Property EnableQuantization
- Namespace
- LMKit.Finetuning
- Assembly
- LM-Kit.NET.dll
EnableQuantization
Gets or sets a flag indicating whether the merged model is re-quantized back to the base model's precision after the merge.
public bool EnableQuantization { get; set; }
Property Value
Remarks
The merge itself always writes the adapted tensors at F16 precision,
so a merged quantized base grows accordingly. When true, and the base
precision differs from the merge output, a quantization pass converts the
merged model back to the base model's precision (for example Q4_K_M), so the
final artifact stays the size of the base at a small fidelity cost.
The default is false: the merged model keeps its F16 tensors.