Presets

A preset defines the model source, context size, chat options, and sampler parameters used by the API. You pass a preset when creating the API instance.

Base type

All presets derive from PresetCoreBase (namespace Aspose.LLM.Abstractions.Parameters.Presets). This type exposes:

  • BaseModelSourceParameters — Where to load the main model (e.g. Hugging Face repo and file name).
  • MmprojSourceParameters — Optional vision projector (for vision presets).
  • BinaryManagerParameters — Release tag and binary management.
  • EngineParameters — Engine and cache settings.
  • ChatParameters — Chat template and role names.
  • ContextParameters — Context size, batch sizes, threading, GPU layers.
  • SamplerParameters — Sampling (temperature, top-p, etc.).
  • BaseModelInferenceParameters — Inference options (GPU layers, memory mapping, etc.).

Creating the API with a preset

using Aspose.LLM;
using Aspose.LLM.Abstractions.Parameters.Presets;

var preset = new Qwen25Preset();
using var api = AsposeLLMApi.Create(preset);

You can use a built-in preset (e.g. Qwen25Preset, Gemma3Preset, Llama32Preset) or a custom class that extends PresetCoreBase and sets the properties you need.

Default preset

api.GetDefaultPreset() returns a preset filled with the engine’s default parameters. You can use it to create another API instance or inspect defaults. For raw parameter tuples, use await api.GetDefaultParametersAsync().

Supported presets

See Supported presets in the product overview for a list of built-in text and vision presets.

Chat sessions