MaxView

2026-04-16 20:39:38.410341: 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 20:39:38.528092 129328134261888 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 20:40:34.438323 129328134261888 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 20:40:34.438431 129328134261888 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 20:40:34.438468 129328134261888 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 20:40:34.438495 129328134261888 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system
I0416 20:40:34.438518 129328134261888 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0416 20:40:34.438603 129328134261888 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 20:40:34.532732 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:40:34.938370 129328134261888 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 20:40:35.037379 129328134261888 checkpointing.py:688] Setting up checkpoint logger...
I0416 20:40:35.037472 129328134261888 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 20:40:35.037514 129328134261888 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 20:40:35.037714 129328134261888 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 0x759ef73fcfe0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 20:40:36.693313 129328134261888 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0416 20:40:36.693487 129328134261888 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 0x75987af49d00>}, handler_registry=None
I0416 20:40:36.693744 129328134261888 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x75987af49d00>` 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 20:40:36.693783 129328134261888 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x759879b856d0>` 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 20:40:36.693811 129328134261888 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 0x75987af49d00>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x75987af49d00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x759879b856d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x759879b856d0>}).
I0416 20:40:36.694133 129328134261888 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 20:40:36.694183 129328134261888 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x75987ab79760> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 20:40:37.124275 129230604076608 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776371826842351608, 'commit_timestamp_nsecs': 1776371859928809791, 'custom_metadata': {}} from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA
I0416 20:40:37.204267 129230604076608 checkpoint_manager.py:1847] Missing metrics for step 0
I0416 20:40:37.253390 129328134261888 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints
I0416 20:40:37.253869 129328134261888 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/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x759879b86ea0>
I0416 20:40:37.253960 129328134261888 checkpointing.py:302] Checkpoint manager created!
W0416 20:40:37.269931 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:40:37.481154 129328134261888 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 20:40:37.481254 129328134261888 checkpointing.py:582] restoring from this run's directory step 9
I0416 20:40:37.652422 129328134261888 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9.
I0416 20:40:38.210230 4120732 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 20:40:38.210293 4120732 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 20:40:44.094215 129328134261888 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 2.029 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host)
I0416 20:40:44.094648 129328134261888 checkpointer.py:318] Finished restoring checkpoint in 6.53 seconds from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9.
I0416 20:40:44.094776 129328134261888 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776372037.5652971, 'checkpointer_duration_secs': 6.5294225215911865, 'checkpoint_manager_start_time': 1776372037.481768, 'checkpoint_manager_duration_secs': 6.612958192825317}
[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 20:40:44.162966 129328134261888 maxtext_utils.py:1805]  decoder/decoder_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0416 20:40:44.163074 129328134261888 maxtext_utils.py:1805]  decoder/layers/mlp/wi_0/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0416 20:40:44.163115 129328134261888 maxtext_utils.py:1805]  decoder/layers/mlp/wi_1/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0416 20:40:44.163148 129328134261888 maxtext_utils.py:1805]  decoder/layers/mlp/wo/kernel/value
    Shape:     float32[7168,16,2048]
    Physical:  (None, None, 'fsdp')
I0416 20:40:44.163175 129328134261888 maxtext_utils.py:1805]  decoder/layers/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0416 20:40:44.163199 129328134261888 maxtext_utils.py:1805]  decoder/layers/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0416 20:40:44.163224 129328134261888 maxtext_utils.py:1805]  decoder/layers/self_attention/key/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:40:44.163249 129328134261888 maxtext_utils.py:1805]  decoder/layers/self_attention/out/kernel/value
    Shape:     float32[16,16,128,2048]
    Physical:  (None, None, None, 'fsdp')
I0416 20:40:44.163272 129328134261888 maxtext_utils.py:1805]  decoder/layers/self_attention/query/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:40:44.163294 129328134261888 maxtext_utils.py:1805]  decoder/layers/self_attention/value/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:40:44.163316 129328134261888 maxtext_utils.py:1805]  decoder/logits_dense/kernel/value
    Shape:     float32[2048,32000]
    Physical:  ('fsdp', None)
I0416 20:40:44.163339 129328134261888 maxtext_utils.py:1805]  token_embedder/embedding/value
    Shape:     float32[32000,2048]
    Physical:  (None, 'fsdp')

I0416 20:40:44.276767 129328134261888 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 20:40:44.276843 129328134261888 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:40:44.290230 129328134261888 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 20:40:44.290280 129328134261888 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:40:44.304970 129328134261888 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 20:40:44.305016 129328134261888 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:40:44.318822 129328134261888 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 20:40:44.318870 129328134261888 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:40:44.351351 129328134261888 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 20:40:44.351409 129328134261888 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:40:44.365386 129328134261888 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 20:40:44.365433 129328134261888 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:40:44.379341 129328134261888 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 20:40:44.379388 129328134261888 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:40:44.437122 129328134261888 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 20:40:44.437194 129328134261888 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:40:44.456306 129328134261888 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 20:40:44.456355 129328134261888 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
W0416 20:40:45.482474 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:40:45.547764 129328134261888 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 20:40:46.357557 129328134261888 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 20:40:46.359711 129328134261888 metric_logger.py:289] number parameters: 1.104 billion
W0416 20:40:47.149898 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:40:47.224149 129328134261888 checkpointing.py:794] Waiting for step 10 to finish before checkpoint...
I0416 20:40:47.340345 129328134261888 checkpointing.py:798] Waited 0.1161966323852539 seconds for step 10 to finish before starting checkpointing.
I0416 20:40:47.340813 129328134261888 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 20:40:47.340985 129328134261888 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10
I0416 20:40:47.341365 129328134261888 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10.
I0416 20:40:47.439933 129328134261888 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 20:40:47.529365 129328134261888 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0416 20:40:47.529453 129328134261888 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 20:40:47.540503 129229073155648 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10
W0416 20:40:48.167177 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:40:48.182559 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:40:48.195009 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:40:48.201058 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:40:48.205401 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:40:48.209533 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:40:48.213465 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:40:48.217740 129229062669888 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10/items
W0416 20:40:48.218146 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:40:48.380844 129229027018304 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776372048132535954, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776372048132535954, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA
I0416 20:40:57.381510 129328134261888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.853148s
I0416 20:40:57.388294 129328134261888 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.241 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host)
I0416 20:40:57.388355 129328134261888 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.947176s (batch_requests_ready=0.080422s, total_serialization_initiated=9.860077s, others=0.006676s)
I0416 20:40:57.388428 129328134261888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.948012s (all_items=0.000028s, per_item={'items': '0.00002789'}, temp_paths=9.947984)
I0416 20:40:57.389719 129229073155648 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 20:40:57.389874 129328134261888 async_checkpointer.py:561] Finished blocking save. Time taken: 10.048847s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10.
I0416 20:40:57.390105 129328134261888 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 20:40:57.390258 129229083641408 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 20:40:57.390338 129328134261888 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776372047.3407993, 'wait_for_prev_duration_secs': 3.981590270996094e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776372047.341006, 'checkpointer_blocking_duration_secs': 10.049000263214111, 'get_old_steps_start_time': 1776372057.390023, 'get_old_steps_duration_secs': 4.1484832763671875e-05, 'checkpoint_manager_blocking_start_time': 1776372047.3406994, 'checkpoint_manager_blocking_duration_secs': 10.049606561660767}
I0416 20:40:57.390425 129328134261888 checkpointing.py:409] Started an asynchronous checkpoint save for step 10
I0416 20:40:57.390479 129328134261888 max_utils.py:750] 
Memstats: After params initialized:
I0416 20:40:57.390528 129328134261888 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0))
I0416 20:40:57.390554 129328134261888 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0416 20:40:57.390575 129328134261888 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0416 20:40:57.390594 129328134261888 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0416 20:40:57.390610 129328134261888 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0416 20:40:57.390627 129328134261888 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0416 20:40:57.390644 129328134261888 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0416 20:40:57.390661 129328134261888 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0416 20:40:57.396609 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:40:57.402947 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:40:57.407580 4118845 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:40:57.598242 129328134261888 metric_logger.py:185] completed step: 10, seconds: 0.863, TFLOP/s/device: 15.741, Tokens/s/device: 2372.707, total_weights: 16384, loss: 3.587
I0416 20:40:57.711790 129328134261888 metric_logger.py:185] completed step: 11, seconds: 10.372, TFLOP/s/device: 1.310, Tokens/s/device: 197.457, total_weights: 16384, loss: 3.086
I0416 20:40:57.830113 129328134261888 metric_logger.py:185] completed step: 12, seconds: 0.007, TFLOP/s/device: 2025.209, Tokens/s/device: 305261.589, total_weights: 16384, loss: 2.671
I0416 20:40:57.948307 129328134261888 metric_logger.py:185] completed step: 13, seconds: 0.115, TFLOP/s/device: 118.185, Tokens/s/device: 17814.117, total_weights: 16384, loss: 2.335
I0416 20:40:57.953906 129328134261888 checkpointing.py:794] Waiting for step 15 to finish before checkpoint...
I0416 20:40:58.169583 129229041698368 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0
I0416 20:40:58.185764 129328134261888 checkpointing.py:798] Waited 0.23183345794677734 seconds for step 15 to finish before starting checkpointing.
I0416 20:40:58.186153 129328134261888 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 20:41:38.350268 129229062669888 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.961262s (commit=40.491507s, array_metadata_write=0.469755s)
I0416 20:41:38.351296 129229073155648 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 248.218 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 50 seconds) (per-host)
I0416 20:41:38.351412 129229073155648 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.961490s.
I0416 20:41:38.813725 129229073155648 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 20:41:39.018956 129229027018304 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776372048132535954, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA
I0416 20:41:39.251675 129229073155648 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 20:41:39.252263 129229073155648 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.578746s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10/items
I0416 20:41:39.252901 129229073155648 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10/items
I0416 20:41:39.497611 129229073155648 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10
I0416 20:41:40.191077 129229073155648 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10`.
I0416 20:41:40.191710 129229073155648 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 52.850684s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/10
I0416 20:41:40.191784 129229073155648 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 42.801867s.
I0416 20:41:40.191899 129229083641408 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 20:41:40.191953 129229083641408 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 20:41:40.192022 129229083641408 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0416 20:41:40.192079 129229083641408 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0416 20:41:40.192431 129328134261888 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 20:41:40.192551 129328134261888 checkpoint_manager.py:1441] Waiting for previous save to complete took 42.006396 seconds. If this number is high, consider checkpointing less frequently.
I0416 20:41:40.193394 129328134261888 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15
I0416 20:41:40.193691 129328134261888 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15.
I0416 20:41:40.379328 129229083641408 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15
I0416 20:41:40.388808 129328134261888 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0416 20:41:40.388902 129328134261888 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 20:41:41.124135 129229062669888 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15/items
I0416 20:41:52.691457 129328134261888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 12.303652s
I0416 20:41:52.697542 129328134261888 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1018.681 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 12 seconds) (per-host)
I0416 20:41:52.697618 129328134261888 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 12.405180s (batch_requests_ready=0.087892s, total_serialization_initiated=12.311290s, others=0.005998s)
I0416 20:41:52.697698 129328134261888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 12.405710s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=12.405692)
I0416 20:41:52.699036 129230530676288 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 20:41:52.699140 129328134261888 async_checkpointer.py:561] Finished blocking save. Time taken: 12.505692s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15.
I0416 20:41:52.699407 129328134261888 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 20:41:52.699582 129229073155648 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 20:41:52.699694 129328134261888 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776372058.1861296, 'wait_for_prev_duration_secs': 42.0063955783844, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776372100.1934175, 'checkpointer_blocking_duration_secs': 12.505859136581421, 'get_old_steps_start_time': 1776372112.6993043, 'get_old_steps_duration_secs': 4.1484832763671875e-05, 'checkpoint_manager_blocking_start_time': 1776372058.1860683, 'checkpoint_manager_blocking_duration_secs': 54.513590812683105}
I0416 20:41:52.699819 129328134261888 checkpointing.py:409] Started an asynchronous checkpoint save for step 15
I0416 20:41:52.700300 129328134261888 metric_logger.py:185] completed step: 14, seconds: 0.118, TFLOP/s/device: 115.182, Tokens/s/device: 17361.523, total_weights: 16384, loss: 2.072
I0416 20:41:52.709301 129328134261888 metric_logger.py:185] completed step: 15, seconds: 0.118, TFLOP/s/device: 115.212, Tokens/s/device: 17365.940, total_weights: 16384, loss: 1.869
I0416 20:41:52.823737 129328134261888 metric_logger.py:185] completed step: 16, seconds: 54.753, TFLOP/s/device: 0.248, Tokens/s/device: 37.404, total_weights: 16384, loss: 1.716
I0416 20:41:52.942195 129328134261888 metric_logger.py:185] completed step: 17, seconds: 0.007, TFLOP/s/device: 1845.325, Tokens/s/device: 278147.494, total_weights: 16384, loss: 1.601
I0416 20:41:53.060646 129328134261888 metric_logger.py:185] completed step: 18, seconds: 0.116, TFLOP/s/device: 117.276, Tokens/s/device: 17677.116, total_weights: 16384, loss: 1.513
I0416 20:41:53.179769 129328134261888 checkpointing.py:794] Waiting for step 19 to finish before checkpoint...
I0416 20:41:53.181012 129328134261888 checkpointing.py:798] Waited 0.001253366470336914 seconds for step 19 to finish before starting checkpointing.
I0416 20:41:53.181823 129328134261888 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 20:41:53.445637 129228439815744 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0
I0416 20:42:35.013778 129230583105088 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.315495s (commit=41.884406s, array_metadata_write=0.431089s)
I0416 20:42:35.014820 129230530676288 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 230.926 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 54 seconds) (per-host)
I0416 20:42:35.014923 129230530676288 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.315725s.
I0416 20:42:35.450502 129230530676288 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 20:42:35.889259 129230530676288 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 20:42:35.889877 129230530676288 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.581229s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15/items
I0416 20:42:35.890526 129230530676288 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15/items
I0416 20:42:36.140982 129230530676288 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15
I0416 20:42:36.822969 129230530676288 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15`.
I0416 20:42:36.823599 129230530676288 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 56.630168s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/15
I0416 20:42:36.823672 129230530676288 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 44.124476s.
I0416 20:42:36.823835 129229073155648 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 20:42:36.823939 129229073155648 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 20:42:36.824001 129229073155648 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0416 20:42:36.824069 129229073155648 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0416 20:42:36.824351 129328134261888 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 20:42:36.824772 129328134261888 checkpoint_manager.py:1441] Waiting for previous save to complete took 43.642968 seconds. If this number is high, consider checkpointing less frequently.
I0416 20:42:36.824990 129328134261888 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19
I0416 20:42:36.825277 129328134261888 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19.
I0416 20:42:36.994654 129229073155648 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19
I0416 20:42:37.022245 129328134261888 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0416 20:42:37.022337 129328134261888 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 20:42:37.703111 129230520190528 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19/items
I0416 20:42:50.536414 129328134261888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 13.515165s
I0416 20:42:50.543148 129328134261888 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 926.980 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 13 seconds) (per-host)
I0416 20:42:50.543211 129328134261888 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 13.632309s (batch_requests_ready=0.103346s, total_serialization_initiated=13.522334s, others=0.006629s)
I0416 20:42:50.543288 129328134261888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 13.633037s (all_items=0.000012s, per_item={'items': '0.00001192'}, temp_paths=13.633025)
I0416 20:42:50.544662 129230509704768 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 20:42:50.544820 129328134261888 async_checkpointer.py:561] Finished blocking save. Time taken: 13.719788s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19.
I0416 20:42:50.545011 129328134261888 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 20:42:50.545223 129230488733248 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 20:42:50.545332 129328134261888 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776372113.1817896, 'wait_for_prev_duration_secs': 43.64296841621399, 'time_between_consecutive_saves_sec': 12.989694118499756, 'checkpointer_blocking_start_time': 1776372156.8250153, 'checkpointer_blocking_duration_secs': 13.719935894012451, 'get_old_steps_start_time': 1776372170.544962, 'get_old_steps_duration_secs': 1.621246337890625e-05, 'checkpoint_manager_blocking_start_time': 1776372113.18176, 'checkpoint_manager_blocking_duration_secs': 57.363547563552856}
I0416 20:42:50.545450 129328134261888 checkpointing.py:409] Started an asynchronous checkpoint save for step 19
I0416 20:42:50.545480 129328134261888 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 20:42:51.299551 129230572619328 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0
I0416 20:43:37.118569 129230583105088 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 46.574675s (commit=46.127044s, array_metadata_write=0.447630s)
I0416 20:43:37.119902 129230509704768 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 209.882 MiB/s (total gbytes: 12.3 GiB) (time elapsed: a minute) (per-host)
I0416 20:43:37.120009 129230509704768 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.575147s.
I0416 20:43:37.526779 129230509704768 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 20:43:37.940615 129230509704768 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 20:43:37.941281 129230509704768 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.564402s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19/items
I0416 20:43:37.941965 129230509704768 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19/items
I0416 20:43:38.179538 129230509704768 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19
I0416 20:43:38.858392 129230509704768 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19`.
I0416 20:43:38.859021 129230509704768 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 62.033994s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/19
I0416 20:43:38.859104 129230509704768 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 48.314243s.
I0416 20:43:38.859314 129230488733248 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 20:43:38.859419 129230488733248 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 20:43:38.859474 129230488733248 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0416 20:43:38.859513 129230488733248 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0416 20:43:38.859659 129328134261888 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 20:43:38.859835 129328134261888 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 20:43:38.861145 129328134261888 metric_logger.py:185] completed step: 19, seconds: 0.119, TFLOP/s/device: 114.609, Tokens/s/device: 17275.120, total_weights: 16384, loss: 1.441
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops