-
Notifications
You must be signed in to change notification settings - Fork 4
/
changelog.txt
116 lines (89 loc) · 5.22 KB
/
changelog.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
Changelog
---.---.---.---
V1.3-R2
- Fixed a minor bug where if there wasn't enough chat history, the program would crash, as it would attempt to load chat that wasn't there.
---.---.---.---
V1.3
- Added the Tag & Task menu
- Tags can be used to classify info for future use.
- Applies tags automatically to chats that you put in.
- Tasks can be used to swap between character cards, allowing you to swap out parts of the memory.
- Tasks are hyphenated between the "WaifuName" and "Task"
- For example, if your waifu is named "Ember", and you have a task called "Party", you would want a character card in Oobabooga to be defined "Ember-Party"
- Your bot can now use keyboard input to control the keyboard.
- Be sure to toggle "MODULE_GAMING" to "ON".
- Changing the task will change what JSON file it uses (i.e. the task "Emerald" will use the button mappings in "/Configurables/GamingInputs/Emerald.json").
- By default, this is set to "None" with no mappings.
- You can add mappings by copy/paste the file, and renaming to something else.
- Try to use lowercase letters for the keyboard input, capital letters did not work.
- Warning: They can also trigger their own hotkeys, if not turned off!
- Automatic gaming can now be toggled on in the Task menu. This is done by taking a picture, then asking for an input.
- Note: Very bad at the moment, don't expect much of anything. May require more prompting and tuning.
- Vision can now use the main monitor's screenshot as the image input.
- Turn on "Use Screenshot" in the Visual menu.
- Timestamps will now be included in the encoding, telling them the current date and time.
- You may want to ask them in the character card to not mention the current time, as they may spam it.
- Timestamps are also included and stored in message metadata.
- Can be toggled in the .env
- Messages can now be undone / redone while they are speaking, cutting them off.
- Messages are now chunked out and read, instead of all at once.
- Asterisks can now be banned from generating, for conversational mode and stopping roleplay.
- Discord token is now stored in "Configurables/Tokens/Discord.json", for security reasons.
- Fixed an issue where the lorebook was not giving lore for messages with a "?" at the end.
- Added a "is_live_pipe" state to the main script, which will tell us if we are currently running/processing something.
---.---.---.---
V1.2
- Lorebook messages are now directly infused into the encoding as it is sent.
- This now sends all relevant lore triggered within the past 3 message sets, instead of just 1 with a required cooldown.
- Lore triggering requirements were improved, to add plurals and fix edgecases.
- You can still view what lore is triggered via the UI Logs.
- Random Memories will now trigger before the alarm.
- This allows your bot to randomly scan your chat history, and remember past times.
- You can also trigger random memories manually via the UI.
- Your VTuber can now look around, either Following Faces or Randomly.
- This requires setting up 6 emotes for your VTuber. In order, they should have your VTuber's eyes doing the following (they can be named anything);
- "Look Slight Right"
- "Look Right"
- "Look Very Right"
- "Look Slight Left"
- "Look Left"
- "Look Very Left"
- In the .env, change "EYES_FOLLOW" to "Random" or "Faces". Set the "EYES_START_ID" to whatever emote slot the "Look Slight Right" is set up as.
- Make sure all the eye looking emotes follow eachother in order. You can re-order them in VTube Studio if needed.
- Obviously, you need a camera for the VTuber to follow faces, as well as the Vision module enabled.
- Other Roleplay Suppression is now disabled if you have "Cutoff at Newlines" off.
- This will allow the bot to send messages containing character lines, such as "User:" or "Riley:".
- This is to allow lists, info, and multi-user RP scenarios, if you want.
- Fixed issues with the RAG history desyncing when undoing messages.
---.---.---.---
v1.1-R2
- Fixed a few major bugs:
- Fixed the "Error" taking over all of the Gradio WebUI
- Happened due to Gradio & FastAPI dependency conflict (reminder: always vet your stuff~!)
- Fixed issues with the software failing gently when you have no mic
- Fixed crashes relating to searching for "Minecraft" logs, it now check to see if the module is enabled first
---.---.---.---
v1.1
- Visual System
- Toggleable as a module
- Able to take new images or upload them directly for the AI to see
- Runs using Ooba, like with the text
- Can set the port to the existing, default one, or load another instance to dual wield
- Option to see images before being sent
- Can retake them
- Use C/X on the keyboard to confirm
- Automatically shrinks images to a proper size
- Fixed bits of the Minecraft module
- Configurable "MinecraftUsername" to set your AI's name (stops feedback loops)
- Configurable "MinecraftUsernameFollow" to set who your AI follows when doing "#follow"
---.---.---.---
V1.0
- Initial public release of Z-Waif. Contains:
- WebUI
- RAG
- Discord
- Semi-Minecraft Functionality
- VTuber Emotes
- Hotkeys
- Various other initial release items
---.---.---.---