Back to Sillytavern

Settings

public/scripts/extensions/vectors/settings.html

1.18.03.7 KB
Original Source

Vector Storage

Vectorization Source ChutesCloudflare Workers AICohereElectron HubExtras (deprecated)Google AI StudioGoogle Vertex AIKoboldCppllama.cppLocal (Transformers)MistralAINanoGPTNomicAIOllamaOpenAIOpenRouterSiliconFlowTogetherAIvLLMWebLLM Extension

Vectorization Model Hint: Set your Chutes API key in API Connections.

Vectorization Model Hint: Set your NanoGPT API key in API Connections.

Vectorization Model Hint: Set your Electron Hub API key in API Connections.

Use secondary URL Secondary Embedding endpoint URL

Vectorization Model

Requires the WebLLM extension to be installed. Click here to install.

Vectorization Model Keep model in memory The model must be downloaded first! Do it with the ollama pull command or click here. Hint: Set the URL in the API connection settings.

Set the KoboldCpp URL in the Text Completion API connection settings. Must use version 1.87 or higher and have an embedding model loaded.

The server MUST be started with the --embedding flag to use this feature! Hint: Set the URL in the API connection settings.

Vectorization Model text-embedding-ada-002text-embedding-3-smalltext-embedding-3-large

Vectorization Model embed-v4.0embed-english-v3.0embed-multilingual-v3.0embed-english-light-v3.0embed-multilingual-light-v3.0embed-english-v2.0embed-english-light-v2.0embed-multilingual-v2.0

Vectorization Model M2-BERT-Retrieval-32kM2-BERT-Retrieval-8kM2-BERT-Retrieval-2KUAE-Large-V1BAAI-Bge-Large-1p5BAAI-Bge-Base-1p5Sentence-BERTBert Base Uncased

Vectorization Model Hint: Set the URL in the API connection settings.

Vectorization Model gemini-embedding-001gemini-embedding-2-previewgemini-embedding-exp-03-07text-embedding-004text-embedding-005embedding-001

NomicAI API Key Click to set

Vectorization Model Hint: Set your OpenRouter API key in API Connections.

Vectorization Model Hint: Set your SiliconFlow API key in API Connections.

Vectorization Model Hint: Set your Workers AI API key and Account ID in API Connections.

Query messages

Score threshold

Chunk boundary

Include in World Info Scanning


World Info settings

Enable for World Info

Enabled for all entries

  • Checked: all entries except ❌ status can be activated.
  • Unchecked: only entries with 🔗 status can be activated.

Max Entries

File vectorization settings

Enable for files Only chunk on custom boundary Translate files into English before processing Message attachments

Size threshold (KB)

Chunk size (chars)

Chunk overlap (%)

Retrieve chunks

Data Bank files

Size threshold (KB)

Chunk size (chars)

Chunk overlap (%)

Retrieve chunks

Injection TemplateInjection Position Before Main Prompt / Story StringAfter Main Prompt / Story StringIn-chat @ DepthasSystemUserAssistant

Vectorize All

Purge Vectors


Chat vectorization settings

Enabled for chat messages


Include hidden messages Injection Template Injection Position Before Main Prompt / Story StringAfter Main Prompt / Story StringIn-chat @ Depth

Chunk size (chars)

Retain#

Insert#


Vector Summarization Summarize chat messages for vector generation_Warning: This will slow down vector generation drastically, as all messages have to be summarized first._Summarize chat messages when sending_Warning: This might cause your sent messages to take a bit to process and slow down response time._Summarize with:Main APIExtras APIWebLLM ExtensionSummary Prompt:Only used when Main API or WebLLM Extension is selected.Summarization retries per messageSummarization min length (chars)

Old messages are vectorized gradually as you chat. To process all previous messages, click the button below.

Vectorize All

Purge Vectors

View Stats

Processed 0% of messages. ETA: ... seconds.