);
-	//Special Case for Story Text Length
-	} else if ((data.classname == "actions") && (data.name == "Selected Text Length")) {
-		//do_story_text_length_updates(data);
-	//Special Case for Story Text Length
-	} else if ((data.classname == "actions") && (data.name == "In AI Input")) {
-		//console.log(data.value);
-		//if (data.value['In AI Input']) {
-		//	document.getElementById('Selected Text Chunk '+data.value.id).classList.add("within_max_length");
-		//} else {
-		//	document.getElementById('Selected Text Chunk '+data.value.id).classList.remove("within_max_length");
-		//}
 	//Special Case for Presets
 	} else if ((data.classname == 'model') && (data.name == 'presets')) {
 		do_presets(data);
@@ -1239,7 +1262,7 @@ function world_info_entry(data) {
 	}
 	
 	$('#world_info_constant_'+data.uid).bootstrapToggle();
-	assign_world_info_to_action(uid=data.uid);
+	assign_world_info_to_action(null, data.uid);
 	
 	update_token_lengths();
 	
@@ -1902,11 +1925,11 @@ function dragend(e) {
 	e.preventDefault();
 }
 
-function assign_world_info_to_action(uid=null, action_item=null) {
+function assign_world_info_to_action(action_item, uid) {
 	if (Object.keys(world_info_data).length > 0) {
 		if (uid != null) {
 			var worldinfo_to_check = {};
-			worldinfo_to_check[uid] = world_info_data[uid]
+			worldinfo_to_check[uid] = world_info_data[uid];
 		} else {
 			var worldinfo_to_check = world_info_data;
 		}
@@ -1915,6 +1938,7 @@ function assign_world_info_to_action(uid=null, action_item=null) {
 		} else {
 			var actions = document.getElementById("Selected Text").children;
 		}
+		
 		for (action of actions) {
 			//First check to see if we have a key in the text
 			var words = Array.prototype.slice.call( action.children );

templates/settings flyout.html

@@ -77,6 +77,9 @@
 			{% endwith %}
 		{% endwith %}
 		</div>
+		<div id="probabilities">
+			
+		</div>
 	</div>
 	<div id="setting_menu_settings" class="hidden settings_category_area">
 		<div class="force_center">