XPK Start: Fri Apr 17 13:41:07 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-17 13:41:31.245773: 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)
I0417 13:41:31.431360 139563380156224 max_utils.py:273] Attempting to initialize the jax distributed system...
I0417 13:41:40.471627 139563380156224 distributed.py:149] Starting JAX distributed service on [::]:8482
I0417 13:41:40.473932 139563380156224 distributed.py:172] Connecting to JAX distributed service on mt-08-checkpoint-async--23lo4-slice-job-0-0.mt-08-checkpoint-async--23lo4:8482
I0417 13:41:41.975427 139563380156224 max_utils.py:284] Jax distributed system initialized!
I0417 13:41:47.926304 139563380156224 max_utils.py:800] System Information: Jax Version: 0.9.2
I0417 13:41:47.926409 139563380156224 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0417 13:41:47.926451 139563380156224 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365
I0417 13:41:47.926487 139563380156224 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0417 13:41:48.973000 139563380156224 maxtext_utils.py:1548] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0417 13:41:48.973270 139563380156224 checkpointing.py:677] Setting up checkpoint logger...
I0417 13:41:48.973321 139563380156224 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0417 13:41:48.973365 139563380156224 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0417 13:41:48.973692 139563380156224 base_pytree_checkpoint_handler.py:441] 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 0x7eee1a58bc50>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0417 13:41:50.727131 139563380156224 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0417 13:41:50.727366 139563380156224 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eecddb91190>}, handler_registry=None
I0417 13:41:50.727603 139563380156224 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eecddb91190>` 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`.
I0417 13:41:50.727651 139563380156224 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eecddbb74a0>` 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`.
I0417 13:41:50.727688 139563380156224 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 0x7eecddb91190>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eecddb91190>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eecddbb74a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eecddbb74a0>}).
I0417 13:41:50.728031 139563380156224 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0417 13:41:50.728104 139563380156224 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ed8b0443ec0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0417 13:41:56.530233 139441859446528 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433005626338797, 'commit_timestamp_nsecs': 1776433056611638814, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA
I0417 13:41:56.775298 139441859446528 checkpoint_manager.py:1871] Missing metrics for step 0
I0417 13:41:57.913606 139563380156224 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints
I0417 13:41:58.354677 139563380156224 checkpoint_manager.py:929] [process=6][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=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, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7eecddba6a20>
I0417 13:41:58.354865 139563380156224 checkpointing.py:301] Checkpoint manager created!
I0417 13:41:59.291125 139563380156224 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0417 13:41:59.291234 139563380156224 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:41:59.674641 139563380156224 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 13:41:59.674725 139563380156224 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:41:59.691420 139563380156224 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 13:41:59.691484 139563380156224 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:41:59.715418 139563380156224 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:41:59.715493 139563380156224 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:41:59.732057 139563380156224 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:41:59.732127 139563380156224 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:41:59.748547 139563380156224 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:41:59.748612 139563380156224 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:41:59.773397 139563380156224 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0417 13:41:59.773470 139563380156224 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:41:59.795587 139563380156224 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0417 13:41:59.795657 139563380156224 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0417 13:42:00.016122 139563380156224 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0417 13:42:00.016230 139563380156224 checkpointing.py:581] restoring from this run's directory step 9
I0417 13:42:00.790311 139563380156224 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/9.
I0417 13:42:02.229642 139563380156224 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/9.
I0417 13:42:02.230085 139563380156224 event_tracking.py:125] [process=6] [sync] Finished blocking load in 1.44 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/9.
I0417 13:42:04.975327 1465 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0417 13:42:11.533927 139563380156224 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 240.729 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.564528465270996 s) (per-host)
I0417 13:42:11.538579 139563380156224 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 211.607 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.47894287109375 s) (per-host)
I0417 13:42:13.836699 139563380156224 event_tracking.py:138] [process=6] [sync] Finished load in 13.05 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/9
I0417 13:42:13.836909 139563380156224 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1776433320.7902145, 'checkpointer_duration_secs': 13.046662330627441, 'checkpoint_manager_start_time': 1776433320.016468, 'checkpoint_manager_duration_secs': 13.820411205291748}
fsdp: 32
I0417 13:42:14.257132 139563380156224 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0417 13:42:14.257257 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0417 13:42:14.257311 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0417 13:42:14.257374 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0417 13:42:14.257427 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0417 13:42:14.257465 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0417 13:42:14.257517 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/key/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0417 13:42:14.257581 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/out/kernel
Shape: float32[16,16,128,2048]
Logical: P('heads', 'layers', 'kv', 'embed')
Physical: (None, None, None, 'fsdp')
I0417 13:42:14.257621 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/query/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'q_heads', 'kv')
Physical: ('fsdp', None, None, None)
I0417 13:42:14.257657 139563380156224 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/value/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0417 13:42:14.257703 139563380156224 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0417 13:42:14.257748 139563380156224 maxtext_utils.py:1651] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0417 13:42:14.743349 139563380156224 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 13:42:14.743444 139563380156224 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 13:42:14.759695 139563380156224 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 13:42:14.759759 139563380156224 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 13:42:25.651126 139563380156224 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0417 13:42:25.651967 139563380156224 metric_logger.py:301] number parameters: 1.104 billion
I0417 13:42:36.911904 139563380156224 checkpointing.py:772] Waiting for step 10 to finish before checkpoint...
I0417 13:42:37.064163 139563380156224 checkpointing.py:776] Waited 0.15223383903503418 seconds for step 10 to finish before starting checkpointing.
I0417 13:42:37.066540 139563380156224 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 13:42:37.068120 139563380156224 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10
I0417 13:42:37.071230 139563380156224 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/10.
I0417 13:42:37.894287 139563380156224 signaling_client.py:364] Using JaxDistributedSignalingClient
I0417 13:42:37.895313 139563380156224 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0417 13:42:37.895369 139563380156224 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
I0417 13:42:38.170908 139563380156224 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.276694s
I0417 13:42:38.171069 139563380156224 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.499 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2805347442626953 s) (per-host)
I0417 13:42:38.171122 139563380156224 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.280599s (batch_requests_ready=0.002160s, total_serialization_initiated=0.278367s, others=0.000072s)
I0417 13:42:38.171212 139563380156224 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.284705s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.284688)
I0417 13:42:38.172023 139563380156224 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.10 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/10.
I0417 13:42:38.172296 139438285891328 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 14:02:38.172270
I0417 13:42:38.178883 139563380156224 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 13:42:38.179153 139441070925568 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0417 13:42:38.179302 139563380156224 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433357.0665221, 'wait_for_prev_duration_secs': 5.936622619628906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776433357.0681572, 'checkpointer_blocking_duration_secs': 1.1042487621307373, 'get_old_steps_start_time': 1776433358.1724286, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776433357.0646799, 'checkpoint_manager_blocking_duration_secs': 1.114586591720581}
I0417 13:42:38.179409 139563380156224 checkpointing.py:408] Started an asynchronous checkpoint save for step 10
I0417 13:42:38.179457 139563380156224 max_utils.py:750]
Memstats: After params initialized:
I0417 13:42:38.179506 139563380156224 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0))
I0417 13:42:38.179539 139563380156224 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0))
I0417 13:42:38.179566 139563380156224 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0))
I0417 13:42:38.179591 139563380156224 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0))
I0417 13:42:38.493832 139563380156224 metric_logger.py:196] completed step: 10, seconds: 11.260, TFLOP/s/device: 1.207, Tokens/s/device: 181.886, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869
I0417 13:42:38.668314 139563380156224 metric_logger.py:196] completed step: 11, seconds: 1.580, TFLOP/s/device: 8.598, Tokens/s/device: 1295.911, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659
I0417 13:42:38.815712 139563380156224 metric_logger.py:196] completed step: 12, seconds: 0.027, TFLOP/s/device: 496.352, Tokens/s/device: 74815.518, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068
I0417 13:42:38.963052 139563380156224 metric_logger.py:196] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.724, Tokens/s/device: 13373.471, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001
I0417 13:42:38.967599 139563380156224 checkpointing.py:772] Waiting for step 15 to finish before checkpoint...
I0417 13:42:39.257176 139563380156224 checkpointing.py:776] Waited 0.28954577445983887 seconds for step 15 to finish before starting checkpointing.
I0417 13:42:39.260156 139563380156224 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 13:42:49.544167 139438294284032 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/10/items/array_metadatas/process_6
I0417 13:43:23.695962 139438285891328 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 34.485 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.80538821220398 s) (per-host)
I0417 13:43:23.696085 139438285891328 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.523708s.
I0417 13:43:35.411041 139438285891328 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 57.238647s.
I0417 13:43:35.411307 139441070925568 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 13:43:35.411430 139441070925568 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0417 13:43:35.411477 139441070925568 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0417 13:43:35.416238 139441070925568 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0417 13:43:35.416434 139563380156224 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10.
W0417 13:43:35.416575 139563380156224 checkpoint_manager.py:1452] Waiting for previous save to complete took 56.156422 seconds. If this number is high, consider checkpointing less frequently.
I0417 13:43:35.418806 139563380156224 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15
I0417 13:43:35.421900 139563380156224 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/15.
I0417 13:43:36.554716 139563380156224 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0417 13:43:36.554811 139563380156224 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
I0417 13:43:36.580319 139563380156224 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026721s
I0417 13:43:36.580447 139563380156224 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 51.259 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030093908309936523 s) (per-host)
I0417 13:43:36.580496 139563380156224 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030150s (batch_requests_ready=0.001773s, total_serialization_initiated=0.028315s, others=0.000062s)
I0417 13:43:36.580574 139563380156224 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035058s (all_items=0.000014s, per_item={'items': '0.00001359'}, temp_paths=0.035045)
I0417 13:43:36.581269 139563380156224 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.16 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/15.
I0417 13:43:36.581582 139435626714880 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 14:03:36.581545
I0417 13:43:36.594892 139563380156224 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 13:43:36.595203 139441070925568 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0417 13:43:36.595368 139563380156224 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433359.2601254, 'wait_for_prev_duration_secs': 56.15642166137695, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776433415.4188595, 'checkpointer_blocking_duration_secs': 1.1628637313842773, 'get_old_steps_start_time': 1776433416.5817475, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776433359.2577841, 'checkpoint_manager_blocking_duration_secs': 57.3375506401062}
I0417 13:43:36.595481 139563380156224 checkpointing.py:408] Started an asynchronous checkpoint save for step 15
I0417 13:43:36.596368 139563380156224 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.262, Tokens/s/device: 13906.808, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193
I0417 13:43:36.602463 139563380156224 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.140, Tokens/s/device: 13888.324, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661
I0417 13:43:45.207943 139438294284032 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/15/items/array_metadatas/process_6
I0417 13:43:59.764008 139563380156224 metric_logger.py:196] completed step: 16, seconds: 57.633, TFLOP/s/device: 0.236, Tokens/s/device: 35.535, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872
I0417 13:43:59.911283 139563380156224 metric_logger.py:196] completed step: 17, seconds: 23.016, TFLOP/s/device: 0.590, Tokens/s/device: 88.982, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975
I0417 13:44:00.058498 139563380156224 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.363, Tokens/s/device: 13469.696, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333
I0417 13:44:00.205294 139563380156224 checkpointing.py:772] Waiting for step 19 to finish before checkpoint...
I0417 13:44:00.205981 139563380156224 checkpointing.py:776] Waited 0.0007061958312988281 seconds for step 19 to finish before starting checkpointing.
I0417 13:44:00.208148 139563380156224 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 13:44:19.802111 139435626714880 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.521 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.25173091888428 s) (per-host)
I0417 13:44:19.802212 139435626714880 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.220515s.
I0417 13:44:29.129872 139435626714880 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 52.548159s.
I0417 13:44:29.130143 139441070925568 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 13:44:29.130283 139441070925568 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0417 13:44:29.130331 139441070925568 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0417 13:44:29.132329 139441070925568 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0417 13:44:29.132529 139563380156224 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15.
W0417 13:44:29.132666 139563380156224 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.924524 seconds. If this number is high, consider checkpointing less frequently.
I0417 13:44:29.134047 139563380156224 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19
I0417 13:44:29.136158 139563380156224 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/19.
I0417 13:44:29.840591 139563380156224 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0417 13:44:29.840685 139563380156224 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
I0417 13:44:29.866428 139563380156224 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026686s
I0417 13:44:29.866566 139563380156224 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 51.318 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030058860778808594 s) (per-host)
I0417 13:44:29.866626 139563380156224 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030128s (batch_requests_ready=0.001642s, total_serialization_initiated=0.028413s, others=0.000074s)
I0417 13:44:29.866725 139563380156224 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034224s (all_items=0.000010s, per_item={'items': '0.00000954'}, temp_paths=0.034214)
I0417 13:44:29.867443 139563380156224 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.73 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/19.
I0417 13:44:29.867727 139441070925568 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 14:04:29.867695
I0417 13:44:29.872108 139563380156224 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 13:44:29.872385 139438294284032 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0417 13:44:29.872540 139563380156224 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433440.2081187, 'wait_for_prev_duration_secs': 28.924524307250977, 'time_between_consecutive_saves_sec': 24.79183793067932, 'checkpointer_blocking_start_time': 1776433469.1340866, 'checkpointer_blocking_duration_secs': 0.7337954044342041, 'get_old_steps_start_time': 1776433469.867899, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776433440.2062104, 'checkpoint_manager_blocking_duration_secs': 29.666297435760498}
I0417 13:44:29.872644 139563380156224 checkpointing.py:408] Started an asynchronous checkpoint save for step 19
I0417 13:44:29.872686 139563380156224 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 13:44:37.727257 139437218219776 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_08_checkpoint_async_true/checkpoints/19/items/array_metadatas/process_6
I0417 13:45:13.528699 139441070925568 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.153 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.69215679168701 s) (per-host)
I0417 13:45:13.528842 139441070925568 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.660977s.
I0417 13:45:27.283663 139441070925568 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 57.415801s.
I0417 13:45:27.283921 139438294284032 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 13:45:27.284042 139438294284032 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0417 13:45:27.284084 139438294284032 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0417 13:45:27.289346 139438294284032 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0417 13:45:27.289502 139563380156224 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19.
I0417 13:45:27.289648 139563380156224 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 13:45:27.290682 139563380156224 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.405, Tokens/s/device: 13928.278, total_weights: 65536, loss: 5.391, lm_loss: 5.391, perplexity: 219.349
Per train step:
Total TFLOPs: 13.59
split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Fri Apr 17 13:45:38 UTC 2026
EXIT_CODE=0