Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 5 additions & 3 deletions src/maxtext/common/checkpointing.py
Original file line number Diff line number Diff line change
Expand Up @@ -198,19 +198,21 @@ def _load_full_state_from_path(
if source_checkpoint_layout == "orbax":
context = ocp_v1.Context(checkpoint_layout=ocp_v1.options.CheckpointLayout.ORBAX)
with context:
return ocp_v1.load_pytree(path, abstract_unboxed_pre_state)
return ocp_v1.load(path, abstract_state=abstract_unboxed_pre_state)
elif source_checkpoint_layout == "safetensors":
context = ocp_v1.Context(checkpoint_layout=ocp_v1.options.CheckpointLayout.SAFETENSORS)
with context:
metadata = ocp_v1.pytree_metadata(path)
metadata = ocp_v1.metadata(path)
simple_abstract_state = metadata.metadata
shardings = sharding_utils.construct_maximal_shardings(simple_abstract_state)

def combine_sharding(sds, shardings):
return jax.ShapeDtypeStruct(shape=sds.shape, dtype=sds.dtype, sharding=shardings)

sharded_abstract_state = jax.tree.map(combine_sharding, simple_abstract_state, shardings)
pre_transformed_state = ocp_v1.load_pytree(path, sharded_abstract_state)
pre_transformed_state = ocp_v1.load(
path, abstract_state=sharded_abstract_state
)
state = checkpoint_conversion_fn(pre_transformed_state)
return state
else:
Expand Down
Loading