MaxView

← Back to run

Log Summary

2026-04-16 20:46:28.866083: 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 20:46:28.983972 127964142390400 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 20:47:22.554646 127964142390400 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 20:47:22.554754 127964142390400 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 20:47:22.554789 127964142390400 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 20:47:22.554812 127964142390400 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system
I0416 20:47:22.554833 127964142390400 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0416 20:47:22.554905 127964142390400 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 20:47:22.647032 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:47:23.050451 127964142390400 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 20:47:23.148801 127964142390400 checkpointing.py:688] Setting up checkpoint logger...
I0416 20:47:23.148894 127964142390400 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True
I0416 20:47:23.148937 127964142390400 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 20:47:23.149132 127964142390400 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x746162ffce30>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 20:47:24.871482 127964142390400 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0416 20:47:24.871657 127964142390400 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 0x745ae5700800>}, handler_registry=None
I0416 20:47:24.871886 127964142390400 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x745ae5700800>` 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 20:47:24.871926 127964142390400 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x745ae67b4da0>` 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 20:47:24.871954 127964142390400 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 0x745ae5700800>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x745ae5700800>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x745ae67b4da0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x745ae67b4da0>}).
I0416 20:47:24.872265 127964142390400 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 20:47:24.872317 127964142390400 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x745ae6781760> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 20:47:25.334198 127858007606848 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776372310022678643, 'commit_timestamp_nsecs': 1776372318751895182, '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_false/checkpoints/0/_CHECKPOINT_METADATA
I0416 20:47:25.412688 127858007606848 checkpoint_manager.py:1847] Missing metrics for step 0
I0416 20:47:25.478914 127964142390400 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_false/checkpoints
I0416 20:47:25.479393 127964142390400 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_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x745ae67b40e0>
I0416 20:47:25.479481 127964142390400 checkpointing.py:302] Checkpoint manager created!
W0416 20:47:25.495705 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:47:25.709475 127964142390400 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 20:47:25.709569 127964142390400 checkpointing.py:582] restoring from this run's directory step 9
I0416 20:47:25.865211 127964142390400 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_false/checkpoints/9.
I0416 20:47:26.787007 4127794 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 20:47:26.787082 4127794 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 20:47:33.562899 127964142390400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.688 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host)
I0416 20:47:33.563802 127964142390400 checkpointer.py:318] Finished restoring checkpoint in 7.77 seconds from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/9.
I0416 20:47:33.563941 127964142390400 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_false/checkpoints', 'checkpointer_start_time': 1776372445.791484, 'checkpointer_duration_secs': 7.772392749786377, 'checkpoint_manager_start_time': 1776372445.7100835, 'checkpoint_manager_duration_secs': 7.853797197341919}
[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 20:47:33.634088 127964142390400 maxtext_utils.py:1805]  decoder/decoder_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0416 20:47:33.634178 127964142390400 maxtext_utils.py:1805]  decoder/layers/mlp/wi_0/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0416 20:47:33.634217 127964142390400 maxtext_utils.py:1805]  decoder/layers/mlp/wi_1/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0416 20:47:33.634248 127964142390400 maxtext_utils.py:1805]  decoder/layers/mlp/wo/kernel/value
    Shape:     float32[7168,16,2048]
    Physical:  (None, None, 'fsdp')
I0416 20:47:33.634275 127964142390400 maxtext_utils.py:1805]  decoder/layers/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0416 20:47:33.634300 127964142390400 maxtext_utils.py:1805]  decoder/layers/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0416 20:47:33.634326 127964142390400 maxtext_utils.py:1805]  decoder/layers/self_attention/key/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:47:33.634351 127964142390400 maxtext_utils.py:1805]  decoder/layers/self_attention/out/kernel/value
    Shape:     float32[16,16,128,2048]
    Physical:  (None, None, None, 'fsdp')
I0416 20:47:33.634375 127964142390400 maxtext_utils.py:1805]  decoder/layers/self_attention/query/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:47:33.634398 127964142390400 maxtext_utils.py:1805]  decoder/layers/self_attention/value/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:47:33.634421 127964142390400 maxtext_utils.py:1805]  decoder/logits_dense/kernel/value
    Shape:     float32[2048,32000]
    Physical:  ('fsdp', None)
I0416 20:47:33.634445 127964142390400 maxtext_utils.py:1805]  token_embedder/embedding/value
    Shape:     float32[32000,2048]
    Physical:  (None, 'fsdp')

I0416 20:47:33.748763 127964142390400 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 20:47:33.748845 127964142390400 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:47:33.797303 127964142390400 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 20:47:33.797369 127964142390400 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:47:33.813549 127964142390400 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 20:47:33.813597 127964142390400 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:47:33.833226 127964142390400 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 20:47:33.833310 127964142390400 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:47:33.868716 127964142390400 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 20:47:33.868778 127964142390400 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:47:33.882565 127964142390400 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 20:47:33.882612 127964142390400 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:47:33.896349 127964142390400 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 20:47:33.896395 127964142390400 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:47:34.005221 127964142390400 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 20:47:34.005303 127964142390400 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:47:34.028218 127964142390400 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 20:47:34.028272 127964142390400 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
W0416 20:47:34.943514 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:47:35.026161 127964142390400 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 20:47:35.026687 127964142390400 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 20:47:35.028028 127964142390400 metric_logger.py:289] number parameters: 1.104 billion
W0416 20:47:35.637400 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:47:35.713913 127964142390400 checkpointing.py:794] Waiting for step 10 to finish before checkpoint...
I0416 20:47:35.830032 127964142390400 checkpointing.py:798] Waited 0.11610746383666992 seconds for step 10 to finish before starting checkpointing.
I0416 20:47:35.830446 127964142390400 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 20:47:35.830599 127964142390400 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10
I0416 20:47:35.830896 127964142390400 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_false/checkpoints/10.
I0416 20:47:35.919831 127964142390400 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 20:47:36.000857 127856399091264 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_false/checkpoints/10
I0416 20:47:36.013249 127964142390400 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0416 20:47:36.013337 127964142390400 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 20:47:36.642555 127856388605504 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_false/checkpoints/10/items
W0416 20:47:36.651458 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:36.666642 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:36.682430 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:36.687273 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:36.693061 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:36.697198 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:36.701213 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:36.705269 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:47:36.795096 127856357148224 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776372456554862064, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776372456554862064, "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_false/checkpoints/10/_CHECKPOINT_METADATA
I0416 20:47:45.214280 127964142390400 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.202016s
I0416 20:47:45.220169 127964142390400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.327 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host)
I0416 20:47:45.220236 127964142390400 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.299501s (batch_requests_ready=0.084190s, total_serialization_initiated=9.209523s, others=0.005789s)
I0416 20:47:45.220320 127964142390400 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.300033s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=9.300014)
I0416 20:47:45.221531 127856378119744 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 20:47:45.221626 127964142390400 async_checkpointer.py:561] Finished blocking save. Time taken: 9.390987s. 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_false/checkpoints/10.
I0416 20:47:45.221813 127964142390400 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 20:47:45.222002 127856409577024 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 20:47:45.222119 127964142390400 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776372455.830434, 'wait_for_prev_duration_secs': 3.743171691894531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776372455.8306198, 'checkpointer_blocking_duration_secs': 9.391114950180054, 'get_old_steps_start_time': 1776372465.2217498, 'get_old_steps_duration_secs': 2.47955322265625e-05, 'checkpoint_manager_blocking_start_time': 1776372455.8303525, 'checkpoint_manager_blocking_duration_secs': 9.391732931137085}
I0416 20:47:45.222252 127964142390400 checkpointing.py:409] Started an asynchronous checkpoint save for step 10
I0416 20:47:45.222304 127964142390400 max_utils.py:750] 
Memstats: After params initialized:
I0416 20:47:45.222350 127964142390400 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0))
I0416 20:47:45.222374 127964142390400 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0416 20:47:45.222394 127964142390400 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0416 20:47:45.222412 127964142390400 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0416 20:47:45.222429 127964142390400 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0416 20:47:45.222446 127964142390400 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0416 20:47:45.222462 127964142390400 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0416 20:47:45.222478 127964142390400 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0416 20:47:45.228697 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:45.234905 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:47:45.239464 4126140 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:47:45.431726 127964142390400 metric_logger.py:185] completed step: 10, seconds: 0.685, TFLOP/s/device: 19.846, Tokens/s/device: 2991.406, total_weights: 16384, loss: 3.587
I0416 20:47:45.545462 127964142390400 metric_logger.py:185] completed step: 11, seconds: 9.715, TFLOP/s/device: 1.399, Tokens/s/device: 210.797, total_weights: 16384, loss: 3.086
I0416 20:47:45.664293 127964142390400 metric_logger.py:185] completed step: 12, seconds: 0.007, TFLOP/s/device: 1962.607, Tokens/s/device: 295825.509, total_weights: 16384, loss: 2.671
I0416 20:47:45.783606 127964142390400 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 116.143, Tokens/s/device: 17506.368, total_weights: 16384, loss: 2.335
I0416 20:47:45.790972 127964142390400 checkpointing.py:794] Waiting for step 15 to finish before checkpoint...
I0416 20:47:46.022165 127964142390400 checkpointing.py:798] Waited 0.23116302490234375 seconds for step 15 to finish before starting checkpointing.
I0416 20:47:46.022615 127964142390400 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 20:47:46.037117 127856367633984 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_false/checkpoints/10/items/array_metadatas/process_0
I0416 20:47:53.946792 127856388605504 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 8.725880s (commit=8.314103s, array_metadata_write=0.411777s)
I0416 20:47:53.947828 127856378119744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 700.990 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 18 seconds) (per-host)
I0416 20:47:53.947880 127856378119744 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 8.726212s.
W0416 20:47:54.279576 127856378119744 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/10/items, 
I0416 20:47:54.354664 127856378119744 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 20:47:54.354779 127856378119744 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.135261s. 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_false/checkpoints/10/items
I0416 20:47:54.355412 127856378119744 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_false/checkpoints/10/items
I0416 20:47:54.553840 127856357148224 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776372456554862064, '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_false/checkpoints/10/_CHECKPOINT_METADATA
I0416 20:47:54.594471 127856378119744 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_false/checkpoints/10
I0416 20:47:55.233693 127856378119744 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_false/checkpoints/10`.
I0416 20:47:55.234352 127856378119744 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 19.403720s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/10
I0416 20:47:55.234436 127856378119744 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 10.012769s.
I0416 20:47:55.234619 127856409577024 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 20:47:55.234725 127856409577024 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 20:47:55.234780 127856409577024 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0416 20:47:55.234827 127856409577024 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0416 20:47:55.235167 127964142390400 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 20:47:55.235592 127964142390400 checkpoint_manager.py:1441] Waiting for previous save to complete took 9.212986 seconds. If this number is high, consider checkpointing less frequently.
I0416 20:47:55.236353 127964142390400 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15
I0416 20:47:55.236639 127964142390400 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_false/checkpoints/15.
I0416 20:47:55.406816 127964142390400 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0416 20:47:55.406971 127964142390400 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 20:47:55.415224 127856409577024 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_false/checkpoints/15
I0416 20:47:56.079938 127856399091264 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_false/checkpoints/15/items
I0416 20:48:06.914182 127964142390400 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.508362s
I0416 20:48:06.920490 127964142390400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.064 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host)
I0416 20:48:06.920559 127964142390400 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.601257s (batch_requests_ready=0.079448s, total_serialization_initiated=11.515598s, others=0.006211s)
I0416 20:48:06.920636 127964142390400 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.601999s (all_items=0.000013s, per_item={'items': '0.00001287'}, temp_paths=11.601986)
I0416 20:48:06.922643 127856378119744 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 20:48:06.922739 127964142390400 async_checkpointer.py:561] Finished blocking save. Time taken: 11.686343s. 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_false/checkpoints/15.
I0416 20:48:06.922995 127964142390400 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 20:48:06.923166 127857651091008 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 20:48:06.923237 127964142390400 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776372466.022591, 'wait_for_prev_duration_secs': 9.212985515594482, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776372475.236378, 'checkpointer_blocking_duration_secs': 11.686540603637695, 'get_old_steps_start_time': 1776372486.9229345, 'get_old_steps_duration_secs': 2.0503997802734375e-05, 'checkpoint_manager_blocking_start_time': 1776372466.0225186, 'checkpoint_manager_blocking_duration_secs': 20.90069270133972}
I0416 20:48:06.923350 127964142390400 checkpointing.py:409] Started an asynchronous checkpoint save for step 15
I0416 20:48:06.923784 127964142390400 metric_logger.py:185] completed step: 14, seconds: 0.117, TFLOP/s/device: 116.385, Tokens/s/device: 17542.808, total_weights: 16384, loss: 2.072
I0416 20:48:06.932582 127964142390400 metric_logger.py:185] completed step: 15, seconds: 0.120, TFLOP/s/device: 112.951, Tokens/s/device: 17025.239, total_weights: 16384, loss: 1.869
I0416 20:48:07.046331 127964142390400 metric_logger.py:185] completed step: 16, seconds: 21.139, TFLOP/s/device: 0.643, Tokens/s/device: 96.880, total_weights: 16384, loss: 1.716
I0416 20:48:07.164430 127964142390400 metric_logger.py:185] completed step: 17, seconds: 0.008, TFLOP/s/device: 1806.079, Tokens/s/device: 272231.822, total_weights: 16384, loss: 1.601
I0416 20:48:07.282901 127964142390400 metric_logger.py:185] completed step: 18, seconds: 0.115, TFLOP/s/device: 117.711, Tokens/s/device: 17742.662, total_weights: 16384, loss: 1.513
I0416 20:48:07.402699 127964142390400 checkpointing.py:794] Waiting for step 19 to finish before checkpoint...
I0416 20:48:07.404545 127964142390400 checkpointing.py:798] Waited 0.0018444061279296875 seconds for step 19 to finish before starting checkpointing.
I0416 20:48:07.404939 127964142390400 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 20:48:07.701891 127855602173504 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_false/checkpoints/15/items/array_metadatas/process_0
I0416 20:48:14.644329 127855591687744 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 7.722415s (commit=7.294992s, array_metadata_write=0.427423s)
I0416 20:48:14.645412 127856378119744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 653.873 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 19 seconds) (per-host)
I0416 20:48:14.645516 127856378119744 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 7.722734s.
W0416 20:48:14.940488 127856378119744 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/15/items, 
I0416 20:48:15.017423 127856378119744 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 20:48:15.017525 127856378119744 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.125966s. 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_false/checkpoints/15/items
I0416 20:48:15.018181 127856378119744 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_false/checkpoints/15/items
I0416 20:48:15.260570 127856378119744 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_false/checkpoints/15
I0416 20:48:15.929877 127856378119744 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_false/checkpoints/15`.
I0416 20:48:15.930516 127856378119744 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 20.694125s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/15
I0416 20:48:15.930582 127856378119744 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 9.007802s.
I0416 20:48:15.930681 127857651091008 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 20:48:15.930728 127857651091008 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 20:48:15.930774 127857651091008 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0416 20:48:15.930810 127857651091008 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0416 20:48:15.931036 127964142390400 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 20:48:15.931156 127964142390400 checkpoint_manager.py:1441] Waiting for previous save to complete took 8.526234 seconds. If this number is high, consider checkpointing less frequently.
I0416 20:48:15.931346 127964142390400 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19
I0416 20:48:15.931622 127964142390400 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_false/checkpoints/19.
I0416 20:48:16.095199 127856378119744 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_false/checkpoints/19
I0416 20:48:16.106793 127964142390400 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0416 20:48:16.106882 127964142390400 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 20:48:16.774550 127855591687744 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_false/checkpoints/19/items
I0416 20:48:27.758254 127964142390400 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.652412s
I0416 20:48:27.763864 127964142390400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.050 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host)
I0416 20:48:27.763931 127964142390400 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.753053s (batch_requests_ready=0.087874s, total_serialization_initiated=11.659670s, others=0.005509s)
I0416 20:48:27.764010 127964142390400 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.753527s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=11.753516)
I0416 20:48:27.765338 127857661576768 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 20:48:27.765433 127964142390400 async_checkpointer.py:561] Finished blocking save. Time taken: 11.834046s. 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_false/checkpoints/19.
I0416 20:48:27.765596 127964142390400 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 20:48:27.765725 127857651091008 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 20:48:27.765805 127964142390400 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776372487.4049077, 'wait_for_prev_duration_secs': 8.52623438835144, 'time_between_consecutive_saves_sec': 12.170065641403198, 'checkpointer_blocking_start_time': 1776372495.9313676, 'checkpointer_blocking_duration_secs': 11.8341703414917, 'get_old_steps_start_time': 1776372507.7655487, 'get_old_steps_duration_secs': 1.5735626220703125e-05, 'checkpoint_manager_blocking_start_time': 1776372487.4048584, 'checkpoint_manager_blocking_duration_secs': 20.360921382904053}
I0416 20:48:27.765953 127964142390400 checkpointing.py:409] Started an asynchronous checkpoint save for step 19
I0416 20:48:27.765984 127964142390400 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 20:48:28.494652 127856768190016 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_false/checkpoints/19/items/array_metadatas/process_0
I0416 20:48:36.601249 127856220833344 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 8.836621s (commit=8.399890s, array_metadata_write=0.436731s)
I0416 20:48:36.602199 127857661576768 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 613.697 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 20 seconds) (per-host)
I0416 20:48:36.602297 127857661576768 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 8.836821s.
W0416 20:48:36.916749 127857661576768 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/19/items, 
I0416 20:48:36.999455 127857661576768 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 20:48:36.999560 127857661576768 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.142189s. 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_false/checkpoints/19/items
I0416 20:48:37.000204 127857661576768 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_false/checkpoints/19/items
I0416 20:48:37.231804 127857661576768 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_false/checkpoints/19
I0416 20:48:37.886135 127857661576768 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_false/checkpoints/19`.
I0416 20:48:37.886764 127857661576768 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 21.955386s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/19
I0416 20:48:37.886827 127857661576768 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 10.121352s.
I0416 20:48:37.886929 127857651091008 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 20:48:37.886976 127857651091008 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 20:48:37.887024 127857651091008 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0416 20:48:37.887076 127857651091008 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0416 20:48:37.887193 127964142390400 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 20:48:37.887345 127964142390400 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 20:48:37.887900 127964142390400 metric_logger.py:185] completed step: 19, seconds: 0.121, TFLOP/s/device: 112.551, Tokens/s/device: 16964.877, 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