MaxView

XPK Start: Wed Apr 22 18:09:44 UTC 2026
2026-04-22 18:09:48.676446: 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 00:00:1776881388.689517      10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1776881388.693326      10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
W0000 00:00:1776881388.704522      10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776881388.704541      10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776881388.704543      10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776881388.704546      10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
2026-04-22 18:10:07.847242: 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)
I0422 18:10:08.366782 135706623592256 max_utils.py:273] Attempting to initialize the jax distributed system...
INFO:2026-04-22 18:10:17,407:jax._src.distributed:140: Starting JAX distributed service on [::]:8482
I0422 18:10:17.407266 135706623592256 distributed.py:140] Starting JAX distributed service on [::]:8482
INFO:2026-04-22 18:10:17,409:jax._src.distributed:157: Connecting to JAX distributed service on mt-13-scan-layers-false-hckxg-slice-job-0-0.mt-13-scan-layers-false-hckxg:8482
I0422 18:10:17.409446 135706623592256 distributed.py:157] Connecting to JAX distributed service on mt-13-scan-layers-false-hckxg-slice-job-0-0.mt-13-scan-layers-false-hckxg:8482
I0422 18:10:18.552544 135706623592256 max_utils.py:284] Jax distributed system initialized!
I0422 18:10:25.411123 135706623592256 max_utils.py:800] System Information: Jax Version: 0.8.1
I0422 18:10:25.411230 135706623592256 max_utils.py:801] System Information: Jaxlib Version: 0.8.1
I0422 18:10:25.411272 135706623592256 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
I0422 18:10:25.411307 135706623592256 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0422 18:10:26.029325 135706623592256 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0422 18:10:27.198670 135706623592256 checkpointing.py:688] Setting up checkpoint logger...
I0422 18:10:27.198795 135706623592256 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0422 18:10:27.198842 135706623592256 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None
I0422 18:10:27.199087 135706623592256 base_pytree_checkpoint_handler.py:415] 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 0x7b6c2c374fb0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0422 18:10:30.299353 135706623592256 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0422 18:10:30.299591 135706623592256 checkpoint_manager.py:709] [process=5][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b58ed2ae390>}, handler_registry=None
I0422 18:10:30.299853 135706623592256 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b58ed2ae390>` 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`.
I0422 18:10:30.299904 135706623592256 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b58ed2853a0>` 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`.
I0422 18:10:30.299943 135706623592256 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 0x7b58ed2ae390>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b58ed2ae390>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b58ed2853a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b58ed2853a0>}).
I0422 18:10:30.300291 135706623592256 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33
I0422 18:10:30.300369 135706623592256 async_checkpointer.py:177] [process=5][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b58ee2c00e0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0422 18:10:31.835974 135706623592256 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints
I0422 18:10:32.011222 135706623592256 checkpoint_manager.py:929] [process=5][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_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=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b58ed2857c0>
I0422 18:10:32.011394 135706623592256 checkpointing.py:302] Checkpoint manager created!
I0422 18:10:34.473038 135706623592256 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0422 18:10:34.473167 135706623592256 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0422 18:10:44.569762 135706623592256 maxtext_utils.py:1749]  decoder/decoder_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.569886 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.569941 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.569985 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.570023 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.570060 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.570100 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570137 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.570174 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570210 135706623592256 maxtext_utils.py:1749]  decoder/layers_0/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570249 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.570300 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.570353 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.570389 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.570423 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.570458 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570493 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.570527 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570560 135706623592256 maxtext_utils.py:1749]  decoder/layers_1/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570593 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.570625 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.570668 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.570701 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.570732 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.570764 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570796 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.570830 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570865 135706623592256 maxtext_utils.py:1749]  decoder/layers_10/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.570899 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.570934 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.570968 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.570999 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.571030 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.571064 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571096 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.571128 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571161 135706623592256 maxtext_utils.py:1749]  decoder/layers_11/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571193 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.571225 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.571261 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.571291 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.571320 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.571352 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571383 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.571413 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571444 135706623592256 maxtext_utils.py:1749]  decoder/layers_12/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571475 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.571505 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.571535 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.571562 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.571590 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.571621 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571660 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.571693 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571724 135706623592256 maxtext_utils.py:1749]  decoder/layers_13/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571754 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.571785 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.571819 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.571847 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.571875 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.571905 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571934 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.571964 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.571994 135706623592256 maxtext_utils.py:1749]  decoder/layers_14/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572024 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.572053 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.572082 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.572110 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.572138 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.572167 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572197 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.572227 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572261 135706623592256 maxtext_utils.py:1749]  decoder/layers_15/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572291 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.572323 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.572355 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.572383 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.572410 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.572441 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572471 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.572501 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572530 135706623592256 maxtext_utils.py:1749]  decoder/layers_2/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572560 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.572590 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.572619 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.572659 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.572690 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.572720 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572750 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.572780 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572809 135706623592256 maxtext_utils.py:1749]  decoder/layers_3/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.572839 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.572869 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.572899 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.572927 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.572954 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.572986 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573016 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.573045 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573075 135706623592256 maxtext_utils.py:1749]  decoder/layers_4/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573105 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.573134 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.573163 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.573191 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.573218 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.573254 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573285 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.573315 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573346 135706623592256 maxtext_utils.py:1749]  decoder/layers_5/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573375 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.573405 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.573434 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.573462 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.573489 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.573518 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573548 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.573578 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573608 135706623592256 maxtext_utils.py:1749]  decoder/layers_6/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573637 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.573680 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.573711 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.573739 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.573767 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.573797 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573827 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.573856 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573885 135706623592256 maxtext_utils.py:1749]  decoder/layers_7/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.573915 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.573945 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.573975 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.574002 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.574029 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.574060 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.574090 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.574120 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.574149 135706623592256 maxtext_utils.py:1749]  decoder/layers_8/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.574179 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/mlp/wi_0/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.574209 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/mlp/wi_1/kernel/value
    Shape:     float32[2048,7168]
    Physical:  ('fsdp', None)
I0422 18:10:44.574238 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/mlp/wo/kernel/value
    Shape:     float32[7168,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:44.574270 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.574297 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0422 18:10:44.574327 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/self_attention/key/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.574356 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/self_attention/out/kernel/value
    Shape:     float32[16,128,2048]
    Physical:  (None, None, 'fsdp')
I0422 18:10:44.574386 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/self_attention/query/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.574415 135706623592256 maxtext_utils.py:1749]  decoder/layers_9/self_attention/value/kernel/value
    Shape:     float32[2048,16,128]
    Physical:  ('fsdp', None, None)
I0422 18:10:44.574445 135706623592256 maxtext_utils.py:1749]  decoder/logits_dense/kernel/value
    Shape:     float32[2048,32000]
    Physical:  ('fsdp', None)

I0422 18:10:44.574475 135706623592256 maxtext_utils.py:1749]  token_embedder/embedding/value
    Shape:     float32[32000,2048]
    Physical:  (None, 'fsdp')
I0422 18:10:45.453980 135706623592256 nnx_decoders.py:770] nnx_decoders/y_in Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0422 18:10:45.454080 135706623592256 nnx_decoders.py:770] nnx_decoders/y_in Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 18:10:45.459314 135706623592256 nnx_decoders.py:770] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0422 18:10:45.459369 135706623592256 nnx_decoders.py:770] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 18:10:45.475117 135706623592256 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0422 18:10:45.475180 135706623592256 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 18:10:45.490439 135706623592256 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0422 18:10:45.490493 135706623592256 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 18:10:45.514802 135706623592256 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 18:10:45.514872 135706623592256 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 18:10:45.530399 135706623592256 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 18:10:45.530461 135706623592256 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 18:10:45.545662 135706623592256 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 18:10:45.545728 135706623592256 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 18:10:45.576821 135706623592256 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0422 18:10:45.576890 135706623592256 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 18:10:45.596255 135706623592256 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0422 18:10:45.596316 135706623592256 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0422 18:11:46.152111 135706623592256 max_utils.py:791] Total memory size: 1.8 GB, Output size: 0.4 GB, Temp size: 1.4 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0422 18:11:46.164174 135706623592256 metric_logger.py:289] number parameters: 1.104 billion
I0422 18:12:49.202872 135706623592256 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0422 18:12:49.474838 135706623592256 checkpointing.py:798] Waited 0.27184367179870605 seconds for step 0 to finish before starting checkpointing.
I0422 18:12:49.478928 135706623592256 checkpoint_manager.py:2013] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 18:12:49.480589 135706623592256 checkpoint_manager.py:1518] [process=5] Saving checkpoint at step 0
I0422 18:12:49.483370 135706623592256 async_checkpointer.py:452] [process=5] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints/0.
I0422 18:12:50.455753 135706623592256 signaling_client.py:364] Using JaxDistributedSignalingClient
I0422 18:12:50.456995 135706623592256 jax_array_handlers.py:358] Scheduling D2H of 648 prioritized jax.Array.
I0422 18:12:50.457114 135706623592256 replica_slices.py:424] 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
I0422 18:12:50.767826 135706623592256 base_pytree_checkpoint_handler.py:153] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.318938s
I0422 18:12:50.767993 135706623592256 base_pytree_checkpoint_handler.py:129] [process=5] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.292 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.6729526519775391 s) (per-host)
I0422 18:12:50.768043 135706623592256 base_pytree_checkpoint_handler.py:737] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.673012s (batch_requests_ready=0.318706s, total_serialization_initiated=0.354239s, others=0.000067s)
I0422 18:12:50.768409 135706623592256 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.677434s (all_items=0.000021s, per_item={'items': '0.00002146'}, temp_paths=0.677413)
I0422 18:12:50.769377 135584425158400 async_checkpointer.py:79] [process=5][thread=async_save] Background save thread started.
I0422 18:12:50.769537 135706623592256 async_checkpointer.py:561] Finished blocking save. Time taken: 1.288878s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints/0.
I0422 18:12:51.190012 135706623592256 checkpoint_manager.py:1566] [process=5][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 18:12:51.190401 135584964130560 async_checkpointer.py:265] [process=5][thread=save_finalize] Waiting for background save thread=async_save.
I0422 18:12:51.190569 135706623592256 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776881569.4789095, 'wait_for_prev_duration_secs': 7.271766662597656e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776881569.480632, 'checkpointer_blocking_duration_secs': 1.2890076637268066, 'get_old_steps_start_time': 1776881570.769675, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776881569.4769373, 'checkpoint_manager_blocking_duration_secs': 1.713590145111084}
I0422 18:12:51.190821 135706623592256 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0422 18:12:51.191059 135706623592256 max_utils.py:750] 
Memstats: After params initialized:
I0422 18:12:51.191121 135706623592256 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_18(process=5,(2,4,0,0))
I0422 18:12:51.191155 135706623592256 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_19(process=5,(3,4,0,0))
I0422 18:12:51.191183 135706623592256 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_22(process=5,(2,5,0,0))
I0422 18:12:51.191213 135706623592256 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_23(process=5,(3,5,0,0))
I0422 18:12:51.520863 135706623592256 metric_logger.py:185] completed step: 0, seconds: 63.031, TFLOP/s/device: 0.216, Tokens/s/device: 32.492, total_weights: 65536, loss: 10.874
I0422 18:12:51.642198 135706623592256 metric_logger.py:185] completed step: 1, seconds: 2.303, TFLOP/s/device: 5.900, Tokens/s/device: 889.285, total_weights: 65536, loss: 10.874
I0422 18:12:52.180436 135706623592256 metric_logger.py:185] completed step: 2, seconds: 0.060, TFLOP/s/device: 226.226, Tokens/s/device: 34099.234, total_weights: 65536, loss: 10.265
I0422 18:12:52.316720 135706623592256 metric_logger.py:185] completed step: 3, seconds: 0.418, TFLOP/s/device: 32.529, Tokens/s/device: 4903.146, total_weights: 65536, loss: 9.736
I0422 18:12:52.594278 135706623592256 metric_logger.py:185] completed step: 4, seconds: 0.216, TFLOP/s/device: 62.918, Tokens/s/device: 9483.721, total_weights: 65536, loss: 9.279
I0422 18:12:52.620905 135706623592256 metric_logger.py:185] completed step: 5, seconds: 0.136, TFLOP/s/device: 99.794, Tokens/s/device: 15042.012, total_weights: 65536, loss: 8.894
I0422 18:13:14.736184 135706623592256 metric_logger.py:185] completed step: 6, seconds: 0.279, TFLOP/s/device: 48.622, Tokens/s/device: 7328.891, total_weights: 65536, loss: 8.595
I0422 18:13:14.872581 135706623592256 metric_logger.py:185] completed step: 7, seconds: 21.994, TFLOP/s/device: 0.618, Tokens/s/device: 93.114, total_weights: 65536, loss: 8.387
I0422 18:13:15.009002 135706623592256 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.278, Tokens/s/device: 14059.947, total_weights: 65536, loss: 8.258
I0422 18:13:15.153141 135706623592256 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0422 18:13:15.157832 135706623592256 checkpointing.py:798] Waited 0.0046977996826171875 seconds for step 9 to finish before starting checkpointing.
I0422 18:13:15.160712 135706623592256 checkpoint_manager.py:2024] [process=5][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 18:13:15.769843    2798 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0422 18:13:18.456365 135584433551104 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 648 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints/0/items/array_metadatas/process_5
I0422 18:13:46.053508 135584425158400 base_pytree_checkpoint_handler.py:129] [process=5] /jax/checkpoint/write/gbytes_per_sec: 28.228 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 55.958417654037476 s) (per-host)
I0422 18:13:46.053618 135584425158400 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 55.284130s.
I0422 18:13:56.094174 135584425158400 async_checkpointer.py:144] [process=5][thread=async_save] Background save thread done. Time taken: 65.324667s.
I0422 18:13:56.094476 135584964130560 async_checkpointer.py:273] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 18:13:56.094593 135584964130560 async_checkpointer.py:283] [process=5][thread=save_finalize] No errors found in background save thread=async_save.
I0422 18:13:56.094643 135584964130560 checkpoint_manager.py:2133] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0422 18:13:56.095959 135584964130560 checkpoint_manager.py:2142] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0422 18:13:56.096123 135706623592256 checkpoint_manager.py:2036] [process=5][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0422 18:13:56.096283 135706623592256 checkpoint_manager.py:1458] Waiting for previous save to complete took 40.935573 seconds. If this number is high, consider checkpointing less frequently.
I0422 18:13:56.100247 135706623592256 checkpoint_manager.py:1518] [process=5] Saving checkpoint at step 9
I0422 18:13:56.103745 135706623592256 async_checkpointer.py:452] [process=5] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints/9.
I0422 18:13:56.744424 135706623592256 jax_array_handlers.py:358] Scheduling D2H of 648 prioritized jax.Array.
I0422 18:13:56.744559 135706623592256 replica_slices.py:424] 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
I0422 18:13:56.912748 135706623592256 base_pytree_checkpoint_handler.py:153] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.176855s
I0422 18:13:56.912912 135706623592256 base_pytree_checkpoint_handler.py:129] [process=5] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.031 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5088534355163574 s) (per-host)
I0422 18:13:56.912969 135706623592256 base_pytree_checkpoint_handler.py:737] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.508919s (batch_requests_ready=0.297881s, total_serialization_initiated=0.210965s, others=0.000072s)
I0422 18:13:56.913346 135706623592256 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.513210s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.513193)
I0422 18:13:56.914263 135581782767360 async_checkpointer.py:79] [process=5][thread=async_save] Background save thread started.
I0422 18:13:56.914429 135706623592256 async_checkpointer.py:561] Finished blocking save. Time taken: 0.814112s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints/9.
I0422 18:13:57.347414 135706623592256 checkpoint_manager.py:1566] [process=5][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 18:13:57.347796 135584964130560 async_checkpointer.py:265] [process=5][thread=save_finalize] Waiting for background save thread=async_save.
I0422 18:13:57.347957 135706623592256 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776881595.160684, 'wait_for_prev_duration_secs': 40.93557262420654, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776881636.1002865, 'checkpointer_blocking_duration_secs': 0.81424880027771, 'get_old_steps_start_time': 1776881636.914554, 'get_old_steps_duration_secs': 2.4557113647460938e-05, 'checkpoint_manager_blocking_start_time': 1776881595.159036, 'checkpoint_manager_blocking_duration_secs': 42.188884258270264}
I0422 18:13:57.348245 135706623592256 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0422 18:13:57.348296 135706623592256 checkpoint_manager.py:2024] [process=5][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 18:14:02.443366 135584433551104 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 648 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_13_scan_layers_false/checkpoints/9/items/array_metadatas/process_5
I0422 18:14:40.120754 135581782767360 base_pytree_checkpoint_handler.py:129] [process=5] /jax/checkpoint/write/gbytes_per_sec: 36.133 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.71664810180664 s) (per-host)
I0422 18:14:40.120872 135581782767360 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.206499s.
I0422 18:14:48.885951 135581782767360 async_checkpointer.py:144] [process=5][thread=async_save] Background save thread done. Time taken: 51.971562s.
I0422 18:14:48.886183 135584964130560 async_checkpointer.py:273] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 18:14:48.886263 135584964130560 async_checkpointer.py:283] [process=5][thread=save_finalize] No errors found in background save thread=async_save.
I0422 18:14:48.886315 135584964130560 checkpoint_manager.py:2133] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0422 18:14:48.888151 135584964130560 checkpoint_manager.py:2142] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0422 18:14:48.888333 135706623592256 checkpoint_manager.py:2036] [process=5][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0422 18:14:48.888686 135706623592256 checkpoint_manager.py:2013] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 18:14:48.889374 135706623592256 metric_logger.py:185] completed step: 9, seconds: 0.137, TFLOP/s/device: 99.196, Tokens/s/device: 14951.961, total_weights: 65536, loss: 8.182
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Wed Apr 22 18:14:59 UTC 2026
EXIT_CODE=0