sf sweep

Grid expansion for batch submission

sf sweep <EXP> -g KEY=V1,V2 [options]

Each variant is one standard submission (quotas / queuing apply as usual). See Hyperparameter Sweeps for the workflow walkthrough.

Option	Description
-g, --set-grid KEY=V1,V2 (required, repeatable)	Grid axis; multiple axes take the Cartesian product
-s, --set KEY=VALUE	Fixed overrides shared by all variants, repeatable
--profile	Resource declaration, same as sf submit
-m, --method	Method id; if omitted, read from the lock file
--dry-run	Only print the variant list, do not submit
--allow-dirty / --no-validate	Same as sf submit

Examples

# 3 × 2 = 6 variants; review first, then submit
sf sweep my-grpo --profile h200:4 \
  -g policy.optimizer.kwargs.lr=1e-6,2e-6,5e-6 \
  -g grpo.kl_coef=0.01,0.05 \
  --dry-run
sf sweep my-grpo --profile h200:4 \
  -g policy.optimizer.kwargs.lr=1e-6,2e-6,5e-6 \
  -g grpo.kl_coef=0.01,0.05

# Stop the whole group with one command
sf job stop-sweep <SWEEP_ID>

sf sweep

Submit a hyperparameter sweep

sf sweep <EXP> [options]

Argument	Description
EXP	Experiment name or path (required)

Option	Description
--set-grid, -g	A grid axis, repeatable; several axes form a cartesian product. e.g. -g policy.lr=1e-5,2e-5 -g grpo.kl=0.01,0.05 (required · repeatable)
--set, -s	A fixed override shared by every variant, repeatable (repeatable)
--profile	Target hardware and resources in one argument: h200 (the registry default shape), h200:4 (4 GPUs), h200:16 (2 full nodes). Repeat it to split pools per role (the heterogeneous extension point): —profile train=h200:8 —profile rollout=h100:2. Comma-separate roles that share a pool: —profile actor,rollout=h200:8 —profile teacher=h200:8 (repeatable)
--method, -m	Method identifier; read from the experiment’s recipe.lock.json when omitted
--allow-dirty	Allow uncommitted changes in the working tree
--no-validate	Skip the pre-submission validation
--dry-run	Only print the variant list, submit nothing
--early-stop	Enable ASHA early stopping: cut the clearly hopeless variants at a few step rungs
--stop-metric	Which metric early stopping watches; defaults to the method’s declared primary metric (see sf methods <method>)
--rungs	Early-stopping rungs (steps), comma separated. Defaults to 100,300,900
--keep <int range>	Keep 1/keep of the variants at each rung. 3 is the usual value: cutting harder hurts slow starters (default 3)