MaxView

← Back to run

Log Summary

2026-04-23 04:00:24.364000: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1776916824.377934 2405275 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1776916824.381979 2405275 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
W0000 00:00:1776916824.394581 2405275 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776916824.394597 2405275 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776916824.394598 2405275 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776916824.394600 2405275 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
`rope_scaling`'s factor field must be a float >= 1, got 40
`rope_scaling`'s beta_fast field must be a float, got 32
`rope_scaling`'s beta_slow field must be a float, got 1
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
2026-04-23 04:00:28.046803: 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)
I0423 04:00:28.181604 133686539519808 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
~/maxtext/maxtext_venv/lib/python3.12/site-packages/jax/_src/xla_bridge.py:204: UserWarning: TPU backend initialization is taking more than 60.0 seconds. Did you run your code on all TPU hosts? See https://docs.jax.dev/en/latest/multi_process.html for more information.
  warnings.warn(
I0423 04:01:35.407818 133686539519808 max_utils.py:800] System Information: Jax Version: 0.8.1
I0423 04:01:35.408089 133686539519808 max_utils.py:801] System Information: Jaxlib Version: 0.8.1
I0423 04:01:35.408138 133686539519808 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709
I0423 04:01:35.408170 133686539519808 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system
I0423 04:01:35.408194 133686539519808 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0423 04:01:35.408356 133686539519808 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
I0423 04:01:36.071462 133686539519808 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0423 04:01:36.071727 133686539519808 checkpointing.py:677] Setting up checkpoint logger...
I0423 04:01:36.071773 133686539519808 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0423 04:01:36.071810 133686539519808 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0423 04:01:36.072261 133686539519808 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 0x79960bbaf170>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0423 04:01:37.784850 133686539519808 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0423 04:01:37.785219 133686539519808 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 0x7994ad123bf0>}, handler_registry=None
I0423 04:01:37.785445 133686539519808 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7994ad123bf0>` 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`.
I0423 04:01:37.785486 133686539519808 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79943cb35070>` 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`.
I0423 04:01:37.785515 133686539519808 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 0x7994ad123bf0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7994ad123bf0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79943cb35070>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79943cb35070>}).
I0423 04:01:37.785795 133686539519808 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0423 04:01:37.785844 133686539519808 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x79943cbf8540> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0423 04:01:38.225538 133656876156480 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776916703087360117, 'commit_timestamp_nsecs': 1776916750421268062, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/5/_CHECKPOINT_METADATA
I0423 04:01:38.306797 133656865670720 checkpoint_manager.py:1847] Missing metrics for step 9
I0423 04:01:38.326034 133686539519808 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints
I0423 04:01:38.326516 133686539519808 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_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7994ab3d2180>
I0423 04:01:38.326615 133686539519808 checkpointing.py:301] Checkpoint manager created!
I0423 04:01:38.713806 133686539519808 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0423 04:01:38.713898 133686539519808 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0423 04:01:38.990736 133686539519808 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 04:01:38.990818 133686539519808 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0423 04:01:39.003103 133686539519808 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 04:01:39.003152 133686539519808 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0423 04:01:39.022016 133686539519808 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 04:01:39.022085 133686539519808 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0423 04:01:39.034348 133686539519808 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 04:01:39.034399 133686539519808 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0423 04:01:39.046581 133686539519808 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 04:01:39.046630 133686539519808 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0423 04:01:39.065916 133686539519808 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0423 04:01:39.065971 133686539519808 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0423 04:01:39.081352 133686539519808 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0423 04:01:39.081406 133686539519808 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0423 04:01:39.230539 133686539519808 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0423 04:01:39.230632 133686539519808 checkpointing.py:581] restoring from this run's directory step 9
I0423 04:01:39.413574 133686539519808 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/9.
I0423 04:01:40.038998 2406836 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0423 04:01:40.039073 2406836 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0423 04:01:48.899719 133686539519808 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.354 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host)
I0423 04:01:48.900076 133686539519808 checkpointer.py:318] Finished restoring checkpoint in 9.57 seconds from gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/9.
I0423 04:01:48.900182 133686539519808 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1776916899.332782, 'checkpointer_duration_secs': 9.567360639572144, 'checkpoint_manager_start_time': 1776916899.2308388, 'checkpoint_manager_duration_secs': 9.669309139251709}
[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
I0423 04:01:49.295222 133686539519808 maxtext_utils.py:1707]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   PartitionSpec('norm',)
    Physical:  (None,)
I0423 04:01:49.295333 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/mlp/wi_0/kernel
    Shape:     float32[2048,16,7168]
    Logical:   PartitionSpec('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0423 04:01:49.295376 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/mlp/wi_1/kernel
    Shape:     float32[2048,16,7168]
    Logical:   PartitionSpec('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0423 04:01:49.295422 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/mlp/wo/kernel
    Shape:     float32[7168,16,2048]
    Logical:   PartitionSpec('mlp', 'layers', 'embed')
    Physical:  (None, None, 'fsdp')
I0423 04:01:49.295463 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/post_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   PartitionSpec('norm', 'layers')
    Physical:  (None, None)
I0423 04:01:49.295492 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/pre_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   PartitionSpec('norm', 'layers')
    Physical:  (None, None)
I0423 04:01:49.295532 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/self_attention/key/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0423 04:01:49.295573 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/self_attention/out/kernel
    Shape:     float32[16,16,128,2048]
    Logical:   PartitionSpec('heads', 'layers', 'kv', 'embed')
    Physical:  (None, None, None, 'fsdp')
I0423 04:01:49.295602 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/self_attention/query/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None, None)
I0423 04:01:49.295629 133686539519808 maxtext_utils.py:1707]  params/params/decoder/layers/self_attention/value/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0423 04:01:49.295665 133686539519808 maxtext_utils.py:1707]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   PartitionSpec('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0423 04:01:49.295702 133686539519808 maxtext_utils.py:1707]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   PartitionSpec('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')

I0423 04:01:49.692582 133686539519808 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0423 04:01:49.692662 133686539519808 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0423 04:01:49.704082 133686539519808 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0423 04:01:49.704130 133686539519808 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
I0423 04:01:50.498016 133686539519808 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.
I0423 04:01:50.501698 133686539519808 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0423 04:01:50.501919 133686539519808 metric_logger.py:301] number parameters: 1.104 billion
I0423 04:01:52.369032 133686539519808 checkpointing.py:772] Waiting for step 10 to finish before checkpoint...
I0423 04:01:52.486699 133686539519808 checkpointing.py:776] Waited 0.11765480041503906 seconds for step 10 to finish before starting checkpointing.
I0423 04:01:52.487318 133686539519808 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 04:01:52.487476 133686539519808 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10
I0423 04:01:52.487764 133686539519808 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10.
I0423 04:01:52.571032 133686539519808 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0423 04:01:52.651663 133654841919040 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10
I0423 04:01:52.664182 133686539519808 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0423 04:01:52.664268 133686539519808 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
I0423 04:01:53.413447 133654831433280 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10/items
I0423 04:01:53.576828 133654445557312 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776916913324873408, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776916913324873408, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA
I0423 04:01:54.289543 133686539519808 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.625729s
I0423 04:01:54.294322 133686539519808 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 7.167 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host)
I0423 04:01:54.294463 133686539519808 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.722097s (batch_requests_ready=0.088865s, total_serialization_initiated=1.628606s, others=0.004626s)
I0423 04:01:54.294656 133686539519808 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.723159s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=1.723139)
I0423 04:01:54.296368 133654852404800 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0423 04:01:54.296483 133686539519808 async_checkpointer.py:561] Finished blocking save. Time taken: 1.808964s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10.
I0423 04:01:54.296676 133686539519808 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 04:01:54.296871 133654841919040 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0423 04:01:54.296974 133686539519808 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776916912.4873052, 'wait_for_prev_duration_secs': 4.220008850097656e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776916912.4874978, 'checkpointer_blocking_duration_secs': 1.80910325050354, 'get_old_steps_start_time': 1776916914.296617, 'get_old_steps_duration_secs': 2.2411346435546875e-05, 'checkpoint_manager_blocking_start_time': 1776916912.4872227, 'checkpoint_manager_blocking_duration_secs': 1.8097255229949951}
I0423 04:01:54.297102 133686539519808 checkpointing.py:408] Started an asynchronous checkpoint save for step 10
I0423 04:01:54.297140 133686539519808 max_utils.py:750] 
Memstats: After params initialized:
I0423 04:01:54.297189 133686539519808 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0))
I0423 04:01:54.297216 133686539519808 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0423 04:01:54.297237 133686539519808 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0423 04:01:54.297257 133686539519808 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0423 04:01:54.297276 133686539519808 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0423 04:01:54.297295 133686539519808 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0423 04:01:54.297312 133686539519808 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0423 04:01:54.297330 133686539519808 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
I0423 04:01:54.535910 133686539519808 metric_logger.py:196] completed step: 10, seconds: 1.866, TFLOP/s/device: 7.280, Tokens/s/device: 1097.251, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686
I0423 04:01:54.652272 133686539519808 metric_logger.py:196] completed step: 11, seconds: 2.165, TFLOP/s/device: 6.276, Tokens/s/device: 946.015, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211
I0423 04:01:54.771131 133686539519808 metric_logger.py:196] completed step: 12, seconds: 0.006, TFLOP/s/device: 2367.096, Tokens/s/device: 356794.425, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644
I0423 04:01:54.890006 133686539519808 metric_logger.py:196] completed step: 13, seconds: 0.118, TFLOP/s/device: 115.327, Tokens/s/device: 17383.333, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461
I0423 04:01:54.895428 133686539519808 checkpointing.py:772] Waiting for step 15 to finish before checkpoint...
I0423 04:01:55.061339 133654672049728 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10/items/array_metadatas/process_0
I0423 04:01:55.127895 133686539519808 checkpointing.py:776] Waited 0.23244452476501465 seconds for step 15 to finish before starting checkpointing.
I0423 04:01:55.128888 133686539519808 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 04:02:49.325411 133654831433280 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 55.030180s (commit=54.516785s, array_metadata_write=0.513394s)
I0423 04:02:49.326541 133654852404800 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 222.658 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 56 seconds) (per-host)
I0423 04:02:49.326653 133654852404800 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 55.030124s.
I0423 04:02:49.805343 133654852404800 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0423 04:02:50.018222 133654445557312 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776916913324873408, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA
I0423 04:02:50.270230 133654852404800 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0423 04:02:50.270400 133654852404800 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.638963s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10/items
I0423 04:02:50.271173 133654852404800 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10/items
I0423 04:02:50.509554 133654852404800 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10
I0423 04:02:51.246386 133654852404800 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10`.
I0423 04:02:51.247011 133654852404800 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 58.759501s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/10
I0423 04:02:51.247109 133654852404800 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 56.950583s.
I0423 04:02:51.247307 133654841919040 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 04:02:51.247421 133654841919040 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0423 04:02:51.247494 133654841919040 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0423 04:02:51.247540 133654841919040 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0423 04:02:51.247675 133686539519808 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.
W0423 04:02:51.247798 133686539519808 checkpoint_manager.py:1441] Waiting for previous save to complete took 56.118906 seconds. If this number is high, consider checkpointing less frequently.
I0423 04:02:51.248747 133686539519808 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15
I0423 04:02:51.249027 133686539519808 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15.
I0423 04:02:51.418782 133654852404800 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15
I0423 04:02:51.443283 133686539519808 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0423 04:02:51.443376 133686539519808 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
I0423 04:02:52.136843 133654672049728 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15/items
I0423 04:02:52.788916 133686539519808 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.346008s
I0423 04:02:52.792784 133686539519808 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 8.488 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host)
I0423 04:02:52.792848 133686539519808 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.454057s (batch_requests_ready=0.100582s, total_serialization_initiated=1.349721s, others=0.003755s)
I0423 04:02:52.792921 133686539519808 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.454589s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=1.454575)
I0423 04:02:52.794449 133656670635584 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0423 04:02:52.794534 133686539519808 async_checkpointer.py:561] Finished blocking save. Time taken: 1.545742s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15.
I0423 04:02:52.794729 133686539519808 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 04:02:52.794898 133654841919040 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0423 04:02:52.794995 133686539519808 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776916915.1288645, 'wait_for_prev_duration_secs': 56.11890625953674, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776916971.2487726, 'checkpointer_blocking_duration_secs': 1.5458731651306152, 'get_old_steps_start_time': 1776916972.794659, 'get_old_steps_duration_secs': 2.7418136596679688e-05, 'checkpoint_manager_blocking_start_time': 1776916915.1288, 'checkpoint_manager_blocking_duration_secs': 57.66617178916931}
I0423 04:02:52.795130 133686539519808 checkpointing.py:408] Started an asynchronous checkpoint save for step 15
I0423 04:02:52.795773 133686539519808 metric_logger.py:196] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.449, Tokens/s/device: 17250.964, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030
I0423 04:02:52.803769 133686539519808 metric_logger.py:196] completed step: 15, seconds: 0.119, TFLOP/s/device: 114.128, Tokens/s/device: 17202.567, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551
I0423 04:02:52.920325 133686539519808 metric_logger.py:196] completed step: 16, seconds: 57.907, TFLOP/s/device: 0.235, Tokens/s/device: 35.367, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616
I0423 04:02:53.039324 133686539519808 metric_logger.py:196] completed step: 17, seconds: 0.005, TFLOP/s/device: 2478.951, Tokens/s/device: 373654.443, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002
I0423 04:02:53.158757 133686539519808 metric_logger.py:196] completed step: 18, seconds: 0.118, TFLOP/s/device: 115.012, Tokens/s/device: 17335.805, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577
I0423 04:02:53.276540 133686539519808 checkpointing.py:772] Waiting for step 19 to finish before checkpoint...
I0423 04:02:53.277521 133686539519808 checkpointing.py:776] Waited 0.0010063648223876953 seconds for step 19 to finish before starting checkpointing.
I0423 04:02:53.277796 133686539519808 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 04:02:53.572068 133656844699200 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15/items/array_metadatas/process_0
I0423 04:03:21.326011 133654852404800 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 28.532395s (commit=28.044442s, array_metadata_write=0.487953s)
I0423 04:03:21.329517 133656670635584 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 421.357 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 29 seconds) (per-host)
I0423 04:03:21.329572 133656670635584 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 28.534995s.
I0423 04:03:21.853204 133656670635584 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0423 04:03:22.267024 133656670635584 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0423 04:03:22.267268 133656670635584 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.586996s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15/items
I0423 04:03:22.268088 133656670635584 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15/items
I0423 04:03:22.522721 133656670635584 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15
I0423 04:03:23.227697 133656670635584 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15`.
I0423 04:03:23.228348 133656670635584 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 31.979563s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/15
I0423 04:03:23.228422 133656670635584 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 30.433846s.
I0423 04:03:23.228586 133654841919040 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 04:03:23.228691 133654841919040 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0423 04:03:23.228755 133654841919040 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0423 04:03:23.228798 133654841919040 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0423 04:03:23.228945 133686539519808 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.
W0423 04:03:23.229075 133686539519808 checkpoint_manager.py:1441] Waiting for previous save to complete took 29.951287 seconds. If this number is high, consider checkpointing less frequently.
I0423 04:03:23.229518 133686539519808 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19
I0423 04:03:23.229813 133686539519808 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19.
I0423 04:03:23.405621 133656670635584 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19
I0423 04:03:23.410090 133686539519808 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0423 04:03:23.410181 133686539519808 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
I0423 04:03:24.139229 133656844699200 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19/items
I0423 04:03:25.435778 133686539519808 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.026087s
I0423 04:03:25.439723 133686539519808 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.815 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host)
I0423 04:03:25.439800 133686539519808 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.122197s (batch_requests_ready=0.088507s, total_serialization_initiated=2.029847s, others=0.003843s)
I0423 04:03:25.439891 133686539519808 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.122728s (all_items=0.000013s, per_item={'items': '0.00001335'}, temp_paths=2.122715)
I0423 04:03:25.441361 133656670635584 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0423 04:03:25.441442 133686539519808 async_checkpointer.py:561] Finished blocking save. Time taken: 2.211881s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19.
I0423 04:03:25.441618 133686539519808 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 04:03:25.441843 133656458823232 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0423 04:03:25.441953 133686539519808 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776916973.2777703, 'wait_for_prev_duration_secs': 29.951287269592285, 'time_between_consecutive_saves_sec': 2.0302155017852783, 'checkpointer_blocking_start_time': 1776917003.2295432, 'checkpointer_blocking_duration_secs': 2.2119996547698975, 'get_old_steps_start_time': 1776917005.4415576, 'get_old_steps_duration_secs': 2.1696090698242188e-05, 'checkpoint_manager_blocking_start_time': 1776916973.277737, 'checkpoint_manager_blocking_duration_secs': 32.164193868637085}
I0423 04:03:25.442090 133686539519808 checkpointing.py:408] Started an asynchronous checkpoint save for step 19
I0423 04:03:25.442126 133686539519808 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 04:03:26.224229 133656660149824 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19/items/array_metadatas/process_0
I0423 04:04:09.502822 133656844699200 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 44.062355s (commit=43.549576s, array_metadata_write=0.512779s)
I0423 04:04:09.508347 133656670635584 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 273.578 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 46 seconds) (per-host)
I0423 04:04:09.508402 133656670635584 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.066918s.
I0423 04:04:10.007595 133656670635584 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0423 04:04:10.386424 133656670635584 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0423 04:04:10.386638 133656670635584 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.553191s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19/items
I0423 04:04:10.387320 133656670635584 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19/items
I0423 04:04:10.635396 133656670635584 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19
I0423 04:04:11.344039 133656670635584 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19`.
I0423 04:04:11.344668 133656670635584 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 48.115114s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_true/checkpoints/19
I0423 04:04:11.344736 133656670635584 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 45.903253s.
I0423 04:04:11.344916 133656458823232 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 04:04:11.345022 133656458823232 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0423 04:04:11.345086 133656458823232 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0423 04:04:11.345127 133656458823232 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0423 04:04:11.345243 133686539519808 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.
I0423 04:04:11.345367 133686539519808 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 04:04:11.346104 133686539519808 metric_logger.py:196] completed step: 19, seconds: 0.119, TFLOP/s/device: 114.024, Tokens/s/device: 17186.975, total_weights: 16384, loss: 1.449, lm_loss: 1.449, perplexity: 4.260
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops