Tune 30b

apps/grpo/main.py

@@ -32,7 +32,7 @@
 from forge.observability.metrics import record_metric, Reduce
 from forge.observability.perf_tracker import Tracer
 
-from forge.types import LauncherConfig, ProvisionerConfig
+from forge.types import LauncherConfig, ProvisionerConfig, Launcher
 from forge.util.ops import compute_logprobs
 from monarch.actor import endpoint
 from omegaconf import DictConfig
@@ -314,12 +314,12 @@ async def main(cfg: DictConfig):
     max_res_tokens = cfg.max_res_tokens
 
     # ---- Global setups ---- #
-    if cfg.get("provisioner", None) is not None:
-        await init_provisioner(
-            ProvisionerConfig(
-                launcher_config=LauncherConfig(**cfg.provisioner.launcher)
-            )
+    # if cfg.get("provisioner", None) is not None:
+    await init_provisioner(
+        ProvisionerConfig(
+            launcher_config=LauncherConfig(launcher=Launcher("slurm")),
         )
+    )
     metric_logging_cfg = cfg.get("metric_logging", {"console": {"log_per_rank": False}})
     mlogger = await get_or_create_metric_logger()
     await mlogger.init_backends.call_one(metric_logging_cfg)

apps/grpo/qwen3_32b.yaml

@@ -3,15 +3,16 @@
 # NOTE - This has not been tested for correctness yet! All testing so far has been only for infrastructure stability
 
 # Global configuration
-group_size: 2
+group_size: 8
 batch_size: 8
 max_req_tokens: 512
 max_res_tokens: 512
 model: "Qwen/Qwen3-32B"
 off_by_n: 1 # Off by one by default
 
 provisioner:
-  launcher: slurm
+  launcher_config:
+    launcher: slurm
 
 # Main loop configuration
 rollout_threads: 1   # Recommended to set equal to policy.num_replicas
@@ -37,7 +38,7 @@ dataset:
 policy:
   engine_config:
     model: ${model}
-    tensor_parallel_size: 4
+    tensor_parallel_size: 8
     pipeline_parallel_size: 1
     enforce_eager: false
   sampling_config:
@@ -69,8 +70,8 @@ trainer:
     enable: false
   parallelism:
     data_parallel_replicate_degree: 1
-    data_parallel_shard_degree: -1
-    tensor_parallel_degree: 1
+    data_parallel_shard_degree: 1
+    tensor_parallel_degree: 4
     pipeline_parallel_degree: 1
     context_parallel_degree: 1
     expert_parallel_degree: 1
@@ -136,8 +137,8 @@ actors:
     procs: 1
     with_gpus: false
   trainer:
-    procs: 8
-    hosts: 1
+    procs: 4
+    # hosts: 1
     with_gpus: true
   replay_buffer:
     procs: 1

src/forge/controller/actor.py

@@ -166,6 +166,8 @@ async def launch(cls, *args, **kwargs) -> "ForgeActor":
             mesh_name=cls.mesh_name,
         )
 
+        print(f"Spawning proc mesh {cls.mesh_name} with gpus {cls.with_gpus}")
+
         proc_mesh = await get_proc_mesh(process_config=cfg)
 
         actor_name = kwargs.pop("name", cls.__name__)

src/forge/types.py

@@ -159,9 +159,9 @@ class LauncherConfig:
     """A launcher config for the scheduler."""
 
     launcher: Launcher
-    job_name: str
-    services: dict[str, ServiceConfig]
-    actors: dict[str, ProcessConfig]
+    job_name: str | None = None
+    services: dict[str, ServiceConfig] | None = None
+    actors: dict[str, ProcessConfig] | None = None
 
 
 @dataclass