Expose mixed_precision dtype arguments

add training.mixed_precision_param and .mixed_precision_reduce options

refactor a util to map strings to torch dtypes

ghstack-source-id: 387e1ca13ad23e859d21d7760f858ee6e269a796
Pull Request resolved: #348

torchtitan/checkpoint.py

@@ -22,17 +22,10 @@
     set_optimizer_state_dict,
 )
 from torch.distributed.checkpoint.stateful import Stateful
-from torchtitan.config_manager import JobConfig
+from torchtitan.config_manager import JobConfig, TORCH_DTYPE_MAP
 from torchtitan.logging_utils import init_logger, logger
 
 
-DTYPE_MAP = {
-    "float16": torch.float16,
-    "float32": torch.float32,
-    "bfloat16": torch.bfloat16,
-}
-
-
 class IntervalType(enum.Enum):
     SECONDS = enum.auto()
     STEPS = enum.auto()
@@ -141,7 +134,7 @@ def __init__(
         self.pg = dist.new_group(backend="gloo")
 
         self.model_weights_only = ckpt_config.model_weights_only
-        self.export_dtype = DTYPE_MAP[ckpt_config.export_dtype]
+        self.export_dtype = TORCH_DTYPE_MAP[ckpt_config.export_dtype]
 
         self.mp = None
         async_mode = ckpt_config.async_mode.lower()

torchtitan/config_manager.py

@@ -9,13 +9,21 @@
 from collections import defaultdict
 from typing import Tuple, Union
 
+import torch
+
 try:
     import tomllib
 except ModuleNotFoundError:
     import tomli as tomllib
 
 from torchtitan.logging_utils import logger
 
+TORCH_DTYPE_MAP = {
+    "float16": torch.float16,
+    "float32": torch.float32,
+    "bfloat16": torch.bfloat16,
+}
+
 
 class JobConfig:
     """
@@ -207,6 +215,26 @@ def __init__(self):
             default=1,
             help="Pipeline Parallelism degree. 1 means disabled.",
         )
+        self.parser.add_argument(
+            "--training.mixed_precision_param",
+            type=str,
+            default="bfloat16",
+            choices=["bfloat16", "float32"],
+            help="""
+                torch dtype to use for parameters when applying mixed precision via FSDP.
+                This feature only takes effect when data_parallel_degree > 1
+            """,
+        )
+        self.parser.add_argument(
+            "--training.mixed_precision_reduce",
+            type=str,
+            default="float32",
+            choices=["float32"],
+            help="""
+                torch dtype to use for reductions when applying mixed precision via FSDP.
+                This feature only takes effect when data_parallel_degree > 1
+            """,
+        )
         self.parser.add_argument(
             "--training.compile",
             action="store_true",
@@ -275,6 +303,7 @@ def __init__(self):
             "--checkpoint.export_dtype",
             type=str,
             default="float32",
+            choices=["float16", "bfloat16", "float32"],
             help="""
                 Converts to the specified precision when training completes and model_weights_only=true.
                 Currently supports float32, float16, and bfloat16.

torchtitan/parallelisms/parallelize_llama.py

@@ -28,7 +28,7 @@
 
 from torch.utils.checkpoint import _pt2_selective_checkpoint_context_fn_gen, checkpoint
 
-from torchtitan.config_manager import JobConfig
+from torchtitan.config_manager import JobConfig, TORCH_DTYPE_MAP
 from torchtitan.logging_utils import logger
 
 
@@ -209,9 +209,9 @@ def parallelize_llama(model, world_mesh, parallel_dims, job_config: JobConfig):
     if parallel_dims.dp_enabled:
         dp_mesh = world_mesh["dp"] if world_mesh.ndim > 1 else world_mesh
         assert dp_mesh.mesh_dim_names == ("dp",), dp_mesh.mesh_dim_names
-        # TODO: Expose `reduce_dtype` as a config option.
         mp_policy = MixedPrecisionPolicy(
-            param_dtype=torch.bfloat16, reduce_dtype=torch.float32
+            param_dtype=TORCH_DTYPE_MAP[job_config.training.mixed_precision_param],
+            reduce_dtype=TORCH_DTYPE_MAP[job_config.training.mixed_precision_reduce],
         )
         ac_mode = job_config.activation_checkpoint.mode
         fsdp_config = {"mesh": dp_mesh, "mp_policy": mp_policy}