XPK Start: Tue Apr 21 02:17:26 UTC 2026 2026-04-21 02:17:30.926671: 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 00:00:1776737850.939510 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776737850.943435 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776737850.954912 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776737850.954930 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776737850.954932 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776737850.954935 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 02:17:50.264013: 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) I0421 02:17:50.781059 140707638953792 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-21 02:17:59,821:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0421 02:17:59.821453 140707638953792 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-21 02:17:59,823:jax._src.distributed:157: Connecting to JAX distributed service on mt-16-zarr3-true-rs-jppcx-slice-job-0-0.mt-16-zarr3-true-rs-jppcx:8482 I0421 02:17:59.823764 140707638953792 distributed.py:157] Connecting to JAX distributed service on mt-16-zarr3-true-rs-jppcx-slice-job-0-0.mt-16-zarr3-true-rs-jppcx:8482 I0421 02:18:01.140863 140707638953792 max_utils.py:284] Jax distributed system initialized! I0421 02:18:08.026577 140707638953792 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 02:18:08.026703 140707638953792 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 02:18:08.026746 140707638953792 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 I0421 02:18:08.026782 140707638953792 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 02:18:08.647927 140707638953792 maxtext_utils.py:1398] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 02:18:08.648210 140707638953792 checkpointing.py:677] Setting up checkpoint logger... I0421 02:18:08.648271 140707638953792 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 02:18:08.648316 140707638953792 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0421 02:18:08.648684 140707638953792 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 0x7ff888775940>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 02:18:09.886682 140707638953792 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 02:18:09.886918 140707638953792 checkpoint_manager.py:709] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fe612551430>}, handler_registry=None I0421 02:18:09.887149 140707638953792 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fe612551430>` 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`. I0421 02:18:09.887198 140707638953792 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe6136a7ef0>` 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`. I0421 02:18:09.887236 140707638953792 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 0x7fe612551430>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fe612551430>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe6136a7ef0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe6136a7ef0>}). I0421 02:18:09.889928 140707638953792 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0421 02:18:09.890012 140707638953792 async_checkpointer.py:177] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ff8763fb2e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 02:18:11.903387 140591744653056 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776737726948902873, 'commit_timestamp_nsecs': 1776737773939021293, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/9/_CHECKPOINT_METADATA I0421 02:18:12.150735 140591744653056 checkpoint_manager.py:1877] Missing metrics for step 9 I0421 02:18:12.550538 140707638953792 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints I0421 02:18:13.736736 140707638953792 checkpoint_manager.py:929] [process=7][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_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7fe6846c5760> I0421 02:18:13.736904 140707638953792 checkpointing.py:301] Checkpoint manager created! I0421 02:18:14.630313 140707638953792 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 02:18:14.630420 140707638953792 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 02:18:15.001684 140707638953792 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 02:18:15.001776 140707638953792 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 02:18:15.017496 140707638953792 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 02:18:15.017559 140707638953792 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 02:18:15.040906 140707638953792 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 02:18:15.040975 140707638953792 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 02:18:15.056847 140707638953792 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 02:18:15.056907 140707638953792 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 02:18:15.072650 140707638953792 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 02:18:15.072722 140707638953792 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 02:18:15.096530 140707638953792 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 02:18:15.096604 140707638953792 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 02:18:15.116096 140707638953792 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 02:18:15.116161 140707638953792 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0421 02:18:15.305779 140707638953792 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 02:18:15.305894 140707638953792 checkpointing.py:581] restoring from this run's directory step 9 I0421 02:18:15.811438 140707638953792 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/9. I0421 02:18:17.638754 1408 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0421 02:18:23.869399 140707638953792 jax_array_handlers.py:837] [process=7] /jax/orbax/read/worker/io/requested throughput: 253.421 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.235767602920532 s) (per-host) I0421 02:18:23.883560 140707638953792 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/read/gbytes_per_sec: 237.930 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.651496887207031 s) (per-host) I0421 02:18:24.807637 140707638953792 checkpointer.py:318] Finished restoring checkpoint in 9.24 seconds from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/9. I0421 02:18:24.807841 140707638953792 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1776737895.56378, 'checkpointer_duration_secs': 9.244030237197876, 'checkpoint_manager_start_time': 1776737895.3061352, 'checkpoint_manager_duration_secs': 9.50167727470398} fsdp: 32 I0421 02:18:25.207159 140707638953792 maxtext_utils.py:1501] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 02:18:25.207295 140707638953792 maxtext_utils.py:1501] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 02:18:25.207349 140707638953792 maxtext_utils.py:1501] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 02:18:25.207407 140707638953792 maxtext_utils.py:1501] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0421 02:18:25.207461 140707638953792 maxtext_utils.py:1501] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0421 02:18:25.207501 140707638953792 maxtext_utils.py:1501] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0421 02:18:25.207555 140707638953792 maxtext_utils.py:1501] 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) I0421 02:18:25.207608 140707638953792 maxtext_utils.py:1501] 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') I0421 02:18:25.207651 140707638953792 maxtext_utils.py:1501] 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) I0421 02:18:25.207735 140707638953792 maxtext_utils.py:1501] 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) I0421 02:18:25.207789 140707638953792 maxtext_utils.py:1501] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0421 02:18:25.207839 140707638953792 maxtext_utils.py:1501] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0421 02:18:25.705343 140707638953792 train.py:156] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 02:18:25.705437 140707638953792 train.py:156] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0421 02:18:25.720716 140707638953792 train.py:163] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 02:18:25.720777 140707638953792 train.py:163] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0421 02:18:36.361753 140707638953792 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. I0421 02:18:36.362616 140707638953792 metric_logger.py:289] number parameters: 1.104 billion I0421 02:18:47.324909 140707638953792 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0421 02:18:47.867695 140707638953792 checkpointing.py:776] Waited 0.542733907699585 seconds for step 10 to finish before starting checkpointing. I0421 02:18:47.870554 140707638953792 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 02:18:47.872589 140707638953792 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 10 I0421 02:18:47.875544 140707638953792 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/10. I0421 02:18:48.885055 140707638953792 signaling_client.py:364] Using JaxDistributedSignalingClient I0421 02:18:48.887593 140707638953792 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0421 02:18:48.887649 140707638953792 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 I0421 02:18:49.143161 140707638953792 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.258204s I0421 02:18:49.143336 140707638953792 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.005 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5133907794952393 s) (per-host) I0421 02:18:49.143391 140707638953792 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.513456s (batch_requests_ready=0.253235s, total_serialization_initiated=0.260147s, others=0.000074s) I0421 02:18:49.143482 140707638953792 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.517608s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.517589) I0421 02:18:49.144365 140591157458688 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0421 02:18:49.144538 140707638953792 async_checkpointer.py:561] Finished blocking save. Time taken: 1.271880s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/10. I0421 02:18:50.018822 140707638953792 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0421 02:18:50.019217 140586845705984 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 02:18:50.019384 140707638953792 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776737927.8705354, 'wait_for_prev_duration_secs': 6.4849853515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776737927.8726273, 'checkpointer_blocking_duration_secs': 1.272017002105713, 'get_old_steps_start_time': 1776737929.1446743, 'get_old_steps_duration_secs': 2.5272369384765625e-05, 'checkpoint_manager_blocking_start_time': 1776737927.8682704, 'checkpoint_manager_blocking_duration_secs': 2.1510753631591797} I0421 02:18:50.019494 140707638953792 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0421 02:18:50.019543 140707638953792 max_utils.py:750] Memstats: After params initialized: I0421 02:18:50.019603 140707638953792 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0)) I0421 02:18:50.019637 140707638953792 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0)) I0421 02:18:50.019683 140707638953792 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0)) I0421 02:18:50.019711 140707638953792 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0)) I0421 02:18:50.289696 140707638953792 metric_logger.py:185] completed step: 10, seconds: 10.962, TFLOP/s/device: 1.239, Tokens/s/device: 186.824, total_weights: 65536, loss: 6.833 I0421 02:18:50.450107 140707638953792 metric_logger.py:185] completed step: 11, seconds: 2.963, TFLOP/s/device: 4.585, Tokens/s/device: 691.092, total_weights: 65536, loss: 6.528 I0421 02:18:50.597764 140707638953792 metric_logger.py:185] completed step: 12, seconds: 0.013, TFLOP/s/device: 1036.869, Tokens/s/device: 156288.156, total_weights: 65536, loss: 6.265 I0421 02:18:50.745378 140707638953792 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.745, Tokens/s/device: 13376.616, total_weights: 65536, loss: 6.045 I0421 02:18:50.749920 140707638953792 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0421 02:18:51.040210 140707638953792 checkpointing.py:776] Waited 0.2902553081512451 seconds for step 15 to finish before starting checkpointing. I0421 02:18:51.042906 140707638953792 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 02:18:56.973710 140586862491392 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/10/items/array_metadatas/process_7 I0421 02:19:33.684174 140591157458688 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 35.060 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.05420660972595 s) (per-host) I0421 02:19:33.684277 140591157458688 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.539799s. I0421 02:19:38.906056 140591157458688 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 49.761563s. I0421 02:19:38.906270 140586845705984 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 02:19:38.906327 140586845705984 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 02:19:38.906372 140586845705984 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0421 02:19:38.908019 140586845705984 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0421 02:19:38.908199 140707638953792 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0421 02:19:38.908342 140707638953792 checkpoint_manager.py:1458] Waiting for previous save to complete took 47.865439 seconds. If this number is high, consider checkpointing less frequently. I0421 02:19:38.910214 140707638953792 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 15 I0421 02:19:38.914470 140707638953792 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/15. I0421 02:19:39.873995 140707638953792 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0421 02:19:39.874092 140707638953792 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 I0421 02:19:39.918399 140707638953792 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045633s I0421 02:19:39.918564 140707638953792 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.238 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29450559616088867 s) (per-host) I0421 02:19:39.918616 140707638953792 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.294567s (batch_requests_ready=0.247014s, total_serialization_initiated=0.047484s, others=0.000068s) I0421 02:19:39.918721 140707638953792 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.299286s (all_items=0.000014s, per_item={'items': '0.00001359'}, temp_paths=0.299273) I0421 02:19:39.919680 140590108894976 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0421 02:19:39.919841 140707638953792 async_checkpointer.py:561] Finished blocking save. Time taken: 1.009549s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/15. I0421 02:19:40.363358 140707638953792 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0421 02:19:40.363771 140586845705984 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 02:19:40.363950 140707638953792 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776737931.0428777, 'wait_for_prev_duration_secs': 47.86543917655945, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776737978.9102583, 'checkpointer_blocking_duration_secs': 1.0096909999847412, 'get_old_steps_start_time': 1776737979.9199686, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776737931.0408819, 'checkpoint_manager_blocking_duration_secs': 49.32303214073181} I0421 02:19:40.364063 140707638953792 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0421 02:19:40.364679 140707638953792 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 91.921, Tokens/s/device: 13855.344, total_weights: 65536, loss: 5.864 I0421 02:19:40.370097 140707638953792 metric_logger.py:185] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.140, Tokens/s/device: 13888.324, total_weights: 65536, loss: 5.719 I0421 02:19:57.868709 140707638953792 metric_logger.py:185] completed step: 16, seconds: 49.619, TFLOP/s/device: 0.274, Tokens/s/device: 41.275, total_weights: 65536, loss: 5.605 I0421 02:19:58.016069 140707638953792 metric_logger.py:185] completed step: 17, seconds: 17.352, TFLOP/s/device: 0.783, Tokens/s/device: 118.026, total_weights: 65536, loss: 5.517 I0421 02:19:58.163619 140707638953792 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.355, Tokens/s/device: 13468.545, total_weights: 65536, loss: 5.448 I0421 02:19:58.310793 140707638953792 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0421 02:19:58.311463 140707638953792 checkpointing.py:776] Waited 0.0006878376007080078 seconds for step 19 to finish before starting checkpointing. I0421 02:19:58.313469 140707638953792 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 02:20:02.985911 140586862491392 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/15/items/array_metadatas/process_7 I0421 02:20:38.463823 140590108894976 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 26.846 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 58.83972477912903 s) (per-host) I0421 02:20:38.463944 140590108894976 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 58.544153s. I0421 02:20:47.067106 140590108894976 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 67.147300s. I0421 02:20:47.067433 140586845705984 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 02:20:47.067555 140586845705984 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 02:20:47.067603 140586845705984 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0421 02:20:47.070163 140586845705984 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0421 02:20:47.070340 140707638953792 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0421 02:20:47.070476 140707638953792 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.757015 seconds. If this number is high, consider checkpointing less frequently. I0421 02:20:47.072030 140707638953792 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 19 I0421 02:20:47.075640 140707638953792 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/19. I0421 02:20:48.505105 140707638953792 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0421 02:20:48.505201 140707638953792 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 I0421 02:20:48.546643 140707638953792 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042619s I0421 02:20:48.546823 140707638953792 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.080 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.7416903972625732 s) (per-host) I0421 02:20:48.546872 140707638953792 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.741748s (batch_requests_ready=0.697003s, total_serialization_initiated=0.044682s, others=0.000063s) I0421 02:20:48.546959 140707638953792 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.745787s (all_items=0.000012s, per_item={'items': '0.00001192'}, temp_paths=0.745775) I0421 02:20:48.547872 140591149065984 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0421 02:20:48.547998 140707638953792 async_checkpointer.py:561] Finished blocking save. Time taken: 1.475900s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/19. I0421 02:20:48.549977 140707638953792 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0421 02:20:48.550217 140586845705984 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 02:20:48.550348 140707638953792 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776737998.3134418, 'wait_for_prev_duration_secs': 48.757014989852905, 'time_between_consecutive_saves_sec': 19.405381679534912, 'checkpointer_blocking_start_time': 1776738047.072069, 'checkpointer_blocking_duration_secs': 1.476100206375122, 'get_old_steps_start_time': 1776738048.5481825, 'get_old_steps_duration_secs': 2.0742416381835938e-05, 'checkpoint_manager_blocking_start_time': 1776737998.3117025, 'checkpoint_manager_blocking_duration_secs': 50.238616943359375} I0421 02:20:48.550501 140707638953792 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0421 02:20:48.550543 140707638953792 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 02:20:52.872700 140591157458688 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_16_zarr3_true/checkpoints/19/items/array_metadatas/process_7 I0421 02:21:29.950211 140591149065984 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 37.480 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.145036697387695 s) (per-host) I0421 02:21:29.950323 140591149065984 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.402245s. I0421 02:21:37.403509 140591149065984 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 48.855415s. I0421 02:21:37.403866 140586845705984 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 02:21:37.403983 140586845705984 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 02:21:37.404025 140586845705984 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0421 02:21:37.405616 140586845705984 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0421 02:21:37.405764 140707638953792 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0421 02:21:37.405858 140707638953792 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 02:21:37.406569 140707638953792 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.317, Tokens/s/device: 13915.029, total_weights: 65536, loss: 5.391 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Tue Apr 21 02:21:48 UTC 2026 EXIT_CODE=0