102 lines
2.9 KiB
Markdown
102 lines
2.9 KiB
Markdown
---
|
|
type: NodeDoc
|
|
title: LLM Prompt (Text)
|
|
description: Auto-generated node documentation.
|
|
tags: [nodes, docs]
|
|
---
|
|
|
|
# LLM Prompt (Text)
|
|
|
|
* **Node ID:** `Sage_LLMPromptText`
|
|
* **Category:** `Sage Utils/LLM`
|
|
|
|
Unified provider-switching text generation node for REST/OpenAI and Native CLIP.
|
|
|
|
## Inputs
|
|
|
|
### `prompt` — `STRING`
|
|
- **Name:** `prompt`
|
|
- **Description:** The text prompt sent to the selected LLM provider.
|
|
|
|
### `provider` — `COMFY_DYNAMICCOMBO_V3`
|
|
- **Name:** `provider`
|
|
- **Description:** Pick the backend provider and its model/runtime settings.
|
|
|
|
### `lmstudio_rest_model` — `COMBO`
|
|
- **Name:** `model`
|
|
- **Description:** Input value for lmstudio_rest_model.
|
|
|
|
### `lmstudio_rest_load_for_seconds` — `INT`
|
|
- **Name:** `load_for_seconds`
|
|
- **Description:** Compatibility field for LM Studio REST model load duration.
|
|
|
|
### `lmstudio_rest_system_prompt` — `STRING` (optional)
|
|
- **Name:** `system_prompt`
|
|
- **Description:** Optional system instruction prepended as model context.
|
|
|
|
### `ollama_rest_model` — `COMBO`
|
|
- **Name:** `model`
|
|
- **Description:** Input value for ollama_rest_model.
|
|
|
|
### `ollama_rest_keep_alive` — `STRING`
|
|
- **Name:** `keep_alive`
|
|
- **Description:** How long to keep the model loaded after generation (Ollama duration string).
|
|
|
|
### `ollama_rest_options` — `OLLAMA_OPTIONS` (optional)
|
|
- **Name:** `options`
|
|
- **Description:** Optional low-level Ollama generation parameters.
|
|
|
|
### `ollama_rest_system_prompt` — `STRING` (optional)
|
|
- **Name:** `system_prompt`
|
|
- **Description:** Optional system instruction prepended as model context.
|
|
|
|
### `openai_model` — `COMBO`
|
|
- **Name:** `model`
|
|
- **Description:** Input value for openai_model.
|
|
|
|
### `openai_system_prompt` — `STRING` (optional)
|
|
- **Name:** `system_prompt`
|
|
- **Description:** Optional system instruction prepended as model context.
|
|
|
|
### `openai_temperature` — `FLOAT`
|
|
- **Name:** `temperature`
|
|
- **Description:** Input value for openai_temperature.
|
|
|
|
### `openai_max_tokens` — `INT`
|
|
- **Name:** `max_tokens`
|
|
- **Description:** Input value for openai_max_tokens.
|
|
|
|
### `native_clip` — `CLIP`
|
|
- **Name:** `model`
|
|
- **Description:** The native CLIP model used for local LLM generation.
|
|
|
|
### `native_max_length` — `INT`
|
|
- **Name:** `max_length`
|
|
- **Description:** Maximum sequence length for native generation.
|
|
|
|
### `native_thinking` — `BOOLEAN` (optional)
|
|
- **Name:** `thinking`
|
|
- **Description:** Enable model thinking mode if supported by the loaded CLIP model.
|
|
|
|
### `native_sampling` — `COMFY_DYNAMICCOMBO_V3`
|
|
- **Name:** `sampling`
|
|
- **Description:** Choose native sampling settings for the local model.
|
|
|
|
### `seed` — `INT`
|
|
- **Name:** `seed`
|
|
- **Description:** Base seed used by all providers (provider-specific behavior may vary).
|
|
|
|
|
|
## Outputs
|
|
|
|
### `response` — `STRING`
|
|
- **Name:** `response`
|
|
- **Description:** The text response returned by the selected provider.
|
|
|
|
|
|
## Notes
|
|
|
|
|
|
|
|
Generated from the node schema.
|