Merge pull request #488 from WolframResearch/main

Release v1.3.3
WolframResearch · Dec 5, 2023 · b30a0c8 · b30a0c8
2 parents 242740a + e812774
commit b30a0c8
Show file tree

Hide file tree

Showing 10 changed files with 469 additions and 65 deletions.
diff --git a/PacletInfo.wl b/PacletInfo.wl
@@ -1,7 +1,7 @@
 PacletObject[ <|
     "Name"           -> "Wolfram/Chatbook",
     "PublisherID"    -> "Wolfram",
-    "Version"        -> "1.3.2",
+    "Version"        -> "1.3.3",
     "WolframVersion" -> "13.3+",
     "Description"    -> "Wolfram Notebooks + LLMs",
     "License"        -> "MIT",

diff --git a/Scripts/Common.wl b/Scripts/Common.wl
@@ -141,7 +141,7 @@ releaseID[ dir_ ] := FirstCase[
 (* ::Subsection::Closed:: *)
 (*releaseURL*)
 releaseURL[ file_ ] := Enclose[
-    Enclose @ Module[ { pac, repo, ver },
+    Module[ { pac, repo, ver },
         pac  = PacletObject @ Flatten @ File @ file;
         repo = ConfirmBy[ Environment[ "GITHUB_REPOSITORY" ], StringQ ];
         ver  = ConfirmBy[ pac[ "Version" ], StringQ ];

diff --git a/Source/Chatbook/ChatMessages.wl b/Source/Chatbook/ChatMessages.wl
@@ -17,6 +17,7 @@ Wolfram`Chatbook`CellToChatMessage;
 `constructMessages;
 `expandMultimodalString;
 `getTokenizer;
+`getTokenizerName;
 `resizeMultimodalImage;
 
 Begin[ "`Private`" ];
@@ -70,6 +71,10 @@ $styleRoles = <|
     "ChatSystemInput"        -> "System"
 |>;
 
+$cachedTokenizerNames = { "chat-bison", "claude", "gpt-2", "gpt-3.5", "gpt-4-vision", "gpt-4" };
+$cachedTokenizers     = <| |>;
+$fallbackTokenizer    = "gpt-2";
+
 (* ::**************************************************************************************************************:: *)
 (* ::Section::Closed:: *)
 (*CellToChatMessage*)
@@ -1079,58 +1084,120 @@ argumentTokenToString // endDefinition;
 (* ::**************************************************************************************************************:: *)
 (* ::Section::Closed:: *)
 (*Tokenization*)
-$tokenizer := gpt2Tokenizer;
+$tokenizer := $gpt2Tokenizer;
+
+(* ::**************************************************************************************************************:: *)
+(* ::Subsection::Closed:: *)
+(*getTokenizerName*)
+getTokenizerName // beginDefinition;
+
+getTokenizerName[ KeyValuePattern[ "TokenizerName"|"Tokenizer" -> name_String ] ] :=
+    tokenizerName @ name;
+
+getTokenizerName[ KeyValuePattern[ "Tokenizer" -> Except[ $$unspecified ] ] ] :=
+    "Custom";
+
+getTokenizerName[ KeyValuePattern[ "Model" -> model_ ] ] :=
+    With[ { name = tokenizerName @ toModelName @ model },
+        If[ MemberQ[ $cachedTokenizerNames, name ],
+            name,
+            $fallbackTokenizer
+        ]
+    ];
+
+getTokenizerName // endDefinition;
 
 (* ::**************************************************************************************************************:: *)
 (* ::Subsection::Closed:: *)
 (*getTokenizer*)
 getTokenizer // beginDefinition;
 getTokenizer[ KeyValuePattern[ "Tokenizer" -> tokenizer: Except[ $$unspecified ] ] ] := tokenizer;
-getTokenizer[ KeyValuePattern[ "Model" -> model_ ] ] := getTokenizer @ model;
-getTokenizer[ model_ ] := cachedTokenizer @ toModelName @ model;
+getTokenizer[ KeyValuePattern[ "TokenizerName" -> name_String ] ] := cachedTokenizer @ name;
+getTokenizer[ KeyValuePattern[ "Model" -> model_ ] ] := cachedTokenizer @ toModelName @ model;
 getTokenizer // endDefinition;
 
 (* ::**************************************************************************************************************:: *)
 (* ::Subsubsection::Closed:: *)
 (*cachedTokenizer*)
 cachedTokenizer // beginDefinition;
-cachedTokenizer[ All ] := AssociationMap[ cachedTokenizer, $cachedTokenizerNames ];
-cachedTokenizer[ name_String ] := cachedTokenizer0 @ tokenizerName @ toModelName @ name;
+
+cachedTokenizer[ All ] :=
+    AssociationMap[ cachedTokenizer, $cachedTokenizerNames ];
+
+cachedTokenizer[ id_String ] :=
+    With[ { tokenizer = $cachedTokenizers[ tokenizerName @ toModelName @ id ] },
+        tokenizer /; ! MatchQ[ tokenizer, $$unspecified ]
+    ];
+
+cachedTokenizer[ id_String ] := Enclose[
+    Module[ { name, tokenizer },
+        name      = ConfirmBy[ tokenizerName @ toModelName @ id, StringQ, "Name" ];
+        tokenizer = findTokenizer @ name;
+        If[ MissingQ @ tokenizer,
+            (* Fallback to the GPT-2 tokenizer: *)
+            tokenizer = ConfirmMatch[ $gpt2Tokenizer, Except[ $$unspecified ], "GPT2Tokenizer" ];
+            If[ TrueQ @ Wolfram`ChatbookInternal`$BuildingMX,
+                tokenizer, (* Avoid caching fallback values into MX definitions *)
+                cacheTokenizer[ name, tokenizer ]
+            ],
+            cacheTokenizer[ name, ConfirmMatch[ tokenizer, Except[ $$unspecified ], "Tokenizer" ] ]
+        ]
+    ],
+    throwInternalFailure
+];
+
 cachedTokenizer // endDefinition;
 
+(* ::**************************************************************************************************************:: *)
+(* ::Subsubsubsection::Closed:: *)
+(*cacheTokenizer*)
+cacheTokenizer // beginDefinition;
 
-cachedTokenizer0 // beginDefinition;
+cacheTokenizer[ name_String, tokenizer: Except[ $$unspecified ] ] := (
+    $cachedTokenizerNames = Union[ $cachedTokenizerNames, { name } ];
+    $cachedTokenizers[ name ] = tokenizer
+);
 
-cachedTokenizer0[ "chat-bison" ] = ToCharacterCode[ #, "UTF8" ] &;
+cacheTokenizer // endDefinition;
 
-cachedTokenizer0[ "gpt-4-vision" ] :=
-    If[ graphicsQ[ # ],
-        gpt4ImageTokenizer[ # ],
-        cachedTokenizer[ "gpt-4" ][ # ]
-    ] &;
+(* ::**************************************************************************************************************:: *)
+(* ::Subsubsubsection::Closed:: *)
+(*findTokenizer*)
+findTokenizer // beginDefinition;
 
-cachedTokenizer0[ model_String ] := Enclose[
+findTokenizer[ model_String ] := Enclose[
     Quiet @ Module[ { name, tokenizer },
         initTools[ ];
         Quiet @ Needs[ "Wolfram`LLMFunctions`Utilities`Tokenization`" -> None ];
         name      = ConfirmBy[ tokens`FindTokenizer @ model, StringQ, "Name" ];
         tokenizer = ConfirmMatch[ tokens`LLMTokenizer[ Method -> name ], Except[ _tokens`LLMTokenizer ], "Tokenizer" ];
         ConfirmMatch[ tokenizer[ "test" ], _List, "TokenizerTest" ];
-        cachedTokenizer0[ model ] = tokenizer
+        tokenizer
     ],
-    gpt2Tokenizer &
+    Missing[ "NotFound" ] &
 ];
 
-cachedTokenizer0 // endDefinition;
+findTokenizer // endDefinition;
+
+(* ::**************************************************************************************************************:: *)
+(* ::Subsubsubsection::Closed:: *)
+(*Pre-cached small tokenizer functions*)
+$cachedTokenizers[ "chat-bison"   ] = ToCharacterCode[ #, "UTF8" ] &;
+$cachedTokenizers[ "gpt-4-vision" ] = If[ graphicsQ[ # ], gpt4ImageTokenizer[ # ], cachedTokenizer[ "gpt-4" ][ # ] ] &;
 
 (* ::**************************************************************************************************************:: *)
 (* ::Subsubsection::Closed:: *)
 (*tokenizerName*)
 tokenizerName // beginDefinition;
-tokenizerName[ name_String ] := SelectFirst[ $cachedTokenizerNames, StringContainsQ[ name, # ] &, name ];
-tokenizerName // endDefinition;
 
-$cachedTokenizerNames = { "gpt-4-vision", "gpt-4", "gpt-3.5", "gpt-2", "claude-2", "claude-instant-1", "chat-bison" };
+tokenizerName[ name_String ] :=
+    SelectFirst[
+        ReverseSortBy[ $cachedTokenizerNames, StringLength ],
+        StringContainsQ[ name, #, IgnoreCase -> True ] &,
+        name
+    ];
+
+tokenizerName // endDefinition;
 
 (* ::**************************************************************************************************************:: *)
 (* ::Subsection::Closed:: *)
@@ -1182,13 +1249,19 @@ gpt4ImageTokenCount0 // endDefinition;
 (* ::**************************************************************************************************************:: *)
 (* ::Subsection::Closed:: *)
 (*Fallback Tokenizer*)
-gpt2Tokenizer := gpt2Tokenizer = ResourceFunction[ "GPTTokenizer" ][ ];
+$gpt2Tokenizer := $gpt2Tokenizer = gpt2Tokenizer[ ];
+
+(* https://resources.wolframcloud.com/FunctionRepository/resources/GPTTokenizer *)
+importResourceFunction[ gpt2Tokenizer, "GPTTokenizer" ];
 
 (* ::**************************************************************************************************************:: *)
 (* ::Section::Closed:: *)
 (*Package Footer*)
 If[ Wolfram`ChatbookInternal`$BuildingMX,
     cachedTokenizer[ All ];
+    $gpt2Tokenizer;
+    (* This is only needed to generate $gpt2Tokenizer once, so it can be removed to reduce MX file size: *)
+    Remove[ "Wolfram`Chatbook`ResourceFunctions`GPTTokenizer`GPTTokenizer" ];
 ];
 
 (* :!CodeAnalysis::EndBlock:: *)

diff --git a/Source/Chatbook/Formatting.wl b/Source/Chatbook/Formatting.wl
@@ -1137,11 +1137,12 @@ inlineInteractiveCodeCell // beginDefinition;
 
 inlineInteractiveCodeCell[ display_, string_ ] /; $dynamicText := display;
 
+(* TODO: make this switch dynamically depending on $cloudNotebooks (likely as a TemplateBox)*)
 inlineInteractiveCodeCell[ display_, string_ ] :=
     inlineInteractiveCodeCell[ display, string, contentLanguage @ string ];
 
 inlineInteractiveCodeCell[ display_, string_, lang_ ] /; $cloudNotebooks :=
-    Mouseover[ display, Column @ { display, floatingButtonGrid[ string, lang ] } ];
+    cloudInlineInteractiveCodeCell[ display, string, lang ];
 
 inlineInteractiveCodeCell[ display_, string_, lang_ ] :=
     DynamicModule[ { $CellContext`attached, $CellContext`cell },
@@ -1167,6 +1168,54 @@ inlineInteractiveCodeCell[ display_, string_, lang_ ] :=
 
 inlineInteractiveCodeCell // endDefinition;
 
+(* ::**************************************************************************************************************:: *)
+(* ::Subsubsection::Closed:: *)
+(*cloudInlineInteractiveCodeCell*)
+cloudInlineInteractiveCodeCell // beginDefinition;
+
+cloudInlineInteractiveCodeCell[ display_, string_, lang_ ] :=
+    Module[ { padded, buttons },
+
+        padded = Pane[ display, ImageSize -> { { 100, Automatic }, { 30, Automatic } } ];
+
+        buttons = Framed[
+            floatingButtonGrid[ string, lang ],
+            Background     -> White,
+            FrameMargins   -> { { 1, 0 }, { 0, 1 } },
+            FrameStyle     -> White,
+            ImageMargins   -> 1,
+            RoundingRadius -> 3
+        ];
+
+        Mouseover[
+            buttonOverlay[ padded, Invisible @ buttons ],
+            buttonOverlay[ padded, buttons ],
+            ContentPadding -> False,
+            FrameMargins   -> 0,
+            ImageMargins   -> 0,
+            ImageSize      -> All
+        ]
+    ];
+
+cloudInlineInteractiveCodeCell // endDefinition;
+
+(* ::**************************************************************************************************************:: *)
+(* ::Subsubsection::Closed:: *)
+(*buttonOverlay*)
+buttonOverlay // beginDefinition;
+
+buttonOverlay[ a_, b_ ] := Overlay[
+    { a, b },
+    All,
+    2,
+    Alignment      -> { Left, Bottom },
+    ContentPadding -> False,
+    FrameMargins   -> 0,
+    ImageMargins   -> 0
+];
+
+buttonOverlay // endDefinition;
+
 (* ::**************************************************************************************************************:: *)
 (* ::Subsection::Closed:: *)
 (*makeInlineCodeCell*)

diff --git a/Source/Chatbook/Main.wl b/Source/Chatbook/Main.wl
@@ -6,6 +6,7 @@ BeginPackage[ "Wolfram`Chatbook`" ];
 (* ::**************************************************************************************************************:: *)
 (* ::Subsection::Closed:: *)
 (*Declare Symbols*)
+`$AvailableTools;
 `$ChatHandlerData;
 `$ChatPost;
 `$ChatPre;

diff --git a/Source/Chatbook/Prompting.wl b/Source/Chatbook/Prompting.wl
@@ -35,6 +35,10 @@ $basePromptOrder = {
     "Checkboxes",
     "CheckboxesIndeterminate",
     "ConversionFormatting",
+    "SpecialURI",
+    "SpecialURIAudio",
+    "SpecialURIVideo",
+    "SpecialURIDynamic",
     "VisibleUserInput",
     "TrivialCode",
     "Packages",
@@ -51,6 +55,7 @@ $basePromptClasses = <|
     "Math"              -> { "MathExpressions" },
     "Formatting"        -> { "CodeBlocks", "MathExpressions", "EscapedCharacters" },
     "MessageConversion" -> { "ConversionLargeOutputs", "ConversionGraphics", "ConversionFormatting" },
+    "SpecialURIs"       -> { "SpecialURIAudio", "SpecialURIVideo", "SpecialURIDynamic" },
     "All"               -> $basePromptOrder
 |>;
 
@@ -71,6 +76,10 @@ $basePromptDependencies = Append[ "GeneralInstructionsHeader" ] /@ <|
     "ConversionGraphics"           -> { "MessageConversionHeader" },
     "MarkdownImageBox"             -> { "MessageConversionHeader" },
     "ConversionFormatting"         -> { "MessageConversionHeader" },
+    "SpecialURI"                   -> { },
+    "SpecialURIAudio"              -> { "SpecialURI" },
+    "SpecialURIVideo"              -> { "SpecialURI" },
+    "SpecialURIDynamic"            -> { "SpecialURI" },
     "VisibleUserInput"             -> { },
     "TrivialCode"                  -> { },
     "WolframSymbolCapitalization"  -> { },
@@ -159,6 +168,19 @@ $basePromptComponents[ "ConversionFormatting" ] = "\
 ``Cell[TextData[{StyleBox[\"Styled\", FontSlant -> \"Italic\"], \" message\"}], \"ChatInput\"]`` \
 becomes ``Styled message``.";
 
+$basePromptComponents[ "SpecialURI" ] = "\
+* You will occasionally see markdown links with special URI schemes, e.g. ![label](scheme://content-id) that represent \
+interactive interface elements. You can use these in your responses to display the same elements to the user.";
+
+$basePromptComponents[ "SpecialURIAudio" ] = "\
+    * ![label](audio://content-id) represents an interactive audio player.";
+
+$basePromptComponents[ "SpecialURIVideo" ] = "\
+    * ![label](video://content-id) represents an interactive video player.";
+
+$basePromptComponents[ "SpecialURIDynamic" ] = "\
+    * ![label](dynamic://content-id) represents an embedded dynamic UI.";
+
 $basePromptComponents[ "VisibleUserInput" ] = "\
 * The user can still see their input, so there's no need to repeat it in your response";