2026-04-16 00:05:20.026918: 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) I0416 00:05:20.143324 126739104050304 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 00:06:15.669029 126739104050304 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 00:06:15.669155 126739104050304 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 00:06:15.669192 126739104050304 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 I0416 00:06:15.669225 126739104050304 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0416 00:06:15.669248 126739104050304 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 00:06:15.669362 126739104050304 train.py:715] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 00:06:15.763958 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:06:16.167983 126739104050304 maxtext_utils.py:1520] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 00:06:16.168369 126739104050304 checkpointing.py:677] Setting up checkpoint logger... I0416 00:06:16.168422 126739104050304 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 00:06:16.168497 126739104050304 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 00:06:16.169256 126739104050304 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 0x734428ff8d40>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 00:06:17.818887 126739104050304 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0416 00:06:17.819116 126739104050304 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 0x733dac7357c0>}, handler_registry=None I0416 00:06:17.819458 126739104050304 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x733dac7357c0>` 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`. I0416 00:06:17.819503 126739104050304 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x733dac736570>` 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`. I0416 00:06:17.819534 126739104050304 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 0x733dac7357c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x733dac7357c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x733dac736570>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x733dac736570>}). I0416 00:06:17.819847 126739104050304 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 00:06:17.819907 126739104050304 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x733dac7394e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 00:06:18.239959 126637272204864 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776297848141695658, 'commit_timestamp_nsecs': 1776297890655637953, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0416 00:06:18.318311 126637272204864 checkpoint_manager.py:1847] Missing metrics for step 9 I0416 00:06:18.375081 126739104050304 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_15_ocdbt_true/checkpoints I0416 00:06:18.375555 126739104050304 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_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x733dacb44170> I0416 00:06:18.375650 126739104050304 checkpointing.py:301] Checkpoint manager created! W0416 00:06:18.392490 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:06:18.575478 126739104050304 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 00:06:18.575580 126739104050304 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 00:06:18.667852 126739104050304 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 00:06:18.667929 126739104050304 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 00:06:18.682183 126739104050304 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 00:06:18.682231 126739104050304 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 00:06:18.702699 126739104050304 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 00:06:18.702754 126739104050304 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 00:06:18.716937 126739104050304 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 00:06:18.716985 126739104050304 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 00:06:18.731150 126739104050304 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 00:06:18.731196 126739104050304 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 00:06:18.752988 126739104050304 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 00:06:18.753058 126739104050304 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 00:06:18.771355 126739104050304 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 00:06:18.771407 126739104050304 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 00:06:18.955077 126739104050304 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0416 00:06:18.955173 126739104050304 checkpointing.py:581] restoring from this run's directory step 9 I0416 00:06:19.138564 126739104050304 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/9. I0416 00:06:19.665142 3412372 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 00:06:19.665207 3412372 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 00:06:25.845793 126739104050304 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.940 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0416 00:06:25.846201 126739104050304 checkpointer.py:318] Finished restoring checkpoint in 6.80 seconds from gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/9. I0416 00:06:25.846338 126739104050304 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_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776297979.0470273, 'checkpointer_duration_secs': 6.79924464225769, 'checkpoint_manager_start_time': 1776297978.955409, 'checkpoint_manager_duration_secs': 6.890866279602051} [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 I0416 00:06:26.493122 126739104050304 maxtext_utils.py:1623] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 00:06:26.493234 126739104050304 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) I0416 00:06:26.493276 126739104050304 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) I0416 00:06:26.493320 126739104050304 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') I0416 00:06:26.493360 126739104050304 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) I0416 00:06:26.493387 126739104050304 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) I0416 00:06:26.493425 126739104050304 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) I0416 00:06:26.493463 126739104050304 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') I0416 00:06:26.493493 126739104050304 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) I0416 00:06:26.493519 126739104050304 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) I0416 00:06:26.493552 126739104050304 maxtext_utils.py:1623] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed', 'vocab') Physical: ('fsdp', None) I0416 00:06:26.493587 126739104050304 maxtext_utils.py:1623] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed') Physical: (None, 'fsdp') I0416 00:06:27.214426 126739104050304 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 00:06:27.214508 126739104050304 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 00:06:27.228206 126739104050304 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 00:06:27.228254 126739104050304 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 00:06:27.914773 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:06:27.982649 126739104050304 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. I0416 00:06:27.983180 126739104050304 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 00:06:27.983325 126739104050304 metric_logger.py:289] number parameters: 1.104 billion W0416 00:06:28.823025 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:06:28.894216 126739104050304 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0416 00:06:29.010763 126739104050304 checkpointing.py:776] Waited 0.11653399467468262 seconds for step 10 to finish before starting checkpointing. I0416 00:06:29.011187 126739104050304 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 00:06:29.011353 126739104050304 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0416 00:06:29.011638 126739104050304 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_15_ocdbt_true/checkpoints/10. I0416 00:06:29.094649 126739104050304 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 00:06:29.185092 126635701438016 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/10 I0416 00:06:29.199273 126739104050304 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 00:06:29.199359 126739104050304 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 W0416 00:06:29.423692 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:29.452447 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:29.458544 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:29.463151 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:29.467272 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:29.471278 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:29.475365 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:29.479509 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:06:29.912019 126635690952256 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/10/items I0416 00:06:30.083949 126635659494976 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776297989830311666, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776297989830311666, "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_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0416 00:06:37.567915 126739104050304 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.369054s I0416 00:06:37.568544 126739104050304 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.456 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0416 00:06:37.568606 126739104050304 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.473018s (batch_requests_ready=0.100216s, total_serialization_initiated=8.372325s, others=0.000477s) I0416 00:06:37.568671 126739104050304 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.473547s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=8.473528) I0416 00:06:37.573780 126635680466496 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 00:06:37.573904 126739104050304 async_checkpointer.py:561] Finished blocking save. Time taken: 8.562511s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/10. I0416 00:06:37.574168 126739104050304 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0416 00:06:37.574382 126635711923776 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 00:06:37.574491 126739104050304 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_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776297989.0111744, 'wait_for_prev_duration_secs': 4.029273986816406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776297989.011375, 'checkpointer_blocking_duration_secs': 8.562711954116821, 'get_old_steps_start_time': 1776297997.574108, 'get_old_steps_duration_secs': 2.384185791015625e-05, 'checkpoint_manager_blocking_start_time': 1776297989.0110915, 'checkpoint_manager_blocking_duration_secs': 8.563364028930664} I0416 00:06:37.574613 126739104050304 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0416 00:06:37.574658 126739104050304 max_utils.py:750] Memstats: After params initialized: I0416 00:06:37.574701 126739104050304 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0416 00:06:37.574726 126739104050304 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 00:06:37.574746 126739104050304 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 00:06:37.574764 126739104050304 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 00:06:37.574781 126739104050304 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 00:06:37.574798 126739104050304 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 00:06:37.574814 126739104050304 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 00:06:37.574830 126739104050304 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 00:06:37.580521 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:37.586296 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:06:37.590785 3410713 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:06:37.784184 126739104050304 metric_logger.py:185] completed step: 10, seconds: 0.910, TFLOP/s/device: 14.925, Tokens/s/device: 2249.595, total_weights: 16384, loss: 3.602 I0416 00:06:37.900321 126739104050304 metric_logger.py:185] completed step: 11, seconds: 8.888, TFLOP/s/device: 1.529, Tokens/s/device: 230.412, total_weights: 16384, loss: 3.101 I0416 00:06:38.019087 126739104050304 metric_logger.py:185] completed step: 12, seconds: 0.005, TFLOP/s/device: 2728.887, Tokens/s/device: 411327.576, total_weights: 16384, loss: 2.684 I0416 00:06:38.137031 126739104050304 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 116.232, Tokens/s/device: 17519.697, total_weights: 16384, loss: 2.348 I0416 00:06:38.141326 126739104050304 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0416 00:06:38.299391 126635669980736 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_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0 I0416 00:06:38.373124 126739104050304 checkpointing.py:776] Waited 0.23176908493041992 seconds for step 15 to finish before starting checkpointing. I0416 00:06:38.373496 126739104050304 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 00:07:13.769613 126635690952256 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.196788s (commit=35.748125s, array_metadata_write=0.448663s) I0416 00:07:13.770723 126635680466496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 282.860 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0416 00:07:13.770826 126635680466496 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.196878s. I0416 00:07:14.208765 126635680466496 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 00:07:14.404142 126635659494976 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776297989830311666, '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_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0416 00:07:14.617394 126635680466496 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 00:07:14.617561 126635680466496 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.557753s. 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_15_ocdbt_true/checkpoints/10/items I0416 00:07:14.618230 126635680466496 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/10/items I0416 00:07:14.850512 126635680466496 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/10 I0416 00:07:15.504353 126635680466496 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_15_ocdbt_true/checkpoints/10`. I0416 00:07:15.504972 126635680466496 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.493586s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/10 I0416 00:07:15.505061 126635680466496 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.931101s. I0416 00:07:15.505222 126635711923776 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 00:07:15.505321 126635711923776 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 00:07:15.505388 126635711923776 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0416 00:07:15.505435 126635711923776 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0416 00:07:15.505770 126739104050304 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. W0416 00:07:15.506677 126739104050304 checkpoint_manager.py:1441] Waiting for previous save to complete took 37.133178 seconds. If this number is high, consider checkpointing less frequently. I0416 00:07:15.507474 126739104050304 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0416 00:07:15.507756 126739104050304 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_15_ocdbt_true/checkpoints/15. I0416 00:07:15.676056 126635711923776 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/15 I0416 00:07:15.684196 126739104050304 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 00:07:15.684285 126739104050304 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 I0416 00:07:16.424269 126635649009216 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/15/items I0416 00:07:27.276571 126739104050304 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.592765s I0416 00:07:27.281133 126739104050304 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.056 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0416 00:07:27.281199 126739104050304 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.689788s (batch_requests_ready=0.088956s, total_serialization_initiated=11.596401s, others=0.004431s) I0416 00:07:27.281313 126739104050304 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.690575s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=11.690561) I0416 00:07:27.282912 126635680466496 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 00:07:27.283012 126739104050304 async_checkpointer.py:561] Finished blocking save. Time taken: 11.775481s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/15. I0416 00:07:27.283285 126739104050304 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0416 00:07:27.283458 126637188318784 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 00:07:27.283559 126739104050304 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_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776297998.3734727, 'wait_for_prev_duration_secs': 37.133177757263184, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776298035.5074973, 'checkpointer_blocking_duration_secs': 11.775634527206421, 'get_old_steps_start_time': 1776298047.2831705, 'get_old_steps_duration_secs': 4.7206878662109375e-05, 'checkpoint_manager_blocking_start_time': 1776297998.3734105, 'checkpoint_manager_blocking_duration_secs': 48.91011691093445} I0416 00:07:27.283714 126739104050304 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0416 00:07:27.284209 126739104050304 metric_logger.py:185] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.461, Tokens/s/device: 17252.854, total_weights: 16384, loss: 2.083 I0416 00:07:27.291866 126739104050304 metric_logger.py:185] completed step: 15, seconds: 0.118, TFLOP/s/device: 115.183, Tokens/s/device: 17361.670, total_weights: 16384, loss: 1.880 I0416 00:07:27.407431 126739104050304 metric_logger.py:185] completed step: 16, seconds: 49.149, TFLOP/s/device: 0.276, Tokens/s/device: 41.670, total_weights: 16384, loss: 1.726 I0416 00:07:27.525640 126739104050304 metric_logger.py:185] completed step: 17, seconds: 0.006, TFLOP/s/device: 2418.930, Tokens/s/device: 364607.442, total_weights: 16384, loss: 1.610 I0416 00:07:27.643575 126739104050304 metric_logger.py:185] completed step: 18, seconds: 0.117, TFLOP/s/device: 116.541, Tokens/s/device: 17566.281, total_weights: 16384, loss: 1.521 I0416 00:07:27.761846 126739104050304 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0416 00:07:27.762723 126739104050304 checkpointing.py:776] Waited 0.0008983612060546875 seconds for step 19 to finish before starting checkpointing. I0416 00:07:27.762989 126739104050304 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 00:07:28.035278 126634925491776 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_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0 I0416 00:08:03.150350 126637240747584 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.868422s (commit=35.425517s, array_metadata_write=0.442905s) I0416 00:08:03.151365 126635680466496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 265.702 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 47 seconds) (per-host) I0416 00:08:03.151475 126635680466496 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.868509s. I0416 00:08:03.619820 126635680466496 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 00:08:04.053493 126635680466496 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 00:08:04.053663 126635680466496 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.572257s. 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_15_ocdbt_true/checkpoints/15/items I0416 00:08:04.054339 126635680466496 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/15/items I0416 00:08:04.277577 126635680466496 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/15 I0416 00:08:04.962431 126635680466496 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_15_ocdbt_true/checkpoints/15`. I0416 00:08:04.963067 126635680466496 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 49.455539s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/15 I0416 00:08:04.963148 126635680466496 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.680188s. I0416 00:08:04.963247 126637188318784 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 00:08:04.963300 126637188318784 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 00:08:04.963370 126637188318784 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0416 00:08:04.963420 126637188318784 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0416 00:08:04.963719 126739104050304 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. W0416 00:08:04.964121 126739104050304 checkpoint_manager.py:1441] Waiting for previous save to complete took 37.201138 seconds. If this number is high, consider checkpointing less frequently. I0416 00:08:04.964352 126739104050304 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0416 00:08:04.964657 126739104050304 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_15_ocdbt_true/checkpoints/19. I0416 00:08:05.145369 126637188318784 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/19 I0416 00:08:05.151907 126739104050304 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 00:08:05.152001 126739104050304 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 I0416 00:08:05.868779 126637177833024 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/19/items I0416 00:08:19.111239 126739104050304 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 13.959710s I0416 00:08:19.115976 126739104050304 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 898.260 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 14 seconds) (per-host) I0416 00:08:19.116035 126739104050304 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 14.068155s (batch_requests_ready=0.100579s, total_serialization_initiated=13.962970s, others=0.004606s) I0416 00:08:19.116139 126739104050304 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 14.068918s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=14.068902) I0416 00:08:19.117730 126637198804544 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 00:08:19.117828 126739104050304 async_checkpointer.py:561] Finished blocking save. Time taken: 14.153435s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/19. I0416 00:08:19.118036 126739104050304 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0416 00:08:19.118245 126637167347264 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 00:08:19.118341 126739104050304 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_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776298047.762963, 'wait_for_prev_duration_secs': 37.20113801956177, 'time_between_consecutive_saves_sec': 12.25751280784607, 'checkpointer_blocking_start_time': 1776298084.9643745, 'checkpointer_blocking_duration_secs': 14.15356993675232, 'get_old_steps_start_time': 1776298099.1179664, 'get_old_steps_duration_secs': 2.8371810913085938e-05, 'checkpoint_manager_blocking_start_time': 1776298047.7629302, 'checkpoint_manager_blocking_duration_secs': 51.355384826660156} I0416 00:08:19.118674 126739104050304 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0416 00:08:19.118712 126739104050304 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 00:08:20.695905 126635800004160 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_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0 I0416 00:08:59.041933 126637230261824 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.925290s (commit=39.495687s, array_metadata_write=0.429603s) I0416 00:08:59.042982 126637198804544 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 234.036 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 53 seconds) (per-host) I0416 00:08:59.043100 126637198804544 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.925221s. I0416 00:08:59.481289 126637198804544 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 00:08:59.902521 126637198804544 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 00:08:59.903222 126637198804544 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.566472s. 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_15_ocdbt_true/checkpoints/19/items I0416 00:08:59.903851 126637198804544 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/19/items I0416 00:09:00.145230 126637198804544 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/19 I0416 00:09:00.805452 126637198804544 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_15_ocdbt_true/checkpoints/19`. I0416 00:09:00.806110 126637198804544 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 55.841722s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_15_ocdbt_true/checkpoints/19 I0416 00:09:00.806187 126637198804544 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.688313s. I0416 00:09:00.806365 126637167347264 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 00:09:00.806476 126637167347264 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 00:09:00.806550 126637167347264 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0416 00:09:00.806602 126637167347264 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0416 00:09:00.806752 126739104050304 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. I0416 00:09:00.806892 126739104050304 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 00:09:00.807488 126739104050304 metric_logger.py:185] completed step: 19, seconds: 0.118, TFLOP/s/device: 115.035, Tokens/s/device: 17339.328, 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