SillyTavern 1.12.9

Highlight

• Fixed a performance issue with saving replies in large chats.

Improvements

• Added cardsCacheCapacity in config.yaml to control memory usage for parsed character cards (default: 100 MB).
• Added Gemini Experimental 2024-12-06 model.
• Added import of chats from Kobold Lite JSON format.
• Added {{lastGenerationType}} and {{groupNotMuted}} macros.
• Added /getpromptentry command to get a boolean state of the prompt manager entry.
• Added a display of the settings preset used for generation to the Prompt Itemization dialog.
• Added trim_stop parameter to KoboldCpp requests to remove trailing stop tokens from the output.

Bug Fixes

• Fixed missing sampler parameter controls for NanoGPT API type.
• Fixed message-adding (/sys, /send, etc.) commands to properly parse at=0 and at=-0.
• Fixed group candidates filter when advanced character search is enabled.
• Fixed /tools-unregister command not accepting a tool name in the unnamed argument.

Community Updates

• Bump cross-spawn from 7.0.3 to 7.0.6 in /tests by @dependabot in #3137
• aphrodite: add range parameter for DRY by @AlpinDale in #3142
• Add docker tutorial to Russian readme by @Ivruix in #3141
• Fix -0 comparison by @Cohee1207 in #3147
• Small css fix. by @IceFog72 in #3148
• Better console output by @Meus-Artis in #3144
• add /getpromptentry command by @Succubyss in #3150
• Support Gemini-exp-1206 by @M0ch0 in #3153
• Optimize past swipes by @Cohee1207 in #3154
• Add types for SillyTavern.getContext by @Cohee1207 in #3152

New Contributors

• @IceFog72 made their first contribution in #3148
• @Meus-Artis made their first contribution in #3144

Full Changelog: 1.12.8...1.12.9