fw align
DPO, ORPO and KTO loss demo on one synthetic preference pair; no model is aligned
fw align <dpo|orpo|kto> [--beta 0.1] [--steps 60]Aliases
Section titled “Aliases”fw dpo runs fw align. fw orpo runs fw align.
Subcommands
Section titled “Subcommands”fw align dpofw align orpofw align ktoThe names come from the usage line.
| Flag | Argument |
|---|---|
--beta |
0.1 |
--steps |
60 |
Global options
Section titled “Global options”Every command also accepts -C, --directory, --offline, --json, -q and -v. See the overview.
Same group: ML, Inference & Heretic
Section titled “Same group: ML, Inference & Heretic”fw infer, fw train, fw hf, fw abliterate, fw vram, fw heretic
Source
Section titled “Source”Read from fw align --help of fw 2026.11.0.