2026-04-15 23:31:14.756886: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0415 23:31:14.874557 124618438372480 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0415 23:32:13.258915 124618438372480 max_utils.py:800] System Information: Jax Version: 0.9.2 I0415 23:32:13.259022 124618438372480 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0415 23:32:13.259067 124618438372480 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0415 23:32:13.259092 124618438372480 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0415 23:32:13.259114 124618438372480 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0415 23:32:13.259191 124618438372480 train.py:715] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0415 23:32:13.349628 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:32:13.754213 124618438372480 maxtext_utils.py:1520] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0415 23:32:13.754481 124618438372480 checkpointing.py:677] Setting up checkpoint logger... I0415 23:32:13.754527 124618438372480 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0415 23:32:13.754565 124618438372480 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0415 23:32:13.755007 124618438372480 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x715668106660>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0415 23:32:15.410853 124618438372480 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0415 23:32:15.411033 124618438372480 checkpoint_manager.py:702] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71519b78fb60>}, handler_registry=None I0415 23:32:15.411301 124618438372480 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71519b78fb60>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0415 23:32:15.411340 124618438372480 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714feaf394f0>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0415 23:32:15.411372 124618438372480 composite_checkpoint_handler.py:505] Initialized registry DefaultCheckpointHandlerRegistry({('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71519b78fb60>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71519b78fb60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714feaf394f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714feaf394f0>}). I0415 23:32:15.411681 124618438372480 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0415 23:32:15.411732 124618438372480 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x714feaf3d4e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0415 23:32:15.805497 124516615325248 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776295798863739578, 'commit_timestamp_nsecs': 1776295845815975280, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/9/_CHECKPOINT_METADATA I0415 23:32:15.880430 124516615325248 checkpoint_manager.py:1847] Missing metrics for step 9 I0415 23:32:15.958619 124618438372480 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints I0415 23:32:15.959133 124618438372480 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x714feaf387a0> I0415 23:32:15.959218 124618438372480 checkpointing.py:301] Checkpoint manager created! W0415 23:32:15.975948 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:32:16.156349 124618438372480 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0415 23:32:16.156447 124618438372480 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:32:16.248300 124618438372480 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 23:32:16.248378 124618438372480 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:32:16.262465 124618438372480 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 23:32:16.262510 124618438372480 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:32:16.282924 124618438372480 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 23:32:16.282978 124618438372480 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:32:16.297145 124618438372480 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 23:32:16.297194 124618438372480 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:32:16.311316 124618438372480 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 23:32:16.311364 124618438372480 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:32:16.332950 124618438372480 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0415 23:32:16.333006 124618438372480 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:32:16.351223 124618438372480 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0415 23:32:16.351280 124618438372480 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0415 23:32:16.537017 124618438372480 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0415 23:32:16.537126 124618438372480 checkpointing.py:581] restoring from this run's directory step 9 I0415 23:32:16.692804 124618438372480 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/9. I0415 23:32:17.222735 3377946 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0415 23:32:17.222803 3377946 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0415 23:32:23.209908 124618438372480 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.995 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0415 23:32:23.210283 124618438372480 checkpointer.py:318] Finished restoring checkpoint in 6.60 seconds from gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/9. I0415 23:32:23.210397 124618438372480 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1776295936.614243, 'checkpointer_duration_secs': 6.596097946166992, 'checkpoint_manager_start_time': 1776295936.5373619, 'checkpoint_manager_duration_secs': 6.672983407974243} [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0415 23:32:23.792815 124618438372480 maxtext_utils.py:1623] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0415 23:32:23.792923 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 23:32:23.792968 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 23:32:23.793013 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0415 23:32:23.793067 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 23:32:23.793097 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 23:32:23.793135 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0415 23:32:23.793174 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0415 23:32:23.793205 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0415 23:32:23.793232 124618438372480 maxtext_utils.py:1623] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0415 23:32:23.793267 124618438372480 maxtext_utils.py:1623] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed', 'vocab') Physical: ('fsdp', None) I0415 23:32:23.793300 124618438372480 maxtext_utils.py:1623] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed') Physical: (None, 'fsdp') I0415 23:32:24.286800 124618438372480 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 23:32:24.286879 124618438372480 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0415 23:32:24.300436 124618438372480 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 23:32:24.300487 124618438372480 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0415 23:32:24.995693 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:32:25.060853 124618438372480 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0415 23:32:25.061386 124618438372480 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0415 23:32:25.061551 124618438372480 metric_logger.py:289] number parameters: 1.104 billion W0415 23:32:25.908940 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:32:25.979169 124618438372480 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0415 23:32:26.095686 124618438372480 checkpointing.py:776] Waited 0.1165010929107666 seconds for step 10 to finish before starting checkpointing. I0415 23:32:26.096137 124618438372480 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:32:26.096303 124618438372480 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0415 23:32:26.096642 124618438372480 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10. I0415 23:32:26.174521 124618438372480 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0415 23:32:26.252011 124618438372480 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:32:26.252109 124618438372480 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0415 23:32:26.269879 124515048752704 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10 W0415 23:32:26.497828 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:26.513346 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:26.529234 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:26.534210 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:26.538874 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:26.543020 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:26.547073 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:26.551146 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:32:26.974878 124515038266944 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10/items I0415 23:32:27.119680 124515006809664 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776295946884931864, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776295946884931864, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA I0415 23:32:35.532613 124618438372480 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.281004s I0415 23:32:35.533202 124618438372480 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.319 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0415 23:32:35.533262 124618438372480 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.357527s (batch_requests_ready=0.072900s, total_serialization_initiated=9.284148s, others=0.000479s) I0415 23:32:35.533330 124618438372480 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.358357s (all_items=0.000024s, per_item={'items': '0.00002384'}, temp_paths=9.358333) I0415 23:32:35.538814 124515048752704 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 23:32:35.538916 124618438372480 async_checkpointer.py:561] Finished blocking save. Time taken: 9.442568s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10. I0415 23:32:35.539126 124618438372480 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0415 23:32:35.539304 124515059238464 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 23:32:35.539393 124618438372480 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776295946.096121, 'wait_for_prev_duration_secs': 4.363059997558594e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776295946.0963244, 'checkpointer_blocking_duration_secs': 9.442700862884521, 'get_old_steps_start_time': 1776295955.539056, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776295946.0960126, 'checkpoint_manager_blocking_duration_secs': 9.443355321884155} I0415 23:32:35.539504 124618438372480 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0415 23:32:35.539548 124618438372480 max_utils.py:750] Memstats: After params initialized: I0415 23:32:35.539596 124618438372480 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0415 23:32:35.539621 124618438372480 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0415 23:32:35.539645 124618438372480 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0415 23:32:35.539664 124618438372480 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0415 23:32:35.539680 124618438372480 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0415 23:32:35.539698 124618438372480 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0415 23:32:35.539714 124618438372480 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0415 23:32:35.539730 124618438372480 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0415 23:32:35.545813 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:35.552322 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:32:35.557186 3376304 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:32:35.746478 124618438372480 metric_logger.py:185] completed step: 10, seconds: 0.917, TFLOP/s/device: 14.815, Tokens/s/device: 2233.073, total_weights: 16384, loss: 3.602 I0415 23:32:35.862415 124618438372480 metric_logger.py:185] completed step: 11, seconds: 9.766, TFLOP/s/device: 1.391, Tokens/s/device: 209.713, total_weights: 16384, loss: 3.101 I0415 23:32:35.981301 124618438372480 metric_logger.py:185] completed step: 12, seconds: 0.005, TFLOP/s/device: 2727.244, Tokens/s/device: 411079.888, total_weights: 16384, loss: 2.684 I0415 23:32:36.099630 124618438372480 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 116.246, Tokens/s/device: 17521.795, total_weights: 16384, loss: 2.348 I0415 23:32:36.103871 124618438372480 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0415 23:32:36.285897 124515017295424 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10/items/array_metadatas/process_0 I0415 23:32:36.336071 124618438372480 checkpointing.py:776] Waited 0.23215389251708984 seconds for step 15 to finish before starting checkpointing. I0415 23:32:36.336468 124618438372480 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 23:33:10.332145 124515038266944 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 34.794018s (commit=34.293009s, array_metadata_write=0.501009s) I0415 23:33:10.333143 124515048752704 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 286.176 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0415 23:33:10.333205 124515048752704 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 34.794240s. I0415 23:33:10.769118 124515048752704 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 23:33:11.005397 124515006809664 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776295946884931864, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA I0415 23:33:11.227664 124515048752704 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:33:11.228276 124515048752704 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.601943s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10/items I0415 23:33:11.228956 124515048752704 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10/items I0415 23:33:11.478671 124515048752704 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10 I0415 23:33:12.154888 124515048752704 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10`. I0415 23:33:12.155506 124515048752704 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.059164s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/10 I0415 23:33:12.155583 124515048752704 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 36.616623s. I0415 23:33:12.155730 124515059238464 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 23:33:12.155835 124515059238464 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 23:33:12.155927 124515059238464 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:33:12.155981 124515059238464 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0415 23:33:12.156240 124618438372480 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0415 23:33:12.156861 124618438372480 checkpoint_manager.py:1441] Waiting for previous save to complete took 35.820381 seconds. If this number is high, consider checkpointing less frequently. I0415 23:33:12.157570 124618438372480 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0415 23:33:12.157889 124618438372480 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15. I0415 23:33:12.340094 124618438372480 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:33:12.340234 124515059238464 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15 I0415 23:33:12.340311 124618438372480 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0415 23:33:13.047010 124514996323904 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15/items I0415 23:33:22.091571 124618438372480 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.752184s I0415 23:33:22.092094 124618438372480 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.254 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0415 23:33:22.092157 124618438372480 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.842157s (batch_requests_ready=0.085978s, total_serialization_initiated=9.755755s, others=0.000425s) I0415 23:33:22.092222 124618438372480 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.842675s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=9.842658) I0415 23:33:22.097210 124514996323904 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 23:33:22.097316 124618438372480 async_checkpointer.py:561] Finished blocking save. Time taken: 9.939685s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15. I0415 23:33:22.097612 124618438372480 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0415 23:33:22.097756 124515048752704 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 23:33:22.097830 124618438372480 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776295956.336444, 'wait_for_prev_duration_secs': 35.8203809261322, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776295992.1576087, 'checkpointer_blocking_duration_secs': 9.939880609512329, 'get_old_steps_start_time': 1776296002.0975137, 'get_old_steps_duration_secs': 3.981590270996094e-05, 'checkpoint_manager_blocking_start_time': 1776295956.3363788, 'checkpoint_manager_blocking_duration_secs': 45.761423110961914} I0415 23:33:22.097946 124618438372480 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0415 23:33:22.098380 124618438372480 metric_logger.py:185] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.259, Tokens/s/device: 17222.386, total_weights: 16384, loss: 2.083 I0415 23:33:22.104581 124618438372480 metric_logger.py:185] completed step: 15, seconds: 0.118, TFLOP/s/device: 115.091, Tokens/s/device: 17347.699, total_weights: 16384, loss: 1.880 I0415 23:33:22.220916 124618438372480 metric_logger.py:185] completed step: 16, seconds: 45.999, TFLOP/s/device: 0.295, Tokens/s/device: 44.522, total_weights: 16384, loss: 1.726 I0415 23:33:22.339589 124618438372480 metric_logger.py:185] completed step: 17, seconds: 0.005, TFLOP/s/device: 2829.473, Tokens/s/device: 426488.963, total_weights: 16384, loss: 1.610 I0415 23:33:22.458102 124618438372480 metric_logger.py:185] completed step: 18, seconds: 0.117, TFLOP/s/device: 115.796, Tokens/s/device: 17454.000, total_weights: 16384, loss: 1.521 I0415 23:33:22.576322 124618438372480 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0415 23:33:22.577117 124618438372480 checkpointing.py:776] Waited 0.0008091926574707031 seconds for step 19 to finish before starting checkpointing. I0415 23:33:22.577384 124618438372480 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 23:33:22.875858 124514325235264 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15/items/array_metadatas/process_0 I0415 23:33:55.317322 124516583867968 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.220905s (commit=32.767643s, array_metadata_write=0.453262s) I0415 23:33:55.318323 124514996323904 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 293.413 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 43 seconds) (per-host) I0415 23:33:55.318432 124514996323904 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.221069s. I0415 23:33:55.740819 124514996323904 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 23:33:56.194248 124514996323904 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:33:56.194867 124514996323904 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.596134s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15/items I0415 23:33:56.195516 124514996323904 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15/items I0415 23:33:56.441388 124514996323904 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15 I0415 23:33:57.101160 124514996323904 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15`. I0415 23:33:57.101762 124514996323904 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 44.944139s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/15 I0415 23:33:57.101837 124514996323904 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.004478s. I0415 23:33:57.101936 124515048752704 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 23:33:57.101985 124515048752704 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 23:33:57.102041 124515048752704 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:33:57.102100 124515048752704 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0415 23:33:57.102529 124618438372480 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0415 23:33:57.102945 124618438372480 checkpoint_manager.py:1441] Waiting for previous save to complete took 34.525577 seconds. If this number is high, consider checkpointing less frequently. I0415 23:33:57.103322 124618438372480 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0415 23:33:57.103614 124618438372480 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19. I0415 23:33:57.279848 124618438372480 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:33:57.279989 124515048752704 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19 I0415 23:33:57.280074 124618438372480 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0415 23:33:57.963441 124516541924928 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19/items I0415 23:34:09.792823 124618438372480 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 12.513672s I0415 23:34:09.793392 124618438372480 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1002.734 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 12 seconds) (per-host) I0415 23:34:09.793447 124618438372480 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 12.602405s (batch_requests_ready=0.085222s, total_serialization_initiated=12.516746s, others=0.000437s) I0415 23:34:09.793511 124618438372480 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 12.603123s (all_items=0.000012s, per_item={'items': '0.00001240'}, temp_paths=12.603111) I0415 23:34:09.798824 124516541924928 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 23:34:09.798921 124618438372480 async_checkpointer.py:561] Finished blocking save. Time taken: 12.695556s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19. I0415 23:34:09.799106 124618438372480 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0415 23:34:09.799276 124514996323904 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 23:34:09.799388 124618438372480 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776296002.5773528, 'wait_for_prev_duration_secs': 34.52557682991028, 'time_between_consecutive_saves_sec': 10.421356916427612, 'checkpointer_blocking_start_time': 1776296037.1033466, 'checkpointer_blocking_duration_secs': 12.695677757263184, 'get_old_steps_start_time': 1776296049.7990377, 'get_old_steps_duration_secs': 3.552436828613281e-05, 'checkpoint_manager_blocking_start_time': 1776296002.5773213, 'checkpoint_manager_blocking_duration_secs': 47.222041606903076} I0415 23:34:09.799514 124618438372480 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0415 23:34:09.799546 124618438372480 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 23:34:10.537149 124514786608704 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19/items/array_metadatas/process_0 I0415 23:34:40.950357 124516573382208 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.152369s (commit=30.722839s, array_metadata_write=0.429530s) I0415 23:34:40.951521 124516541924928 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 288.772 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 43 seconds) (per-host) I0415 23:34:40.951633 124516541924928 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.152669s. I0415 23:34:41.342288 124516541924928 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 23:34:41.765442 124516541924928 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:34:41.766096 124516541924928 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.559175s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19/items I0415 23:34:41.766730 124516541924928 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19/items I0415 23:34:42.001388 124516541924928 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19 I0415 23:34:42.679275 124516541924928 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19`. I0415 23:34:42.679879 124516541924928 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 45.576522s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_08_checkpoint_async_true/checkpoints/19 I0415 23:34:42.679952 124516541924928 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 32.880990s. I0415 23:34:42.680135 124514996323904 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 23:34:42.680236 124514996323904 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 23:34:42.680294 124514996323904 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:34:42.680333 124514996323904 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0415 23:34:42.680480 124618438372480 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0415 23:34:42.680613 124618438372480 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:34:42.681566 124618438372480 metric_logger.py:185] completed step: 19, seconds: 0.119, TFLOP/s/device: 114.466, Tokens/s/device: 17253.580, total_weights: 16384, loss: 1.449 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops